Differences between digital sources

[Quad]

I've always tried to be an enlightened audiophile. I won't do double
blind tests myself because I'm too lazy, but I usually trust those tests
and I think they are a good way to find out if something is audible or
not.

I'm 36 years old and here is what I claim to be able to hear and what I
don't:

Not audible to me:
- LMS running on different OS (Windows, OS X, Linux)
- different S/PDIF cables (as long as they meet the specs)
- different ethernet cables
- TT3 pimping (buffers, WLAN, screen,...)
- FLAC vs. WAV
- FLAC vs. 320kbps MP3
- 24bit/48kHz vs. higher resolutions (same master, properly
downsampled)

Audible to me:
- RCA cables
- speaker cables
- power cables and power supplies
- differences between DACs
- differences between sampling rate conversion algorithms
> native 192kHz with EDO sounds different than downsampled with SOX to
96kHz (either on-the-fly or not)
> SACD ripps with different DSD->PCM settings sound different (POW-R3
vs. TPDF)
- differences in digital source
> Touch sounds different than my average CD player NAD 514 (S/PDIF
into same external DAC NAD M51)
> Touch and NAD 514 sound different than Digital Music Player NAD M50
(S/PDIF into same external DAC)
> HDMI out of my Laptop into same external DAC sounds different again
(WASAPI event).

What strikes me the most is the last point, differences in digital
sources. Are there any good reasons for that? I can't see any. All of
them are bit perfect. Unfortunately I can't compare Touch USB vs. Touch
S/PDIF.

Thank you for a civilized handling of this topic.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[SBGK]

Quad wrote:
> I've always tried to be an enlightened audiophile. I won't do double
> blind tests myself because I'm too lazy, but I usually trust those tests
> and I think they are a good way to find out if something is audible or
> not.
>
> I'm 36 years old and here is what I claim to be able to hear and what I
> don't:
>
> Not audible to me:
> - LMS running on different OS (Windows, OS X, Linux)
> - different S/PDIF cables (as long as they meet the specs)
> - different ethernet cables
> - TT3 pimping (buffers, WLAN, screen,...)
> - FLAC vs. WAV
> - FLAC vs. 320kbps MP3
> - 24bit/48kHz vs. higher resolutions (same master, properly
> downsampled)
>
> Audible to me:
> - RCA cables
> - speaker cables
> - power cables and power supplies
> - differences between DACs
> - differences between sampling rate conversion algorithms
> > native 192kHz with EDO sounds different than downsampled with SOX to
> 96kHz (either on-the-fly or not)
> > SACD ripps with different DSD->PCM settings sound different (POW-R3
> vs. TPDF)
> - differences in digital source
> > Touch sounds different than my average CD player NAD 514 (S/PDIF
> into same external DAC NAD M51)
> > Touch and NAD 514 sound different than Digital Music Player NAD M50
> (S/PDIF into same external DAC)
> > HDMI out of my Laptop into same external DAC sounds different again
> (WASAPI event).
>
> What strikes me the most is the last point, differences in digital
> sources. Are there any good reasons for that? I can't see any. All of
> them are bit perfect. Unfortunately I can't compare Touch USB vs. Touch
> S/PDIF.
>
> Thank you for a civilized handling of this topic.

do different software players sound the same to you ? do different
wasapi buffer sizes sound the same to you ?
HDMI will use different chips in your dac, so depends on how well it
was implemented.
I would be worried if I couldn't tell the difference between wav of any
resolution and 320 mp3, that's like saying you can't tell the difference
between medium wave radio/internet stream and FM radio.


------------------------------------------------------------------------
SBGK's Profile: http://forums.slimdevices.com/member.php?userid=52003
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[darrenyeats]

SBGK wrote:
>
> I would be worried if I couldn't tell the difference between wav of any
> resolution and 320 mp3, that's like saying you can't tell the difference
> between medium wave radio/internet stream and FM radio.
SBGK,
That is, at the least, an exaggeration isn't it?!
Darren


------------------------------------------------------------------------
darrenyeats's Profile: http://forums.slimdevices.com/member.php?userid=10799
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

SBGK wrote:
>
> I would be worried if I couldn't tell the difference between wav of any
> resolution and 320 mp3, that's like saying you can't tell the difference
> between medium wave radio/internet stream and FM radio.

OMG. Not *THAT* again!


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Quad wrote:
> I've always tried to be an enlightened audiophile. I won't do double
> blind tests myself because I'm too lazy, but I usually trust those tests
> and I think they are a good way to find out if something is audible or
> not.
>
> I'm 36 years old and here is what I claim to be able to hear and what I
> don't:
>
> Not audible to me:
> - LMS running on different OS (Windows, OS X, Linux)
> - different S/PDIF cables (as long as they meet the specs)
> - different ethernet cables
> - TT3 pimping (buffers, WLAN, screen,...)
> - FLAC vs. WAV
> - FLAC vs. 320kbps MP3
> - 24bit/48kHz vs. higher resolutions (same master, properly
> downsampled)
>
> Audible to me:
> - RCA cables
> - speaker cables
> - power cables and power supplies
> - differences between DACs
> - differences between sampling rate conversion algorithms
> > native 192kHz with EDO sounds different than downsampled with SOX to
> 96kHz (either on-the-fly or not)
> > SACD ripps with different DSD->PCM settings sound different (POW-R3
> vs. TPDF)
> - differences in digital source
> > Touch sounds different than my average CD player NAD 514 (S/PDIF
> into same external DAC NAD M51)
> > Touch and NAD 514 sound different than Digital Music Player NAD M50
> (S/PDIF into same external DAC)
> > HDMI out of my Laptop into same external DAC sounds different again
> (WASAPI event).
>
> What strikes me the most is the last point, differences in digital
> sources. Are there any good reasons for that? I can't see any. All of
> them are bit perfect. Unfortunately I can't compare Touch USB vs. Touch
> S/PDIF.
>
> Thank you for a civilized handling of this topic.

Good post Quad.

Don't worry, I don't think in general the folks here are mean spirited
over a hobby :-)

Nice summary list. Good to see you've gone over a number of
possibilities and listened for yourself. I'm 41 now and over the years
have tried a few of the tests and measurements as well. Something I've
run into frequently has been the importance of test methodology. For
example, there were times when a group of guys got together to listen to
2 speaker cables and everyone seem to agree that Cable A is better than
Cable B, but the moment someone switches it and the test becomes 'single
blinded', that effect is gone.

Likewise, testing DAC's, the moment we introduce controls like making
sure the volume is <0.25dB between the two units, there's no way any of
us can tell the difference in an A-B test between decent solid state
DAC's.

Since you mentioned you haven't tried double blind testing, why not do
just a controlled straight forward 'instantaneous' A-B with the Touch
vs. NAD 514 into the M51?
- NAD 514 --> coaxial --> M51
- Touch --> optical --> M51
- Get a pink noise test tone and use a sound pressure meter (like the
Radio Shack one) to make sure the two sources are about the same dB
(should be if bit perfect unless M51 has independent volume controls for
each input)
- Go ahead and use the Touch digital volume control to change the volume
until the output SPL is <0.25dB different (this will bias against the
Touch of course, 100% volume is bit perfect for the Touch)
- Play the same album on both the NAD 514 and Touch at the same time.
- Flip back and forth between coaxial & Toslink (NAD 514 vs. Touch) -
does the music sound tonally different from the 2 inputs?

Realize doing this biases against the Touch both in terms of using the
digital volume control and TosLink generally has worse jitter.

Potentially lots can be talked about but I think the procedure above
should be quite straight forward and would be worth doing to test one of
the assertions about bit perfection and audibility if you haven't
already tried.


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Archimago wrote:
> Good post Quad.
>
> Don't worry, I don't think in general the folks here are mean spirited
> over a hobby :-)
>
> Nice summary list. Good to see you've gone over a number of
> possibilities and listened for yourself. I'm 41 now and over the years
> have tried a few of the tests and measurements as well. Something I've
> run into frequently has been the importance of test methodology. For
> example, there were times when a group of guys got together to listen to
> 2 speaker cables and everyone seem to agree that Cable A is better than
> Cable B, but the moment someone switches it and the test becomes 'single
> blinded', that effect is gone.
>
> Likewise, testing DAC's, the moment we introduce controls like making
> sure the volume is <0.25dB between the two units, there's no way any of
> us can tell the difference in an A-B test between decent solid state
> DAC's.
>
> Since you mentioned you haven't tried double blind testing, why not do
> just a controlled straight forward 'instantaneous' A-B with the Touch
> vs. NAD 514 into the M51?
> - NAD 514 --> coaxial --> M51
> - Touch --> optical --> M51
> - Get a pink noise test tone and use a sound pressure meter (like the
> Radio Shack one) to make sure the two sources are about the same dB
> (should be if bit perfect unless M51 has independent volume controls for
> each input)
> - Go ahead and use the Touch digital volume control to change the volume
> until the output SPL is <0.25dB different (this will bias against the
> Touch of course, 100% volume is bit perfect for the Touch)
> - Play the same album on both the NAD 514 and Touch at the same time.
> - Flip back and forth between coaxial & Toslink (NAD 514 vs. Touch) -
> does the music sound tonally different from the 2 inputs? Resolution and
> dynamics differences? Would you or someone sitting with you be able to
> differentiate the two if blinded to the source of the input?
>
> Realize doing this biases against the Touch both in terms of using the
> digital volume control and TosLink generally has worse jitter.
>
> Potentially lots can be talked about but I think the procedure above
> should be quite straight forward and would be worth doing to test one of
> the assertions about bit perfection and audibility if you haven't
> already tried. (If you need to adjust the Touch's volume it would of
> course not be bit perfect and you would need to ask why the M51 plays
> digital input at different volumes if this were to happen!)

Thanks for your reply Archimago. I will get a TosLink cable and follow
your instructions. I really hope I can't reproduce my findings.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Quad wrote:
> I've always tried to be an enlightened audiophile. I won't do double
> blind tests myself because I'm too lazy, but I usually trust those tests
> and I think they are a good way to find out if something is audible or
> not.
>
> I'm 36 years old and here is what I claim to be able to hear and what I
> don't:
>
> Not audible to me:
> - LMS running on different OS (Windows, OS X, Linux)
> - different S/PDIF cables (as long as they meet the specs)
> - different ethernet cables
> - TT3 pimping (buffers, WLAN, screen,...)
> - FLAC vs. WAV
> - FLAC vs. 320kbps MP3
> - 24bit/48kHz vs. higher resolutions (same master, properly
> downsampled)
>
> Audible to me:
> - RCA cables
> - speaker cables
> - power cables and power supplies
> - differences between DACs
> - differences between sampling rate conversion algorithms
> > native 192kHz with EDO sounds different than downsampled with SOX to
> 96kHz (either on-the-fly or not)
> > SACD ripps with different DSD->PCM settings sound different (POW-R3
> vs. TPDF)
> - differences in digital source
> > Touch sounds different than my average CD player NAD 514 (S/PDIF
> into same external DAC NAD M51)
> > Touch and NAD 514 sound different than Digital Music Player NAD M50
> (S/PDIF into same external DAC)
> > HDMI out of my Laptop into same external DAC sounds different again
> (WASAPI event).
>
> What strikes me the most is the last point, differences in digital
> sources. Are there any good reasons for that? I can't see any. All of
> them are bit perfect. Unfortunately I can't compare Touch USB vs. Touch
> S/PDIF.
>
> Thank you for a civilized handling of this topic.

You really lost me when you say you hear differences in power
cables(that really nothing to the music signal), but you can NOT hear
differences in FLAC vs MP3/320 that
removes over 70% of the original music data ????? It is audible......
Indeed....


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

callesoroe wrote:
> hear differences in FLAC vs MP3/320 that removes over 70% of the
> original music data ????? It is audible....

Do you have any credible evidence showing it is audible?


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Julf wrote:
> Do you have any credible evidence showing it is audible?

It is quite logic...... and there is really huge differences in
sound.... Room and life in the music dissapears ..... and your ears gets
tired of listning to it in longer time.
Drum cymbals say "DING" and not "Diiiiiiiiiing" when listning to MP3
....... Just an examble.....


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

callesoroe wrote:
> It is quite logic...... and there is really huge differences in
> sound.... Room and life in the music dissapears ..... and your ears gets
> tired of listning to it in longer time.
> Drum cymbals say "DING" and not "Diiiiiiiiiing" when listning to MP3
> ....... Just an examble.....

Have you done an ABX test (foobar2000 has a nice component for
automating this). If you can always tell the difference between a
320kbs mp3 and the original lossless file, you're in a very, very tiny
majority based on lots of tests, and even they typically only note
differences in problem tracks after they've trained themselves to hear
mp3 artifacts.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

callesoroe wrote:
> It is quite logic...... and there is really huge differences in
> sound.... Room and life in the music dissapears ..... and your ears gets
> tired of listning to it in longer time.
> Drum cymbals say "DING" and not "Diiiiiiiiiing" when listning to MP3
> ....... Just an examble.....

No. It's not quite logic. In fact it is pretty far from logic. MP3 is a
perceptual codec. It throws away data - but only data that the
perceptual algorithms deem unlikely to be audible. The rest of what you
report is purely subjective observations. Please note I asked for
"credible evidence". Subjective impressions, especially if they confirm
something you believe, are often the result of perceptive/cognitive
bias, as you probably know.


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Julf wrote:
> No. It's not quite logic. In fact it is pretty far from logic. MP3 is a
> perceptual codec. It throws away data - but only data that the
> perceptual algorithms deem unlikely to be audible. The rest of what you
> report is purely subjective observations. Please note I asked for
> "credible evidence". Subjective impressions, especially if they confirm
> something you believe, are often the result of perceptive/cognitive
> bias, as you probably know.

My only evidence is that the data of the original music file has been
dramatic reduced, and that is very audible in my setup. And I am NOT a
cable freak og
anything like that. The differences is just so big, that I can not
understand if anyone should have problems telling what is what.
There must be something wrong in the setups....

That said. A very well recorded track can sound better as a MP3/320
file, than a bad recorded track as a FLAC file.
But if you listen to the well recorded track, the FLAC version will be
much better than the MP3. If your system can
not reveal that, then there is definitely something wrong in the setup.


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

callesoroe wrote:
> My only evidence is that the data of the original music file has been
> dramatic reduced, and that is very audible in my setup. And I am NOT a
> cable freak og
> anything like that. The differences is just so big, that I can not
> understand if anyone should have problems telling what is what.
> There must be something wrong in the setups....
>
> That said. A very well recorded track can sound better as a MP3/320
> file, than a bad recorded track as a FLAC file.
> But if you listen to the well recorded track, the FLAC version will be
> much better than the MP3. If your system can
> not reveal that, then there is definitely something wrong in the setup.

? What do you mean "well recorded" or "bad recorded" track? Are you
talking about the source itself (ie. different remasters?) or bad
ripping of the same CD?

I understand your assertion/belief in the first paragraph (whether it's
empirically true is another issue), but your second paragraph was
unclear...


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Archimago wrote:
> ? What do you mean "well recorded" or "bad recorded" track? Are you
> talking about the source itself (ie. different remasters?) or bad
> ripping of the same CD?
>
> I understand your assertion/belief in the first paragraph (whether it's
> empirically true is another issue), but your second paragraph was
> unclear...

What I meant was that you need to have good source material to reveal
the differences. A bad recording compressed and mastered to death in the
studio, then I agree it is difficult to hear the difference(if any). But
when the source material is great, the you will certanly hear it.
I did not mean anything like bad rips or so. I just operate with
accurate rips.


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

callesoroe wrote:
> What I meant was that you need to have good source material to reveal
> the differences. A bad recording compressed and mastered to death in the
> studio, then I agree it is difficult to hear the difference(if any). But
> when the source material is great, the you will certanly hear it.
> I did not mean anything like bad rips or so. I just operate with
> accurate rips.

Interesting. Although my opinion differs. IMO better quality masters
with good dynamic range and generally "quieter" overall presentations
allow modern encoders like LAME to be even more efficient and requiring
less number of bits for the perceptual coding. Forcing them into 320kbps
with these better recordings actually allows the algorithm to encode
even more inaudible & unperceptable information.

The reason I say this is that when I did the MP3 test back in December,
I specifically chose high quality samples PLUS one very loud and clipped
metal track suggested by someone. If you just look at the variance in
the WAV files after encoding & decoding, the clipped, loud, poorly
mastered track was more difficult for the LAME encoder than the others
to handle - hence producing more opportunity to differentiate the sound.


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

callesoroe wrote:
> My only evidence is that the data of the original music file has been
> dramatic reduced, and that is very audible in my setup.

Not quite. The data of the original music has been dramatically reduced,
yes. You can hear a difference. Yes. But one does not necessarily follow
from the other.

What mp3 encoder are you using?


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Julf wrote:
> Not quite. The data of the original music has been dramatically reduced,
> yes. You can hear a difference. Yes. But one does not necessarily follow
> from the other.
>
> What mp3 encoder are you using?

I have tried both foobar2000 and Mediamonkey Gold. Same results.


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

callesoroe wrote:
> I have tried both foobar2000 and Mediamonkey Gold. Same results.

These are not encoders. Within mediamonkey or foobar2000 are you
encoding mp3 using LAME (I assume). And if so, VBR. CBR, ABR?


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[SoftwireEngineer]

Ok, this is a better way of saying than jumping the gun and deciding I
need a transport with less jitter (mea culpa :-) ) . Inspired by
Archimago, I threw away the TT tweak on my Touch, but later I went back
to it, because I felt I was missing the resolution I had earlier. I also
knew that Archimago, measured the jitter differences to be very less or
insignificant. Now, I am wondering whether some sort of noise creeps
into the DACs/Amps from the transport (no pulse transformer/galvanic
isolation) ??? (in that context, a shielded power cord to the touch
linear power supply seemed to reduce some haze and/or increase
resolution).
Another observation or illusion (?), if I unplug all the power cords,
clean the prongs with just a clean paper towel and plug it in, the
system sounds very detailed. I check for this with very low volume
listening. What could be happening ?


------------------------------------------------------------------------
SoftwireEngineer's Profile: http://forums.slimdevices.com/member.php?userid=7000
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[SuperQ]

SoftwireEngineer wrote:
> Another observation or illusion (?), if I unplug all the power cords,
> clean the prongs with just a clean paper towel and plug it in, the
> system sounds very detailed. I check for this with very low volume
> listening. What could be happening ?

Very likely the system is being affected by
Expectation
Bias (http://en.wikipedia.org/wiki/Experimenter's_bias). Simply knowing about the change will change your perception
of the output. Your brain is very bad at dealing with visual vs
auditory stimulus.

For example, the 'McGurk effect'
(https://www.youtube.com/watch?hd=1&v=G-lN8vWm3m0) ('more info'
(http://en.wikipedia.org/wiki/McGurk_effect))


------------------------------------------------------------------------
SuperQ's Profile: http://forums.slimdevices.com/member.php?userid=2139
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

SoftwireEngineer wrote:
> Ok, this is a better way of saying than jumping the gun and deciding I
> need a transport with less jitter (mea culpa :-) ) . Inspired by
> Archimago, I threw away the TT tweak on my Touch, but later I went back
> to it, because I felt I was missing the resolution I had earlier. I also
> knew that Archimago, measured the jitter differences to be very less or
> insignificant. Now, I am wondering whether some sort of noise creeps
> into the DACs/Amps from the transport (no pulse transformer/galvanic
> isolation) ??? (in that context, a shielded power cord to the touch
> linear power supply seemed to reduce some haze and/or increase
> resolution).
> Another observation or illusion (?), if I unplug all the power cords,
> clean the prongs with just a clean paper towel and plug it in, the
> system sounds very detailed. I check for this with very low volume
> listening. What could be happening ?

Cool that you tried anyways Softwire.

I bet your system sounds "very detailed" whether you cleaned with the
paper towel or not :-)

Remember that sensory memory is very short - check out the WiKi on
"echoic memory" (below). We're talking *4-10 seconds*. This is why doing
stuff like testing cable A vs. B which requires fiddling around with
plugging and unplugging is a "no-no". Likewise, when I suggested Quad do
the A-B test, it's best to switch quickly between TosLink and coaxial
with the DAC.

Unless the difference is quite clear, it's VERY HARD if not IMPOSSIBLE
to differentiate minute differences. The mind will look for other
effects like expectation bias, recency/primary effect, slight volume
differences, etc. to make its choice. Like I said before, start
controlling the variables or do blind testing and you'll likely realize
differences will disappear. BTW - since my father has a Touch as well, I
was able to test one unit with the TT mod and the other in stock
condition connected to his ONYX Melody SP3 tube amp, switching between
the 2 inputs while the units played the same song. I could not tell a
difference...

Enjoy the music either way :-)

http://en.wikipedia.org/wiki/Echoic_memory


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Archimago wrote:
> Cool that you tried anyways Softwire.
>
> I bet your system sounds "very detailed" whether you cleaned with the
> paper towel or not :-)
>
> Remember that sensory memory is very short - check out the WiKi on
> "echoic memory" (below). We're talking conservatively *3-10 seconds*
> (possibly much shorter). This is why doing stuff like testing cable A
> vs. B which requires fiddling around with plugging and unplugging is a
> "no-no". Likewise, when I suggested Quad do the A-B test, it's best to
> switch quickly between TosLink and coaxial with the DAC.
>
> Unless the difference is quite clear, it's VERY HARD if not IMPOSSIBLE
> to differentiate minute differences. The mind will look for other
> effects like expectation bias, recency/primary effect, slight volume
> differences, etc. to make its choice. Like I said before, start
> controlling the variables or do blind testing and you'll likely realize
> differences will disappear. BTW - since my father has a Touch as well, I
> was able to test one unit with the TT mod and the other in stock
> condition connected to his ONYX Melody SP3 tube amp, switching between
> the 2 inputs while the units played the same song. I could not tell a
> difference...
>
> Enjoy the music either way :-)
>
> http://en.wikipedia.org/wiki/Echoic_memory

It is a long time since I did this FLAC vs MP3 test, because the results
was so obvious, that there was no doubt at all in my mind.
But inspired of this thread I tried Again..............
The same track in a FLAC version and a MP3 320 CBR in a folder. Same
replay gain information. Testet both with replay gain off/on
Play both files with my Transporter. And the results is still the
same.........
Room and resolution dissapears with MP3. There is no doubt at all.
It is like putting a carpet in front of my Martin Logan speakers...... I
have no doubt for what so ever, that there is a huge difference.


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

callesoroe wrote:
> The same track in a FLAC version and a MP3 320 CBR in a folder. Same
> replay gain information. Testet both with replay gain off/on

And how did you implement the "blind" part of the test?


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

Julf wrote:
> And how did you implement the "blind" part of the test?

random play


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

callesoroe wrote:
> random play

But to see if you got it right, you had to look? Or did you write down
your opinions, and cross-check them afterwards against the list of
tracks actually played?


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[darrenyeats]

Julf wrote:
> But to see if you got it right, you had to look? Or did you write down
> your opinions, and cross-check them afterwards against the list of
> tracks actually played?

Interesting results but, yes, I would like to know the details of how it
was done too.
Darren


------------------------------------------------------------------------
darrenyeats's Profile: http://forums.slimdevices.com/member.php?userid=10799
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

callesoroe wrote:
> It is a long time since I did this FLAC vs MP3 test, because the results
> was so obvious, that there was no doubt at all in my mind.
> But inspired of this thread I tried Again..............
> The same track in a FLAC version and a MP3 320 CBR in a folder. Same
> replay gain information. Testet both with replay gain off/on
> Play both files with my Transporter. And the results is still the
> same.........
> Room and resolution dissapears with MP3. There is no doubt at all.
> It is like putting a carpet in front of my Martin Logan speakers...... I
> have no doubt for what so ever, that there is a huge difference.

Callesoroe, Please don't take offense, and I certainly don't mean any
(it's all about the music after all, and I myself use FLAC files at home
and a Transporter!). But it is simply amazing to me that you are able
to detect "huge" differences (and I assume you've tried a number of
tracks, not simply one track, which could of course be a song that might
be a "problem sample" for the mp3 encoder). I'm amazed because this
flies in the face of controlled experiments of 100s (maybe thousands) of
people, including folks who are taking the comparisons (and test design)
very, very seriously.

So my only interpretation is that you either have very unusual hearing
(and I'm pretty sure you're not 8 years old....as young people do in
fact have very good hearing regarding high frequencies) or there is
something else not setup correctly in your setup. For example, have you
done anything different from default in "file types" in LMS or bitrate
limiting, etc. I'm just grasping for some difference in your system
that makes high bit rate mp3 files sound so clearly worse when this is
simply not supported by widespread empirical evidence. I'm not doubting
that you hear huge differences....I'm just doubting that the obvious
source of this difference can be FLAC vs 320kbs mp3 file. Regards.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[SBGK]

garym wrote:
> Callesoroe, Please don't take offense, and I certainly don't mean any
> (it's all about the music after all, and I myself use FLAC files at home
> and a Transporter!). But it is simply amazing to me that you are able
> to detect "huge" differences (and I assume you've tried a number of
> tracks, not simply one track, which could of course be a song that might
> be a "problem sample" for the mp3 encoder). I'm amazed because this
> flies in the face of controlled experiments of 100s (maybe thousands) of
> people, including folks who are taking the comparisons (and test design)
> very, very seriously.
>
> So my only interpretation is that you either have very unusual hearing
> (and I'm pretty sure you're not 8 years old....as young people do in
> fact have very good hearing regarding high frequencies) or there is
> something else not setup correctly in your setup. For example, have you
> done anything different from default in "file types" in LMS or bitrate
> limiting, etc. I'm just grasping for some difference in your system
> that makes high bit rate mp3 files sound so clearly worse when this is
> simply not supported by widespread empirical evidence. I'm not doubting
> that you hear huge differences....I'm just doubting that the obvious
> source of this difference can be FLAC vs 320kbs mp3 file. Regards.

I guess you could say the differences detected are small where the music
appears to be smeared especialy during more complex passages compared to
a wav file, so it comes down to a matter of which one is preferred. I
use 320vbr quicktime transcoded mp3 on my Ipod Touch as that is good
enough for portable music, but on my main system the resolution is just
not there in mp3. I guess the thousands of people you are quoting don't
have high quality systems or are just not that bothered by how the music
sounds, even so an average pair of headphones plugged into a laptop is
all it takes to differentiate mp3 from Wav/flac.


------------------------------------------------------------------------
SBGK's Profile: http://forums.slimdevices.com/member.php?userid=52003
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

SBGK wrote:
> I guess the thousands of people you are quoting don't have high quality
> systems or are just not that bothered by how the music sounds, even so
> an average pair of headphones plugged into a laptop is all it takes to
> differentiate mp3 from Wav/flac.

No, I'm talking about tests with people that have good systems and do
care. Obviously there are probably millions of people who are perfectly
happy with 128kbps lossy files. That's not the target group I'm
referring to. And please don't play "well I guess if you can't hear the
difference you're system is not resolving enough" card. This is of
course nonsense on several dimensions.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

By the way, I do understand that a people can actually train themselves
to hear artifacts in lossy music. It is possible (but not something I'd
want to do). But even this is based on detecting certain very specific
cues and is not the same as an entire song sounding like a carpet was
laid over the the front of speakers, or the room opened up, etc.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

garym wrote:
> No, I'm talking about tests with people that have good systems and do
> care. Obviously there are probably millions of people who are perfectly
> happy with 128kbps lossy files. That's not the target group I'm
> referring to. And please don't play "well I guess if you can't hear the
> difference you're system is not resolving enough" card. This is of
> course nonsense on several dimensions.

I agree with this comment, garym.

In the MP3 test I ran late last year into Feb 2013, the folks with more
expensive systems (reportedly >$6000) actually PREFERRED the MP3 tracks
even more! This is with the ~150 respondents I received. This was a
surprise finding for me. Obviously audio lovers / audiophiles (aka audio
geeks) who would download ~75MB of test music, sit around to listen to 3
different tracks from different genres, and then fill out a
multi-question survey for me didn't in general hear massive
differences...

callesoroe: What song(s) did you use to test this? I'd love to have a
listen to this track myself.

For those who haven't had a look, here are the posts for the MP3 test a
few months ago:
Preamble:
http://archimago.blogspot.ca/2013/01/high-bitrate-mp3-results-part-0-preamble.html
- I still haven't heard from Neil Young about this :-)
Procedure:
http://archimago.blogspot.ca/2013/02/high-bitrate-mp3-internet-blind-test.html
Results:
http://archimago.blogspot.ca/2013/02/high-bitrate-mp3-internet-blind-test_3422.html
Discussion:
http://archimago.blogspot.ca/2013/02/high-bitrate-mp3-internet-blind-test_2.html
Subjective Comments:
http://archimago.blogspot.ca/2013/02/high-bitrate-internet-blind-test-part-4.html
- Look at the comments from those who chose Set A (incorrectly). They
didn't seem to hear veils and such and thought the lossy was FLAC...

Although I don't claim the data to be the end of the debate, at least I
can say I have tried my best to receive input on this topic in a way
that concealed the MP3 file's identity and thus as best I can tell, the
results represent a blinded test of music going through the LAME encoder
(twice in fact!).


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

garym wrote:
> So my only interpretation is that you either have very unusual hearing
> (and I'm pretty sure you're not 8 years old....as young people do in
> fact have very good hearing regarding high frequencies) or there is
> something else not setup correctly in your setup.

Interestingly, it seems that older people with damaged hearing are more
likely to hear differences between mp3 and lossless, as the masking
works less well as the hearing range of the ear gets narrower and more
selective.


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Julf wrote:
> Interestingly, it seems that older people with damaged hearing are more
> likely to hear differences between mp3 and lossless, as the masking
> works less well as the hearing range of the ear gets narrower and more
> selective.

Quite possible Julf although I haven't seen experiments around this
published (although I haven't looked exhaustively).

Since most of the anomalies people complain of are in the high
frequencies, age-related high frequency loss should not be an issue I
would have thought. Personally, I hope *not* to experience this
phenomenon first hand in another 20 years ;-(


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Archimago wrote:
> Quite possible Julf although I haven't seen experiments around this
> published (although I haven't looked exhaustively).
>
> Since most of the anomalies people complain of are in the high
> frequencies, age-related high frequency loss should not be an issue I
> would have thought. Personally, I hope *not* to experience this
> phenomenon first hand in another 20 years ;-(

just in case I experience greater differences between lossless and mp3
when I'm older, it's a good thing I rip everything to FLAC. Ooops,
wait a minute....I'm already pretty damn old...I was buying Beatles
songs on 45 records *when originally released!* ;-)


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

garym wrote:
> just in case I experience greater differences between lossless and mp3
> when I'm older, it's a good thing I rip everything to FLAC. Ooops,
> wait a minute....I'm already pretty damn old...I was buying Beatles
> songs on 45 records *when originally released!* ;-)

Hey, I'm envious you had a chance to "be there" when it happened and to
grab those 45's :-)

Not so much about the age... :-)


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[SBGK]

Look at this article comparing 48kbps and 160 kbps streams

http://crave.cnet.co.uk/digitalmusic/spotify-vs-sky-songs-sound-quality-blind-test-49303980/

at the end they say

Things to bear in mind

Although conducted fairly, our test was still very subjective. Everybody
hears 'quality' differently -- to some it's that bass sounds deeper or
louder, to some it's that higher-frequency sounds are "crisper" or
better defined.

Also, AAC+ and OGG Vorbis compression methods offer different advantages
in terms of compression efficiency, largely because the psycho-acoustic
algorithms they use are different.

But we were also testing on very high-end equipment. Many people will
simply use iPod earbuds to listen to music, or bog-standard PC speakers.
And through this lower-quality gear, the differences between Sky Songs
and Spotify's free service could be harder to identify.
Conclusion

Whatever the reasoning, Spotify's bit rate is higher, and the majority
of our participants correctly identified this. If you choose to pay for
Spotify's premium service, the quality is higher still. We can only
conclude, therefore, that if you're purely looking for quality of
streaming audio, Spotify is still your best choice.

Update: We have added a paragraph to clarify that this is simply a
casual, anecdotal comparison of two products, and not a definitive study
of the benefits of AAC and OGG Vorbis compression formats. We are well
aware that AAC, OGG Vorbis, MP3 or WMA files of identical bit rates will
not sound the same. If this was a serious study of codec performance, we
would have used 16,000 people, not 16.

Must say, this bears out my findings, different codecs sound different,
that's why I use quicktime 320 vbr for my mp3s and wav for my main
system.

Interesting that out of 16 people the majority of them identified the
higher sample rate.


------------------------------------------------------------------------
SBGK's Profile: http://forums.slimdevices.com/member.php?userid=52003
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

SBGK wrote:
> Look at this article comparing 48kbps and 160 kbps streams
>
> Interesting that out of 16 people the majority of them identified the
> higher sample rate.

A majority identified the difference between 48kbps and 160 kbps
streams? Wow! More news at 11!

It might come as a surprise that a lot of other tests have identified
128k as the limit beyond which most people
can't hear any difference...


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Julf wrote:
> A majority identified the difference between 48kbps and 160 kbps
> streams? Wow! More news at 11!
>
> It might come as a surprise that a lot of other tests have identified
> 128k as the limit beyond which most people
> can't hear any difference...

+1. All my comments have been referencing high bit rate lossy (even
320kbps) transparency relative to lossless.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Archimago wrote:
> Hey, I'm envious you had a chance to "be there" when it happened and to
> grab those 45's :-)
>
> Not so much about the age... :-)

Strangely enough, every decade of my life I've enjoyed more than the
last. I'm in my late 50s now and having a better time than ever. Can't
do some of the things I did in my 20s, but I'm doing things now that I
could never AFFORD to do in my 20s and 30s. Each age has its pros and
cons.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

garym wrote:
> Strangely enough, every decade of my life I've enjoyed more than the
> last. I'm in my late 50s now and having a better time than ever. Can't
> do some of the things I did in my 20s, but I'm doing things now that I
> could never AFFORD to do in my 20s and 30s. Each age has its pros and
> cons.

Nice garym,
That's the way it should be!


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

garym wrote:
> Callesoroe, Please don't take offense, and I certainly don't mean any
> (it's all about the music after all, and I myself use FLAC files at home
> and a Transporter!). But it is simply amazing to me that you are able
> to detect "huge" differences (and I assume you've tried a number of
> tracks, not simply one track, which could of course be a song that might
> be a "problem sample" for the mp3 encoder). I'm amazed because this
> flies in the face of controlled experiments of 100s (maybe thousands) of
> people, including folks who are taking the comparisons (and test design)
> very, very seriously.
>
> So my only interpretation is that you either have very unusual hearing
> (and I'm pretty sure you're not 8 years old....as young people do in
> fact have very good hearing regarding high frequencies) or there is
> something else not setup correctly in your setup. For example, have you
> done anything different from default in "file types" in LMS or bitrate
> limiting, etc. I'm just grasping for some difference in your system
> that makes high bit rate mp3 files sound so clearly worse when this is
> simply not supported by widespread empirical evidence. I'm not doubting
> that you hear huge differences....I'm just doubting that the obvious
> source of this difference can be FLAC vs 320kbs mp3 file. Regards.

Hi Gary!

I can just say what I did was a pretty clear experiment. I can only say
that my Logan speakers are very very detailed, and the can really do
"magic"....... but when playing MP3 the magic dissapears. What I love so
much about my speakers, is the air and room they can make. And
voices......... But all the magic is gone when I listen to MP3. It does
not sound awfull but the magic, that I love so much is gone.....
And when all comes to all. Who cares about diskspace anymore....... No
reason for MP3 with todays diskprices .......


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

callesoroe wrote:
> Hi Gary!
>
> I can just say what I did was a pretty clear experiment. I can only say
> that my Logan speakers are very very detailed, and the can really do
> "magic"....... but when playing MP3 the magic dissapears. What I love so
> much about my speakers, is the air and room they can make. And
> voices......... But all the magic is gone when I listen to MP3. It does
> not sound awfull but the magic, that I love so much is gone.....
> And when all comes to all. Who cares about diskspace anymore....... No
> reason for MP3 with todays diskprices .......

I agree about disk space. Which is why I rip to FLAC. And I listen to
FLAC at home because that's what I have. But in my day job I'm mostly
doing scientific method-based research and I just can't measure "magic"
or "air" in music. I can measure/interpret double blind test results
where subjects in the experiments could not distinguish between high
bitrate mp3 and lossless files. Thus my discussion earlier. But anyhow,
it's your system and your ears so I'm pleased you enjoy it. And
ultimately enjoying the music is what its all about. Actually my
overall preference is seeing LIVE music, even though in many cases the
technical features are not really as good as a well recorded CD of the
same music. (And yes, to be fair, I can't measure that aspect of my
enjoyment either). Regards, Gary


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[SoftwireEngineer]

garym wrote:
> ...But in my day job I'm mostly doing scientific method-based research
> and I just can't measure "magic" or "air" in music...

I have some test tracks (Chesky, I think) which have recordings of
instruments at different distances from the mic. It also has instruments
at different height relative to the mic. I or in my system I dont think
I can notice the height very well. But the distance is very discernible
if the distance is more than 5-10 ft(vague memory). So something like
this could help evaluate the "room" in the original post. "Air" I think
is pretty much decay. We can probably measure the decay trail too. But I
doubt in dbs it will be much significant.
Re: FLAC vs 320Kbps. I have not done much testing but Spotify is easily
discernible for me from my original FLAC tracks. The music sounds flat
(no depth) and I think there is also some sort of "haze" or vagueness to
the sound. My FLACs when the system is at its best (late night/quiet
etc) sounds very 'crisp'.


------------------------------------------------------------------------
SoftwireEngineer's Profile: http://forums.slimdevices.com/member.php?userid=7000
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Here's a first log of this morning. Not quite as good as yesterday, but
still obvious.

I'm somehow flabbergasted.


+-------------------------------------------------------------------+
|Filename: Roll.txt |
|Download: http://forums.slimdevices.com/attachment.php?attachmentid=14665|
+-------------------------------------------------------------------+

------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Quad wrote:
> Here's a first log of this morning. Not quite as good as yesterday, but
> still obvious.
>
> I'm somehow flabbergasted.

That's pretty good Quad. Wondering what settings/program you used for
MP3 encoding?

I have that album so will have a good listen to the song myself...


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Archimago wrote:
> That's pretty good Quad. Wondering what settings/program you used for
> MP3 encoding?
>
> I have that album so will have a good listen to the song myself...

LAME 3.99.5 within foobar. For settings refer to the attachment.

The piano intro was easiest for me to distinguish between the two
versions. It was easier when I started over the track again and again
for ABXY.


+-------------------------------------------------------------------+
|Filename: settings.jpg |
|Download: http://forums.slimdevices.com/attachment.php?attachmentid=14666|
+-------------------------------------------------------------------+

------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Here's another one. Harder but possible.

It is the infamous recording that caused discussions in the version of
HDTracks. I own the redbook version.


+-------------------------------------------------------------------+
|Filename: Concerto.txt |
|Download: http://forums.slimdevices.com/attachment.php?attachmentid=14667|
+-------------------------------------------------------------------+

------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

SoftwireEngineer wrote:
> I have some test tracks (Chesky, I think) which have recordings of
> instruments at different distances from the mic. It also has instruments
> at different height relative to the mic. I or in my system I dont think
> I can notice the height very well. But the distance is very discernible
> if the distance is more than 5-10 ft(vague memory). So something like
> this could help evaluate the "room" in the original post. "Air" I think
> is pretty much decay. We can probably measure the decay trail too. But I
> doubt in dbs it will be much significant.
> Re: FLAC vs 320Kbps. I have not done much testing but Spotify is easily
> discernible for me from my original FLAC tracks. The music sounds flat
> (no depth) and I think there is also some sort of "haze" or vagueness to
> the sound. My FLACs when the system is at its best (late night/quiet
> etc) sounds very 'crisp'.

Are you using Spotify Premium?


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

SoftwireEngineer wrote:
> I have not done much testing but Spotify is easily discernible for me
> from my original FLAC tracks.

Didn't someone show that spotify was using dynamic compression?


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Even harder, but still discernible.

The original here is 24/192 while the MP3 is 24/48, downsampled with the
built-in foobar SRC. I might hear the donwsampling algorithm and not the
codec.


+-------------------------------------------------------------------+
|Filename: Sor.txt |
|Download: http://forums.slimdevices.com/attachment.php?attachmentid=14668|
+-------------------------------------------------------------------+

------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> Even harder, but still discernible.
>
> The original here is 24/192 while the MP3 is 24/48, downsampled with the
> built-in foobar SRC. I might hear the donwsampling algorithm and not the
> codec.

Aha. You're interpreting the test results incorrectly. (You must be
thinking that a high percentage is good...in fact a high percentage
means something very different here) The log you posted indicates that
you can't distinguish between the lossy and lossess tracks. You would
want to see a % (p-value) of, say, 5% or less. But I'd be impressed with
even 10% or less. bottom line, the results you posted indicate that you
can't distinguish between the lossy and lossless file. More info on
interpreting the ABX test here:

http://www.hydrogenaudio.org/forums/index.php?showtopic=16295

a little on p-values and testing:
http://en.wikipedia.org/wiki/P-value


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> Aha. You're interpreting the test results incorrectly. (You must be
> thinking that a high percentage is good...in fact a high percentage
> means something very different here) The log you posted indicates that
> you can't distinguish between the lossy and lossess tracks. You would
> want to see a % (p-value) of, say, 5% or less. But I'd be impressed with
> even 10% or less. bottom line, the results you posted indicate that you
> can't distinguish between the lossy and lossless file. More info on
> interpreting the ABX test here:
>
> http://www.hydrogenaudio.org/forums/index.php?showtopic=16295
>
> a little on p-values and testing:
> http://en.wikipedia.org/wiki/P-value
>
> Hmm... Did you have a look at all 3 logs I posted? The first two have a
> p-value of 5.5% and the last one 24%. I'm not sure if those percentages
> are indicating the p-value though. But 5.5% means, 8 out of 10 were
> guessed correctly.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> Hmm... Did you have a look at all 3 logs I posted? The first two have a
> p-value of 5.5% and the last one 24%. I'm not sure if those percentages
> are indicating the p-value though. But 5.5% means, 8 out of 10 were
> guessed correctly.

I only saw the one log you posted. the one with 24%. I'll go back and
look. So yes, 5.5% means you generally could distinguish...


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> Hmm... Did you have a look at all 3 logs I posted? The first two have a
> p-value of 5.5% and the last one 24%. I'm not sure if those percentages
> are indicating the p-value though. But 5.5% means, 8 out of 10 were
> guessed correctly.

that's not exactly the interpretation. 5.5 means that there is only a
5.5% probability (about 1 out of 20) that you correctly identified the
source (lossy) is from pure chance. But yes, the results indicate that
one can reject the null hypothesis that you can't distinguish between
lossy and lossless. The 5.5 demonstrates that you can with very little
chance of this being a "random" result.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

garym wrote:
> that's not exactly the interpretation. 5.5 means that there is only a
> 5.5% probability (about 1 out of 20) that you correctly identified the
> source (lossy) is from pure chance. But yes, the results indicate that
> one can reject the null hypothesis that you can't distinguish between
> lossy and lossless. The 5.5 demonstrates that you can with very little
> chance of this being a "random" result.

And without getting two bogged down in statistics, when one runs
multiple tests, the p-values have to be adjusted to account for this.
(think of the idea that one asserts he can correctly predict heads or
tails in a coin flip 10 times in a row. If one runs this test enough
times, eventually in one of the tests, the subject will be able to
predict heads/tails 10 times in a row.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> And without getting two bogged down in statistics, when one runs
> multiple tests, the p-values have to be adjusted to account for this.
> (think of the idea that one asserts he can correctly predict heads or
> tails in a coin flip 10 times in a row. If one runs this test enough
> times, eventually in one of the tests, the subject will be able to
> predict heads/tails 10 times in a row.

Ok, I tested the same track as this morning again. This time I tried to
be more focused and clicked the answer only when I was really sure. Is a
p-value of 1.1% in the second shot enough to eliminate the doubts on the
first with 5.5%?

I suspect that it could be so easy for me to judge this track because
I'm playing the piano myself. Probably I know how to listen to it. And
there is nothing hi-res here. No matter how you treat a piano (if your
aim still is music making), there is not much left above 12kHz. And on
top of all, I created this last log on my laptop's headphone out with a
standard Sennheiser in-ear plug.

There are other tracks where I gloriously fail. But nevertheless, I
never thought that I would be able to distinguish FLAC and 320kbps.
Honestly I was a bit afraid that I wouldn't. Now I'm happy and - at
least under certain circumstances - I do hear a difference.

Maybe I should see a doc concerning this:
http://en.wikipedia.org/wiki/Impostor_syndrome

;-)


+-------------------------------------------------------------------+
|Filename: Roll2.txt |
|Download: http://forums.slimdevices.com/attachment.php?attachmentid=14671|
+-------------------------------------------------------------------+

------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> Ok, I tested the same track as this morning again. This time I tried to
> be more focused and clicked the answer only when I was really sure. Is a
> p-value of 1.1% in the second shot enough to eliminate the doubts on the
> first with 5.5%?
>
> I suspect that it could be so easy for me to judge this track because
> I'm playing the piano myself. Probably I know how to listen to it. And
> there is nothing hi-res here. No matter how you treat a piano (if your
> aim still is music making), there is not much left above 12kHz. And on
> top of all, I created this last log on my laptop's headphone out with a
> standard Sennheiser in-ear plug.
>
> There are other tracks where I gloriously fail. But nevertheless, I
> never thought that I would be able to distinguish FLAC and 320kbps.
> Honestly I was a bit afraid that I wouldn't. Now I'm happy and - at
> least under certain circumstances - I do hear a difference.
>
> Maybe I should see a doc concerning this:
> http://en.wikipedia.org/wiki/Impostor_syndrome
>
> ;-)

seems clear that you can distinguish well with this track. But you
yourself have noted why this might be the case for this particular
track. There have always been problem tracks that people can tell
differences in. And by "problem track" I just mean a track with
something unusual that makes it easy to tell or with harpsichords, etc..
This is why testing with real music, multiple tracks, and multiple
genres is always a good idea.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

garym wrote:
> seems clear that you can distinguish well with this track. But you
> yourself have noted why this might be the case for this particular
> track. There have always been problem tracks that people can tell
> differences in. And by "problem track" I just mean a track with
> something unusual that makes it easy to tell or with harpsichords, etc..
> This is why testing with real music, multiple tracks, and multiple
> genres is always a good idea.

True, he can distinguish the 2 tracks based on the ABX results, but as
far as I can tell, there are still too many unanswered questions... As I
suggested above, MP3 needs to be tested on CD data, not hi-res input
shoe-horned into a lossy encoder after SRC and bit-depth reduction.

Also, I didn't know that "There Is Always One More Time" is available as
24/192. Is there a DVD-A out there?


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Archimago wrote:
> True, he can distinguish the 2 tracks based on the ABX results, but as
> far as I can tell, there are still too many unanswered questions... As I
> suggested above, MP3 needs to be tested on CD data, not hi-res input
> shoe-horned into a lossy encoder after SRC and bit-depth reduction.
>
> Also, I didn't know that "There Is Always One More Time" is available as
> 24/192. Is there a DVD-A out there?

I have posted 4 ABX logs so far. No. 1 and 4 are done with Roll, Roll,
Roll. But no. 2 is a track from a Händel Organ Concerto and no.3 is a
guitar tune by Fernando Sor.

Just the last one was hi-res. The others were all 16/44 for Flac and
MP3.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Quad wrote:
> I have posted 4 ABX logs so far. No. 1 and 4 are done with Roll, Roll,
> Roll. But no. 2 is a track from a Händel Organ Concerto and no.3 is a
> guitar tune by Fernando Sor.
>
> Just the last one was hi-res. The others were all 16/44 for Flac and
> MP3.

That's really good!

You should get onto Hydrogen Audio and be one of their CODEC testers!


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Archimago wrote:
> That's really good!
>
> You should get onto Hydrogen Audio and be one of their CODEC testers!

I agree. There are people with uniquely good* hearing for these sorts of
things and you appear to be one of them. (And you've documented this
with ABX tests, unlike the many other "audiophiles" that simply assume
they have really good hearing/revealing systems without doing any blind
testing!)

*or "bad" in the sense that as posted elsewhere, in some cases actual
problems with hearing can make the listener better able to distinguish
lossy from lossless.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> I have posted 4 ABX logs so far. No. 1 and 4 are done with Roll, Roll,
> Roll. But no. 2 is a track from a Händel Organ Concerto and no.3 is a
> guitar tune by Fernando Sor.
>
> Just the last one was hi-res. The others were all 16/44 for Flac and
> MP3.

Ah. Ok. So to clarify, if we ignore the hires downsample track, you
tested 3 tracks and on two of them you were able to distinguish based on
ABX test. Correct?


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> Ah. Ok. So to clarify, if we ignore the hires downsample track, you
> tested 3 tracks and on two of them you were able to distinguish based on
> ABX test. Correct?

Exactly. Not more and not less.

garym wrote:
> I agree. There are people with uniquely good* hearing for these sorts of
> things and you appear to be one of them. (And you've documented this
> with ABX tests, unlike the many other "audiophiles" that simply assume
> they have really good hearing/revealing systems without doing any blind
> testing!)
>
> *or "bad" in the sense that as posted elsewhere, in some cases actual
> problems with hearing can make the listener better able to distinguish
> lossy from lossless.

I actually like the assumption posted in that other blog. If a codec
thinks something is overshadowed by other frequencies and decides to
eliminate that, this might just work for the average (good) ear. As soon
as the the information which supposedly should be dominant is longer
audible, we might have a problem.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> I actually like the assumption posted in that other blog. If a codec
> thinks something is overshadowed by other frequencies and decides to
> eliminate that, this might just work for the average (good) ear. As soon
> as the the information which supposedly should be dominant is longer
> audible, we might have a problem.

exactly! Thanks.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

I still don't have an optical cable. But I A-B-testet:

- Touch -> coax -> M51
- foobar -> HDMI -> M51

The M51 has one volume control for all inputs and I couldn't measure the
db-level due to the lack of a pressure meter.

While these are not optimal double blind conditions, it still is a
controlled straight forward instantaneous A-B test. My wife switched
sources for me.

Unfortunately I can't post a log, but differences are there, easier to
identify than FLAC vs. 320kbps.

What could be the reasons for that? I really want to know.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> I still don't have an optical cable. But I A-B-testet:
>
> - Touch -> coax -> M51
> - foobar -> HDMI -> M51
>
> The M51 has one volume control for all inputs and I couldn't measure the
> db-level due to the lack of a pressure meter.
>
> While these are not optimal double blind conditions, it still is a
> controlled straight forward instantaneous A-B test. My wife switched
> sources for me.
>
> Unfortunately I can't post a log, but differences are there, easier to
> identify than FLAC vs. 320kbps.
>
> What could be the reasons for that? I really want to know.

any DSPs turned on in foobar2000 playback? Foobar is using your
computer's soundcard. Any processing going on there? I'm no expert on
HDMI (just use it for connecting things in my AV setup). But seems that
I've read that some systems using HDMI have some bitrate restrictions,
etc. Really don't know. I suspect something with the computer
soundcard in the foobar setup. (the reason I found squeezeboxes years
ago to replace my foobar2000 on laptop > DAC > stereo was that the
soundcard on my laptop was doing nasty things with my music....wasn't
bit perfect when it arrived at the DAC).


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Quad wrote:
> I still don't have an optical cable. But I A-B-testet:
>
> - Touch -> coax -> M51
> - foobar -> HDMI -> M51
>
> The M51 has one volume control for all inputs and I couldn't measure the
> db-level due to the lack of a pressure meter.
>
> While these are not optimal double blind conditions, it still is a
> controlled straight forward instantaneous A-B test. My wife switched
> sources for me.
>
> Unfortunately I can't post a log, but differences are there, easier to
> identify than FLAC vs. 320kbps.
>
> What could be the reasons for that? I really want to know.

Hard to say Quad. What's the setup for the HDMI - off a graphics card,
motherboard? What computer are you using? I presume there's no sample
rate conversion, ReplayGain, or anything like that going on... As you
probably know, there's been some talk about jitter for HDMI on some gear
being in the multiple-nanosecond range (I think the M51 should be much
better but who knows how the computer-DAC interface functions in this
regard). Stereophile measurements look very good but they didn't bother
checking the HDMI interface.

I presume the Touch is at 100% volume, no transcoding/bit rate
limiting/server resampling.

Here's something you can also try once you get a TosLink - connect BOTH
of the Touch's coaxial and TosLink to the M51 and verify if the
interfaces sound about the same through the same player before
connecting the CDP vs. Touch.


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Archimago wrote:
> Hard to say Quad. What is the difference you're hearing?
>
> What's the setup for the HDMI - off a graphics card, motherboard? What
> computer are you using? I presume there's no sample rate conversion,
> ReplayGain, or anything like that going on in foobar as garym
> suggested... I don't know if any HDMI interface has ASIO to insure
> bit-accurate output without going through an internal resampler,
> dithering, and volume control. DirectSound IMO *always* messes up the
> output - easily demonstrated when I do jitter tests.
>
> As you probably know, there's been some talk about jitter for HDMI on
> some gear being in the multiple-nanosecond range (I think the M51 should
> be much better but who knows how the computer-DAC interface functions in
> this regard). Stereophile measurements look very good but they didn't
> bother checking the HDMI interface.
>
> I presume the Touch is at 100% volume, no transcoding/bit rate
> limiting/server resampling.
>
> Here's something you can also try once you get a TosLink cable - connect
> BOTH of the Touch's coaxial and TosLink to the M51 and verify if the
> interfaces sound the same through the same player before connecting the
> CDP vs. Touch.
>
> I still think an SPL meter is important to make sure volume differences
> isn't creeping into the evaluation.

It's hard to describe the difference in words. I will necessarily sound
like always.... But I'll give it a try.

In favor of the Touch:
- PC sounds somehow thinner
- Touch is somehow warmer

In favor of the PC:
- sounds more effortless
- harsh recording passages have less ringing (piano, violin)
- less fatiguing

My hardware is an Intel HD 3000 with HDMI out (integrated into Sandy
Bridge processor). And as far as I can remember WASAPI is the
bit-perfect Windows alternative to ASIO (Steinberg). I get the same
results if connected through USB and WASAPI. There is no processing
involved, neither on LMS nor on foobar.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

I'll try to put it in another way.

If played on the Touch, a piano attack sounds like if the the upper-mids
and treble appear first. After that first sound impression, the mids and
bass (and with that the sound stage) will be audible.

Played on PC, everything is together on time leading to a less
'compressed' experience. Does that sound familiar to anyone?


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Or explained as a consequence.

On the Touch I have to listen with higher volumes in order to achieve a
nice sound stage, because it is somehow overshadowed by upper-mids and
treble. With louder passages, this can lead to a slightly screaming
impression.

While on the PC, on low and high volumes sound stage is perfect from the
beginning of the note.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Quad wrote:
> Or explained as a consequence.
>
> On the Touch I have to listen with higher volumes in order to achieve a
> nice sound stage, because it is somehow overshadowed by upper-mids and
> treble. With louder passages, this can lead to a slightly screaming
> impression.
>
> While on the PC, on low and high volumes sound stage is perfect from the
> beginning of the note.

Interesting description... Can't say I've ever experienced this. Not
sure how this can be a function of presumably bit-perfect data transfer.
This delaying of frequency response also doesn't seem to jive with
jitter effects around timing. Almost sounds like you're saying the high
frequency tones arrive a little faster than the rest of the sound.

Does the M51 have adjustable digital filters? I guess severe pre-ringing
might cause something like this?


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Wombat]

Archimago wrote:
> Interesting description... Can't say I've ever experienced this. Not
> sure how this can be a function of presumably bit-perfect data transfer.
> This delaying of frequency response also doesn't seem to jive with
> jitter effects around timing. Almost sounds like you're saying the high
> frequency tones arrive a little faster than the rest of the sound.
>
> Does the M51 have adjustable digital filters? I guess severe pre-ringing
> might cause something like this?
No, pre-ringing for sure doesnŽt delay any sounds. The name is
misleading. If anything smears or alters that behaviour it must be some
non-linear nonsense that alters lower frequencies.


------------------------------------------------------------------------
Wombat's Profile: http://forums.slimdevices.com/member.php?userid=4113
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Wombat wrote:
> No, pre-ringing for sure doesnŽt delay any sounds in the clearly audible
> range with a normal acting linear filter. If anything smears or alters
> that behaviour it must be some non-linear nonsense that alters lower
> frequencies.

Good point...


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Wombat]

Archimago wrote:
> Good point...
To be honest i really have a problem finding a reasonable reason for a
beahviour that creates such an effect described. There comes me some
tape pre-echo to mind you could hear back in the good old days :)


------------------------------------------------------------------------
Wombat's Profile: http://forums.slimdevices.com/member.php?userid=4113
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

Wombat wrote:
> To be honest i really have a problem finding a reasonable reason for a
> beahviour that creates such an effect described. There comes me some
> tape pre-echo to mind you could hear back in the good old days :)

Me neither, I can't find a reasonable reason. I don't like that.

Of course the effect is very subtle, not comparable to seconds of tape
pre-echo.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> seems clear that you can distinguish well with this track. But you
> yourself have noted why this might be the case for this particular
> track. There have always been problem tracks that people can tell
> differences in. And by "problem track" I just mean a track with
> something unusual that makes it easy to tell or with harpsichords, etc..
> This is why testing with real music, multiple tracks, and multiple
> genres is always a good idea.
>
> edit: not that harpsichords aren't real music (or your own playing is
> not real music). ;-)

You're perfectly right. If someone wants to test if people in general
can benefit from FLAC for the majority of music, then my tests are
useless. For an answer to this question Archimago's blog is pretty clear
about that.

But my personal null hypothesis was: Under no circumstances I'm able to
hear any difference between FLAC and 320kbps on a properly converted
radio-friendly track.
For this case, the null hypothesis has to be rejected.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> But my personal null hypothesis was: Under no circumstances I'm able to
> hear any difference between FLAC and 320kbps on a properly converted
> radio-friendly track.
> For this case, the null hypothesis has to be rejected.

correct. but you should really follow archimago's advice about using a
straight CD > mp3 compared with a CD > FLAC track without the
downsampling, etc. to get to mp3 first. Much cleaner test without the
possibility of these other confounding effects.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Archimago]

Quad wrote:
> Even harder, but still discernible.
>
> The original here is 24/192 while the MP3 is 24/48, downsampled with the
> built-in foobar SRC. I might hear the donwsampling algorithm and not the
> codec.

Quad, why don't you try downsampling with the best downsampler you have
to 16/44 and make a 320kbps MP3 out of that. Then compare the 16/44
with the MP3 produced?

MP3 isn't really capable of 24-bit encoding and as you suggested,
there's alot going on with the SRC in play as well as bit-depth
reduction including probably addition of dither which if not done right
could introduce aliasing effects.


------------------------------------------------------------------------
Archimago's Profile: http://forums.slimdevices.com/member.php?userid=2207
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[darrenyeats]

Julf wrote:
> Didn't someone show that spotify was using dynamic compression?

As far as I can tell they don't.

However, there are PLENTY of examples where the master they have ripped
is different to the master for my particular CD rip (and as always this
can manifest in dynamic compression, EQ or other more subtle
differences).
Darren


------------------------------------------------------------------------
darrenyeats's Profile: http://forums.slimdevices.com/member.php?userid=10799
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Julf wrote:
> Didn't someone show that spotify was using dynamic compression?

Yes, but I can't locate the info where I saw that right now. Definitely
some evidence was provided that documented that there was additional
processing of the Spotify tracks. So any mp3 > FLAC comparison should
definitely be using an mp3 file that the person ripped themselves vs
spotify (or even an mp3 file of unknown origin).


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[darrenyeats]

I'm not sure I understand this conversation ... Spotify is Ogg Vorbis?


------------------------------------------------------------------------
darrenyeats's Profile: http://forums.slimdevices.com/member.php?userid=10799
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

darrenyeats wrote:
> I'm not sure I understand this conversation ... Spotify is Ogg Vorbis?

correct. So none of these mp3 vs FLAC tests are valid if the comparison
is to spotify (not to mention the evident extra compression folks have
found in spotify files)


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[darrenyeats]

garym wrote:
> correct. So none of these mp3 vs FLAC tests are valid if the comparison
> is to spotify (not to mention the evident extra compression folks have
> found in spotify files)

Please see my previous post re: dynamic compression.


------------------------------------------------------------------------
darrenyeats's Profile: http://forums.slimdevices.com/member.php?userid=10799
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

darrenyeats wrote:
> Please see my previous post re: dynamic compression.

got it. can't seem to locate it (maybe it was a different forum), but
someone had done digital captures, etc. to try and show this
compression. But as you point out use of different masters could also
explain this (particularly if one uses a more recent CD that suffers
from loudness war mastering). And bottom line, use of spotify (or any
service) is not a fair test....definitely should be comparing self
created mp3 to FLAC (where both were created from the same source CD).


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

SoftwireEngineer wrote:
> I have some test tracks (Chesky, I think) which have recordings of
> instruments at different distances from the mic. It also has instruments
> at different height relative to the mic. I or in my system I dont think
> I can notice the height very well. But the distance is very discernible
> if the distance is more than 5-10 ft(vague memory). So something like
> this could help evaluate the "room" in the original post. "Air" I think
> is pretty much decay. We can probably measure the decay trail too. But I
> doubt in dbs it will be much significant.
> Re: FLAC vs 320Kbps. I have not done much testing but Spotify is easily
> discernible for me from my original FLAC tracks. The music sounds flat
> (no depth) and I think there is also some sort of "haze" or vagueness to
> the sound. My FLACs when the system is at its best (late night/quiet
> etc) sounds very 'crisp'.

thanks. that makes more sense (i.e., use of the test tracks). Keep in
mind that when I reference tests showing transparency of mp3 at high
enough bit rate, I'm talking about tests where people are listening to
real music (not test tones or test tracks such as you've mentioned).
However, many different genres of real music are used.

And I'm sure you've seen the other posts re Spotify. Using Spotify files
is not a good test (these are OGG VORBIS transcoded by the squeezebox
and are likely to be subject to some sort of extra compression even
before all that). You should use an mp3 file that you created (so you
know the source).


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[callesoroe]

garym wrote:
> These are not encoders. Within mediamonkey or foobar2000 are you
> encoding mp3 using LAME (I assume). And if so, VBR. CBR, ABR? Although
> if 320, almost any LAME mp3 file *should* be transparent.

Lame and CBR


------------------------------------------------------------------------
callesoroe's Profile: http://forums.slimdevices.com/member.php?userid=22693
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[DJanGo]

Quad wrote:
> Not audible to me:
> - LMS running on different OS (Windows, OS X, Linux)
>

hmm just the idea diff OS (on same Hardware can make a diff sound) is
like "since i use octane booster on my lawnmower my gras is greener and
grows better"

Quad wrote:
>
> Audible to me:
> - RCA cables
> - speaker cables
> - power cables and power supplies
> - differences between DACs
> - differences between sampling rate conversion algorithms
> > native 192kHz with EDO sounds different than downsampled with SOX to
> 96kHz (either on-the-fly or not)
> > SACD ripps with different DSD->PCM settings sound different (POW-R3
> vs. TPDF)
> - differences in digital source
> > Touch sounds different than my average CD player NAD 514 (S/PDIF
> into same external DAC NAD M51)
> > Touch and NAD 514 sound different than Digital Music Player NAD M50
> (S/PDIF into same external DAC)
> > HDMI out of my Laptop into same external DAC sounds different again
> (WASAPI event).
>
you forget same Hardware in a other room sound diff too or more pictures
on the wall other sound ...
Dont mess apples with beans :-(

In case your IN the specs your in the specs and the human ears are like
the human eyes.
If 100 People should compare colours you ve got 100 diff answers and
believe it or not they are all "right" cause everyone see colors diff
than any other.
Same with the ears.

What do i mean with in the specs?
If you (in case its possible but forbidden) put one powerhub into the
wall and cascade that to one other powerhub you will hear that.
Its forbidden but you can do it.
Same if you wrap your Powercable over the speaker cables you will hear
it. But no one who can count up to 3 will do that.

In case your Hardware is bad connected and you had an grounding Problem
- you will hear that too.
Mostly older notebooks can make a sound if you click the mouse or use
the scrollwheel.
If your Hardware is underpowered - you will hear that.

If you got an original CD that is scratched and you ripp it with EAC and
burn it as Redbook again - it could sound better than the original.
If you (and thats the only thing i know about that cheap is better) got
an bad CD i can be that a cheaper Player will play it and a highcass
player wont.

So what do you want to "hear" from us?


------------------------------------------------------------------------
DJanGo's Profile: http://forums.slimdevices.com/member.php?userid=1516
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

DJanGo wrote:
> hmm just the idea diff OS (on same Hardware can make a diff sound) is
> like "since i use octane booster on my lawnmower my gras is greener and
> grows better"

Just look around in this forum (remember, it's called "Audiophile") and
you will find lots of posts about differences in OS. But I'm completely
with you. This is just impossible to happen.

With my list I just wanted to distance myself from the esoteric kind of
audiophiles.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

It's a long time ago that I did my last comparisons. And because of the
burden of proof I didn't question the transparency of 320kbps.

But now I just re-tested. I took 5 tracks I know well and converted them
to 320kbps CBR with lame 3.99.5. Then I put the original and the MP3
together in a foobar playlist and let it play randomly. I listened to
all of the tracks for a couple of seconds/minutes and and tried identify
the MP3s.

Result: 10 of 10 were guessed right.

Do I have to switch sides now and turn into an esoteric? I'm really
confused.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> It's a long time ago that I did my last comparisons. And because of the
> burden of proof I didn't question the transparency of 320kbps.
>
> But now I just re-tested. I took 5 tracks I know well and converted them
> to 320kbps CBR with lame 3.99.5. Then I put the original and the MP3
> together in a foobar playlist and let it play randomly. I listened to
> all of the tracks for a couple of seconds/minutes and and tried to
> identify the MP3s.
>
> Result: 10 of 10 were guessed right.
>
> Do I have to switch sides now and turn into an esoteric? I'm really
> confused.

Impressive. But a better test would be to install the Foobar2000 ABX
component and do a regular ABX test. Try it on the same 5 tracks. (In
your test, the volume difference (even slight) could explain the ability
to pick. The ABX component will control for these things.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> Impressive. But a better test would be to install the Foobar2000 ABX
> component and do a regular ABX test. Try it on the same 5 tracks. (In
> your test, the volume difference (even slight) could explain the ability
> to pick. The ABX component will control for these things.

I just tested 3 of my tracks with the ABX plugin for foobar. 2 of them
could easily be assigned within seconds with 100% accuracy. One was a
bit harder. As soon as I have some time left I will take these tests a
bit further. I didn't expect that, I really didn't.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> I just tested 3 of my tracks with the ABX plugin for foobar. 2 of them
> could easily be assigned within seconds with 100% accuracy. One was a
> bit harder. As soon as I have some time left I will take these tests a
> bit further. I didn't expect that, I really didn't.

Thanks. Interesting. How many trials are you doing (10?). It is true
that some songs achieve better transparency that others. So probably it
is a good idea to do several (maybe even more than 5 and certainly try
different genres). Thanks again for going through the trouble of doing
this. (and by the way, are you under 30 or 40? ....just asking because
of high freq hearing and age, etc.)


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> Thanks. Interesting. How many trials are you doing (10?). It is true
> that some songs achieve better transparency that others. So probably it
> is a good idea to do several (maybe even more than 5 and certainly try
> different genres). Thanks again for going through the trouble of doing
> this. (and by the way, are you under 30 or 40? ....just asking because
> of high freq hearing and age, etc.)

Yes, 10 trials.

It's funny that one record - of which I thought I was able to hear a
difference - almost failed the test, while another one - which I first
doubted to be distinguishable - is quite easy to assign.

Both are piano records, but different genres:
distinguishable: B.B. King - There is always one more time - 08 - Roll,
Roll, Roll (2009)
(almost) not distinguishable: Chopin - Maria João Pires - Nocturne No.1
(1996)

I will do more, extended tests across genres.


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> Yes, 10 trials.
>
> It's funny that one record - of which I thought I was able to hear a
> difference - almost failed the test, while another one - which I first
> doubted to be distinguishable - is quite easy to assign.
>
> Both are piano records, but different genres:
> distinguishable: B.B. King - There is always one more time - 08 - Roll,
> Roll, Roll (2009)
> (almost) not distinguishable: Chopin - Maria João Pires - Nocturne No.1
> (1996)
>
> I will do more, extended tests across genres.

Excellent, can you paste in the table of the results when you do this,
with the p-values. This is more useful that saying "failed or passed" as
there really is no fail or pass, just a p-value on the comparisons
(hypothesis being tested). The complete table of the results of the
trials is just more informative (I'm not suggesting that you're
misleading us, just interested in the details). Thanks.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Quad]

garym wrote:
> Excellent, can you paste in the table of the results when you do this,
> with the p-values. This is more useful that saying "failed or passed" as
> there really is no fail or pass, just a p-value on the comparisons
> (hypothesis being tested). The complete table of the results of the
> trials is just more informative (I'm not suggesting that you're
> misleading us, just interested in the details). Thanks.

Are p-values and result tables part of the automatically generated ABX
log? That's all I can post, I'm not a scientist. Give me some time...


------------------------------------------------------------------------
Quad's Profile: http://forums.slimdevices.com/member.php?userid=20234
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[garym]

Quad wrote:
> Are p-values and result tables part of the automatically generated ABX
> log? That's all I can post, I'm not a scientist. Give me some time...

yep, the log, thanks.


------------------------------------------------------------------------
garym's Profile: http://forums.slimdevices.com/member.php?userid=17325
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Mnyb]

garym wrote:
> Impressive. But a better test would be to install the Foobar2000 ABX
> component and do a regular ABX test. Try it on the same 5 tracks. (In
> your test, the volume difference (even slight) could explain the ability
> to pick. The ABX component will control for these things.

+1

With the very very high confidence in 320kBps being not audible there
are probably countless test performed that actually proves that .

If you for some reason score a positive at at home , there is a
overwhelmingly large possibility that you actually have an experimental
error not a true positive .
So if anyone (like me ) shows some scepticisms it is because of that ,
nothing personal .

And I'm not trying to be smug either , I just believe very much in the
use of scientific methods . I believe you when yours and a couple of
others peer reviewed papers are published :) that shifts the large
knowledge base we have on this .
Or include me in a high quality experiment to see for myself .

There are many possible experimental errors .

In the case of a squeezebox I would server decode mp3 As I have found
the on-board decoder not perfect in all circumstances , it can be quite
audible on some rare tracks .
So the squeezebox performance with mp3 is not as good as mp3 itself ( a
property it shares with most small mp3 players and phones ,basically
everything that’s not a computer ) .
I must point out that there is an explanation (read in our bugzilla and
some older posts ) .

It is not the same as those that believes that FLAC should be server
decoded that is fairy tales . The flac encoder on the squeezebox makes
bitperfect output and the cpu loads seems to be truly insignificant for
the output (as recently measured )

And ofcourse levels , what would an mp3 encoder do is levels actually
preserved ?

Therefore I'm very grateful for Archimagos group test that he provided ,
it had perhaps many "holes" it would not pass as science really ,but he
controlled the file conversion parameters for us and made sure that
there was nothing that could skew the results very much in the files
themselves .

And what about the physiological variations among humans ? Could there
be some unusual person that can hear this ? Some properties in hearing
suggest not , for example frequencies above >20kHz the ear seems not to
conduct that (bone hearing trough implanted metal screws is the way ).
Can you run 100m on 8 seconds too ? What i mean that the results are
good enough to practically include everyone .

But if there are such freak individuals it would sure not be every
audiophile that post on the internet and bash mp3 or redbook CD's they
may exist but be rarer than that .

Btw I do use FLAC almost exclusively if anyone wondered .


------------------------------------------------------------------------
Mnyb's Profile: http://forums.slimdevices.com/member.php?userid=4143
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[darrenyeats]

I don't follow this conversation. The pre-ringing in modern DACs is
really really tiny. You're saying you can hear this?
Darren

Sent from my HTC Sensation Z710e using Tapatalk 2


------------------------------------------------------------------------
darrenyeats's Profile: http://forums.slimdevices.com/member.php?userid=10799
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[mlsstl]

As a somewhat different view, over the years I've noticed sometimes
people assign convoluted or interesting descriptions and explanations to
what they hear. For example, researchers have known for decades that a
slight difference in volume doesn't come off as one source being louder,
but rather clearer or better sounding. Similarly, there can be
differences in frequency response and other commonplace factors that are
just perceptible but still so subtle that the root difference can't be
easily identified.

Combine that with the very human trait of expectation bias and other
subjective influences plus the natural tendency of audiophiles to look
for the most exotic explanation possible and you end up with some very
unusual observations.


------------------------------------------------------------------------
mlsstl's Profile: http://forums.slimdevices.com/member.php?userid=9598
View this thread: http://forums.slimdevices.com/showthread.php?t=98374

[Julf]

mlsstl wrote:
> As a somewhat different view, over the years I've noticed sometimes
> people assign convoluted or interesting descriptions and explanations to
> what they hear. For example, researchers have known for decades that a
> slight difference in volume doesn't come off as one source being louder,
> but rather clearer or better sounding.

Absolutely. In a couple of blind listening tests I have used a control
track that is 1 dB louder than the others. It usually comes out as the
winner, even if it is lower resolution than some of the other tracks.


------------------------------------------------------------------------
Julf's Profile: http://forums.slimdevices.com/member.php?userid=42050
View this thread: http://forums.slimdevices.com/showthread.php?t=98374