[Mauricio]

Something to keep always firmly in mind is that sampling is not a lossy process for the frequency range in question, so long as the sampling rate is at least twice the maximum frequency of the frequency range.  Said another way, a sampling rate of, say, 44.1kHz provides for a lossless (read, perfect, identical) reproduction of a signal at or below a frequency of 22.05kHz.  Above 22.05kHz the process lossy, yes.  Below 22.05kHz, the process of reconstructing the analog signal is lossless, perfect.  In the range of human hearing, between 20-20,000kHz, a sampling rate of 44.1kHz provides for a perfect, lossless representation of the signal.  This talk of "connecting the dots" could lead many to believe that there is some lossy, error-introducing interpolation going on when, in fact, no such thing takes place at all.  This is established scientific fact.

 

[stv014]

Quote:

Originally Posted by jcx /img/forum/go_quote.gif
 
2 lsb p-p triangle dither gives ~ 93 dB  S/N, is used when triangle is selected in Audacity - TPDF dither is pretty universal in DAW, preferred over noise shaped dither when more processing after dither is anticipated

 
Ideally, if the sound is to be processed further, it should not be saved in 16 bit format.
 

 

[jcx]

certainly that's best - at "high enough" word length - where how long depends on how much processing may occur
 
 
with all of the 24 bit DACs today you also as a practical matter don't need to redither 16 bit source with digital Volume or even most "reasonable" "consumer" processing like room or headphone corrective EQ if you can set the system to keep in 24 bit after the DSP
 
with "final step, consumer playback" 24-32 bit processing, 24 bit DAC even with 16  bit source there is no audible "bit loss" from the bits falling below the 24 bit word length - the system's analog electrical noise in DAC, I/V will always be high enough to totally mask any numerical digital "bit loss"

 

[kiteki]

Quote:

It was also in a room with an ambient volume of 19 dB. That's lower than normal. I think my room has an ambient volume of ~40 dB. So to hear what he heard, I'd have to play at 116 dB, with 132 dB peaks. Most rooms are probably around ~30 dB if they're not specifically designed to be quiet so say 106 dB with 122 dB peaks. No one should listen that loud.

Actually it's closer to 95dB.
 
"[size=x-small]The overall system gain is an essential factor in these experiments, or in any attempts to duplicate the work. Our standard system gain was calibrated using an octave of pink noise recorded at -16 dBFS, which produced a wideband SPL of 85 dB at the listening chair.[/size]"
 
"[size=x-small]One of the authors, using a short repeated section of room tone on the Hartke disc mentioned above, obtained a positive result (15/15) at a gain of only 10 dB above our standard level. [/size]"
 
source: http://www.bostonaudiosociety.org/explanation.htm
 
It's valid to note this test only covers the transparency of A/D conversion, not D/A conversion.  For example if there was a NOS DAC with -6dB at 16kHz in their playback system, then you'd hear that on the first CD/SACD plus the second CD, since they are played via the same system.  Likewise all other signatures inherent to audio components via playback.
 
As for the rest of the study, it's referenced in Wiki http://en.wikipedia.org/wiki/Super_Audio_CD#Audible_differences_compared_to_PCM.2FCD, and this article http://people.xiph.org/~xiphmont/demo/neil-young.html#toc_lt, which seem to indicate that this study is the vast representative of the experimental record.
 
I've seen people say there are countless studies which have scientifically disproved SACD, hypersonic effect, 24/192, or even proven "total transparency" of ADC/DAC!  I can't find these "countless studies".
 
Lastly, it wasn't very scientific, there were several acute errors, which are outlined in the first three pages of this thread, with responses from the Author himself - http://www.sa-cd.net/showthread.php?page=1
 
i.e.
- only 4 or 5 discs out of ~20 were actually SACD (DSD) content!
- they set their "confidence level" at a 95% success rate!
 
 
There are - luckily - more scientific studies, such as this one - http://www.nhk.or.jp/strl/publica/labnote/lab486.html
 
However, that is not testing DSD content, only high-frequency content, and their conclusion is actually...
 
"From above results, we can still neither confirm nor deny the possibility that some subjects could discriminate between musical sounds with and without very high frequency components."
 
 
Here's another paper, discussing filters - http://www.physics.sc.edu/kunchur/temporal.pdf
 
Filters - http://www.cirrus.com/en/pubs/whitePaper/DS668WP1.pdf
 
 
A lot of papers which I haven't read or can't access without paying
 

Spoiler: why%20so%20many%20papers

Ultrasonic hearing and Hypersonic effect

Ashihara et al., “Detection threshold for tones about 22 kHz”, 110th AES convention 2001, preprint no. 5401

Ashihara “Hearing thresholds for pure tones above 16 kHz”, The Journal of the Acoustical Society of America 2007, Volume 122, Issue 3, pp. EL52-EL57

D. Griesinger, “Perception of mid frequency and high frequency intermodulation distortion in loudspeakers and its relationship to high-definition audio”, 24th International AES Conference 2003, Banff, Canada
www.davidgriesinger.com/intermod.ppt

Hamasaki et al., “Perceptual Discrimination of Very High Frequency Components in Musical Sound Recorded with a Newly Developed Wide Frequency Range Microphone”, 117th AES convention 2004, preprint no. 6298

Hosoi et al., “Activation of the auditory cortex by ultrasound”, The Lancet, vol. 351, Febr. 14, 1998, p.496

Ishiuchi et al., “Difference tone produced by two ultrasonic components”, Proceedings of the 11th International Symposium on Biotelemetry, Sep 1990, Yokohama, Japan

Lenhardt, “Ultrasonic hearing in humans: applications for tinnitus treatment”, Int. Tinnitus J., vol.9, no.2, pp.69-75 (2003)

Lenhardt, “Eyes as fenestrations to the ears: a novel mechanism for high-frequency and ultrasound hearing”, Int. Tinnitus J., vol.13, no.1, pp.3-10 (2007)

Nakamura et al., “Analysis of music-brain interaction with simultaneous measurement of regional cerebral blood flow and electroencephalogram beta rhythm in human subjects”, Neuroscience Letters 275 (1999), p.222-226

Nishiguchi et al., “Perceptual discrimination between music sounds with and without very high frequency components”, 115th AES convention 2003, preprint no. 5876

Nishiguchi et al., „Perceptual discrimination of very high frequency components in wide frequency range musical sound“, Applied Acoustics 2009, vol. 70, p.921

Nishimura et al., “Ultrasonic masker clarifies ultrasonic perception in man”, Hearing Research 2003, Volume 175, pp.171-177

Omata et al., “A psychoacoustic measurement and ABR for the sound signals in the frequency range between 10 kHz and 24 kHz”, 125th Audio Engineering Society convention 2008, preprint no. 7566

Oohashi et al., “High frequency sound above the audible range affects brain electric activity and sound perception”, 91st AES convention 1991, preprint no. 3207

Oohashi et al., “Inaudible high-frequency sounds affect brain activity: hypersonic effect”, Journal of Neurophysiology 83 (2000), p. 3548-3558
http://www.linearaudio.nl/Documents/...on%20brain.pdf

Oohashi et al., “Multidisciplinary study on the hypersonic effect”, International Congress Series 1226 (2002), pp.27-42

Oohashi et al., “The role of biological systems other than auditory air-conduction in the emergence of the hypersonic effect”, Brain Research 1073-1074 (2006), p. 339-347

Sugimoto et al, “Human perception model for ultrasonic difference tones”, Proceedings of the 24th IASTED International Conference, Feb 16-18, 2005, Innsbruck, Austria

Yagi et al., “Auditory display for deep brain activation: hypersonic effect”,
Proceedings of the 2002 International Conference on Auditory Display, Kyoto, Japan, July 2-5, 2002
http://www.icad.org/websiteV2.0/Conf...gs/Oohashi.pdf

Yagi et al., “Modulatory effect of inaudible high-frequency sounds on human acoustic perception”, Neuroscience Letters 351 (2003), p.191-195

 
 
p.s. Lucky 777th post.
 
x10 multiplier bonus!
 
 

 

[Head Injury]

 

Actually it's closer to 95dB.
 
"[size=x-small]The overall system gain is an essential factor in these experiments, or in any attempts to duplicate the work. Our standard system gain was calibrated using an octave of pink noise recorded at -16 dBFS, which produced a wideband SPL of 85 dB at the listening chair.[/size]"
 
"[size=x-small]One of the authors, using a short repeated section of room tone on the Hartke disc mentioned above, obtained a positive result (15/15) at a gain of only 10 dB above our standard level. [/size]"
 
source: http://www.bostonaudiosociety.org/explanation.htm

Umm, what exactly are you trying to say? I know it was 95 dB. I also know that the ambient volume of the room was ~19 dB.
 

[size=small]

The background noise level in this room is lower than that in most urban listening rooms – 19 dBA.[/size]

 
 
That means the average volume was 76 dB above ambient volume. Say for argument's sake that this allows them to hear noise -76 dB down, while everything else is drowned out by ambient volume. I said my room has an ambient volume of ~40 dB. 76 + 40 = 116 dB average volume for me to be able to hear the same noise floor he heard.
 
The point is, not only was he listening louder than most people will, but he was doing so in a room that's quieter than most people will use. Everyone else will have to listen even louder because of that.
 
 

"From above results, we can still neither confirm nor deny the possibility that some subjects could discriminate between musical sounds with and without very high frequency components."

 
The reason this is their conclusion is because you can't prove something doesn't exist. You have to prove that it does exist. They can't say it doesn't exist because of the results of one test. They can only say that their results don't suggest it exists. Now someone has to show that it does.

 

[jcx]

just for completeness I looked at Audacity triangle dither fft - it does apply a crude high pass, giving slightly better A weighted noise but the variance is the same as the unfiltered flat tpdf dither

 

[stv014]

Quote:

just for completeness I looked at Audacity triangle dither fft - it does apply a crude high pass, giving slightly better A weighted noise but the variance is the same as the unfiltered flat tpdf dither

If you differentiate uniform distribution white noise, the result is highpass filtered noise with triangular distribution. So that is probably what Audacity does.
 

 

[brod]

Quote:

Something to keep always firmly in mind is that sampling is not a lossy process for the frequency range in question, so long as the sampling rate is at least twice the maximum frequency of the frequency range.  Said another way, a sampling rate of, say, 44.1kHz provides for a lossless (read, perfect, identical) reproduction of a signal at or below a frequency of 22.05kHz.  Above 22.05kHz the process lossy, yes.  Below 22.05kHz, the process of reconstructing the analog signal is lossless, perfect.  In the range of human hearing, between 20-20,000kHz, a sampling rate of 44.1kHz provides for a perfect, lossless representation of the signal.  This talk of "connecting the dots" could lead many to believe that there is some lossy, error-introducing interpolation going on when, in fact, no such thing takes place at all.  This is established scientific fact.

If that's the case then why do consumer level 96kHz and 192kHz playback devices and digital media content exist?

 

[Head Injury]

Quote:

If that's the case then why do consumer level 96kHz and 192kHz playback devices and digital media content exist?

Because people are willing to buy it under the false pretense that it matters.

 

[Mauricio]

Quote:

If that's the case then why do consumer level 96kHz and 192kHz playback devices and digital media content exist?

As mentioned in the article, reasons include expectation bias or the placebo effect.  Another reason is that better sounding music resulting from good mastering is available only at those bit-depths and sampling rates.

 

[adamlau]

 

Another reason is that better sounding music resulting from good mastering is available only at those bit-depths and sampling rates.

Depends on the definition of good mastering. In the case of past and present pop, good as in replayable with consistency across general consumer devices (limited and denoised), or good as in a dynamic flat transfer from well-preserved tapes (in the case of analog recordings) with only a touch of EQ when required? 

 

[Mauricio]

For my tastes and choice of equipment, the latter.

 

[brod]

How about for virtualized headphone surround? My sound card takes a 5.1 or 7.1 mix from games/movies and converts it to two channels with HRTF based positional cues... would it be better to use 24/96 or 24/192 instead of 24/48?
 
http://www.creative.com/oem/technology/thx.asp

 

[stv014]

Quote:

just for completeness I looked at Audacity triangle dither fft - it does apply a crude high pass, giving slightly better A weighted noise but the variance is the same as the unfiltered flat tpdf dither

I checked it again, and that is in fact not the case, at least with the 1.3.12 version I tested. I implemented and tested both the white and the colored (differentiated) triangular noise, and the output of Audacity matches the former (93.3 dB RMS, 95.7 dB A-weighted vs. 93.3 dB RMS, 97.3 dB A-weighted). Although a simple FFT with a logarithmic frequency scale may show filtered noise, averaging it over the file makes the displayed response flat.
 

 

[jcx]

I've got Audacity 1.3 beta to be able to it run on Vista
 
its pretty clear that my version of Audacity's "triangle" dither is tilted when plotted along side the "flat" tpdf I created with the diff of 2 SciLab grand() sequences
 
code snippet - though I'm not certain which version: http://code.google.com/p/audacity/source/browse/audacity-src/trunk/src/Dither.cpp?spec=svn11480&r=11480
 
 

// Triangle dither - high pass filtered

inline float Dither::TriangleDither(float sample)

{

float r = DITHER_NOISE;

float result = sample + r - mTriangleState;

mTriangleState = r;

return result;

}

 
seems to be doing the 1st order delay difference "differential"  of the uniform dither input - thanks for the heads up, I hadn't seen this version of tpdf before
 
 
but while I would like to get this really right  - the basic claim that I was refuting is that dithered 16 bit audio has only 84 dB S/N
 

 
Quote:

Yes I know very much about dither & its usefullness at reducing distortion however it does this at the expense of increased noise which still limits your signal to noise ratio to about 84db ...

 
 
we both appear to agree that "proper" tpdf, and normally used definitions of S/N, dynamic range for audio show better than 90 dB for tpdf dithered 16 bit audio