"A passage that is 6dB louder than another passage is said to be twice as loud as the other passage."First obvious mistake - 6dB is four times louder, not two. Sound level goes as the square of pressure.
' date='Jul 5 2008, 13:12' post='575121']There are several small errors around the document, like "4-bit recording would have 16 discrete possible amplitude levels.". a 4bit recording just have 8 amplitude levels, because a signal has a positive and a negative part. This one is done several times.(And what about "Perhaps many are more familiar with 8-bit audio from real-time internet sources like RealAudio". that was audio compressed at 16kbit/s, not just "8-bit" !)But one of the things that made me wonder is how 24bits (as opposed to 16bits) actually makes vinyl lovers happier. AFAIR the SNR of a vinyl is lower (i.e. less range) than that of a CD. Either one doesn't like digital audio (and argues that just analog media can store the signal in enough detail), or accepts the way digital works, and compares what is comparable (i.e SNR)The problem with the document, from my point of view is: It says something that is acceptable for its use (recording), with some correct information, but also with other that are mistakes, misunderstandings or erroneous concepts.I am not implying that the latter are more prominent than the former. Just that they are there.
http://www.ews64.com/mcdecibels.htmlThe topic was dynamic range and PCM quantizations/representations of dynamic range, not sound pressure levels.
Further for anyone to say that 8-bit audio suffers from only noise and not distortion is a most ridiculous statement that really doesn't even deserve the lengths I've gone through to address it up to this point
dither should be applied to smooth out the artifacts from quantization noise generated by the loss of precision
The PCM format provides its optimal resolution when signal levels are at their very highest. As signal levels decrease to lower levels, resolution deteriorates, leaving quiet cymbals and string instruments sounding typically sterile, dry, harsh, and lifeless.
Quote from: pdq on 05 July, 2008, 11:22:44 AM"A passage that is 6dB louder than another passage is said to be twice as loud as the other passage."First obvious mistake - 6dB is four times louder, not two. Sound level goes as the square of pressure.http://www.ews64.com/mcdecibels.htmlThe topic was dynamic range and PCM quantizations/representations of dynamic range, not sound pressure levels.
Might have been nice to hear some of your suggestions for corrections over the past 7 years, especially if you were around back when the doc was written in 2001.
Quote from: ccryder on 06 July, 2008, 05:43:51 AMMight have been nice to hear some of your suggestions for corrections over the past 7 years, especially if you were around back when the doc was written in 2001.Yes it may be a little harsh to criticize in 2007 & 8 an article that contained the following qualification and request for input/feedback:[blockquote]While the contents of this document are specifically targeted to the needs and concerns of field recordists, some of the content can be applied to home recording as well. The submission of additions, corrections, and comments, is requested and encouraged ...[/blockquote]I imagine many people would have noticed the odd academic error in passing, but would have been content to read the article broadly; for its practical guidance, in using what was new technology at the time.
Regarding the Real-Audio reference, never did I mention the phrase "bit-rate". I said 8-bit. As in 8 bits per sample representing a maximum of 8 significant bits required to quantify dynamic range, *not* 8 bits per second. Bits per sample was the topic.
My experience with Real Audio by the time the doc was written was that Real Audio was played back using 8 *significant bits* per PCM sample (regardless of word lengths > 8-bits) once decompressed from Real Audio format and having been fed to a D/A converter.
If that has changed since then (I don't mess with Real Audio, so I don't know how far they've pushed its dynamic range losslessly), then any inaccuracies with respect to bits per sample of the Real Audio product would be a function of the information I believed accurate at the time it was written.
One might get a greater appreciation for what it is that makes audiophiles (and "vinyl lovers") prefer 24-bit audio by reading either some sections of my doc (if you dare), or reading more about how PCM vs. DSD represents sound, especially low level signals. Simply put, the attraction to either high resolution 24 PCM digital audio, or analog audio is a more accurate representation of lower level signals--DSD having its own argument in that regard.
The question of sampling rate and bit depth are a bit unclear to me.Sampling rate is the easiest to understand, as long as you can’t prove Nyquist wrong, a sampling rate double our hearing threshold is good enough. In real life there is a thing called technology so we have to deal with a couple of problems which don’t exist in the pure mathematical world. As Dunn phrased it quit nicely:A direct effect of the higher sampling rate is that for an identical filter design the timedisplacements will scale inversely with sample rate. Hence an improvement can bemade just from raising the sample rate - even for those who cannot hear above20kHz.As far as I knew, there is no theorem proving that a minimum as X bits is needed to reproduce all the details in the signal. We have no Nyquist for the bit depth.From another post in this forum I learned that flipping even the LSB in 16 bit PCM is clearly audible. I’m inclined to think that a resolution well above the hearing threshold is a bit to ‘rough’Listening to recordings at a higher resolution (24 bits) should bring an improvement in detail.Can anybody explains why this doesn’t seems to work in practice?
As far as I knew, there is no theorem proving that a minimum as X bits is needed to reproduce all the details in the signal. We have no Nyquist for the bit depth.From another post in this forum I learned that flipping even the LSB in 16 bit PCM is clearly audible. I’m inclined to think that a resolution well above the hearing threshold is a bit to ‘rough’Listening to recordings at a higher resolution (24 bits) should bring an improvement in detail.Can anybody explains why this doesn’t seems to work in practice?
There's nothing religious about the argument whatsoever, and any pro audio engineer worth their weight in ears will tell you how audibly valuable the 24-bit domain can be. Where people get lost in evaluating these "listening tests" is in understanding what their expectations should be given the test subjects' listening experience, hearing, the quality of the source material, the production methodologies employed, and the reproduction system and environment. People then end up thinking that one test's results regarding a *mastered* recording in some magazine is capable of proving something about all aspects of the 24-bit domain..... The fact that a fully mastered 16-bit recording from a 24bit source can be made to sound as good as its 24bit counterpart is a testament to the fact that there *is* an audible difference between 16-bit and 24-bit *raw source* material....My background and basis for being able to hear what I'm talking about comes purely from listening to raw and/or lightly mastered 24-bit recordings, on good monitors, with a good D/A, in a good room, with little to no EQ, and absolutely no compression. Compare that to the same source's 16-bit counterpart, and it's no contest.-DH
a) in practice, the average person (and sadly enough, many audio engineers) these days has far more hearing damage than they did 10 years ago, due to noise pollution, poorly mixed/mastered recordings, intentionally distorted musical content, excessive compression, improper EQ (can you say "Smiley face"), and high volume volume listening.
These are the things greater dynamic range bring to the table. The loss of detail due to poor representation afforded by shorter word lengths occurs in the smoothness of the dynamic changes (waveform amplitude changes) in lower level signals/components of the recording.
When you start with more significant bits, you have the ability to end up with more significant bits in the final product. Merely playing back a PCM stream that uses 24-bit words doesn't imply that the extra bits are significant by itself. Many listening tests involve a normalized 24 bit recording and a normalized 16 bit recording that has been dithered and noise shaped from the same 24 bit source. There is far less audible difference between those two sources, assuming a good dither/noise shaping algorithm is used. However, that test is not the test that validates the value of recording, mixing, and mastering in the 24 bit domain.
Aside from the fact that it is almost always easy to tell live from a recording, no matter how good the recording, comparison of live vs recording of any particular bit depth is not only unnecessary, it is completely irrelevant to the question of whether there is any audible difference between a 24 bit recording and the same properly converted to 16 bits.
Quote from: ccryder on 06 July, 2008, 06:19:38 PMThese are the things greater dynamic range bring to the table. The loss of detail due to poor representation afforded by shorter word lengths occurs in the smoothness of the dynamic changes (waveform amplitude changes) in lower level signals/components of the recording.No. It is well understood that, for properly dithered quantization the quantization error is not correlated with the original signal (see Oppenheim and Schafer, "Discrete Time Signal Processing" or for a good modern treatment, or W. R. Bennet, "Spectra of Quantized Signals", Bell Systems Technical Journal, vol. 27, 1948 for the foundations of the theory). Stating this mathematically, we define the error samples e[n]:e[n] = Q(x[n]) - x[n]where Q() is the quantization process and x[n] is the samples of the signal under test. When dither is used (or for self dithering signals, where the analog SNR is below is the quantization SNR) the signal e[n] is not correlated with the signals x[n] and Q(x[n]). So what does this mean in practice? It means that the properly dithered quantization process is an additive noise process - it is equivalent to adding independent noise with a particular spectrum to the original signal.Another thing is this belief that more quantization causes "roughness" in the output signal. As you are well aware, the reconstruction process does not produced a stepped output - it produces a bandlimited (and hence fairly smooth) analogue output. The output of a good sampling/reconstruction process is not "rough" or "stepped" in any way - if anything it will be less "rough" because of the strict bandlimiting imposed on the process.This doesn't mean that recording, mixing, mastering an processing at high bit depths are without merit. There are plenty of good technical reasons to argue for this, though, so using audiophile terms like "roughness" and resorting to incorrect interpretations of the process are not necessary.
That explanation would seem to imply to me that there is diminishing perceptibile difference in the slopes of different fade-in and fade-out curves for lower level signals. It also implies that the reconstruction of low amplitude waveforms can create "something" out of virtually nothing, and that "something," when normalized using pure bit shifting, is just as accurate as the reconstruction of high amplitude waveforms, with the same slope error. I beg to differ on at least the latter account. It is my experience that lower level harmonic and reverberant content "drops off the map" faster with 16 bit quantization, and that the timbre of such content changes over time as it decays in a different manner than a 24bit recording.
#2, take 2 recordings, both recorded conservatively so as not to risk ever clipping or hitting 0dBFS for more than 1 sample. Say, -24dBFS is the highest peak on a simultaneouly recorded 24bit and 16bit recording. I don't care how much dither and noise shaping you add, you're not going to end up with more than 12 significant bits on a 16 bit recording, but with good quality analog components & A/D, you will still end up with 16 significant bits of audio on the 24bit version.
Quote from: ccryder on 07 July, 2008, 03:13:21 AM#2, take 2 recordings, both recorded conservatively so as not to risk ever clipping or hitting 0dBFS for more than 1 sample. Say, -24dBFS is the highest peak on a simultaneouly recorded 24bit and 16bit recording. I don't care how much dither and noise shaping you add, you're not going to end up with more than 12 significant bits on a 16 bit recording, but with good quality analog components & A/D, you will still end up with 16 significant bits of audio on the 24bit version.I'm not quite clear whether you're talking about 24 bit recording, or a playback medium that's 24 bits compared to CD 16.Everyone here with an opinion seems to agree that for RECORDING, 24 bits is the way to go, but the consensus is that no-one has demonstrated the difference between 24 and 16 for PLAYBACK. In your example, @ccyrder, I assume, in my ignorance, that it would not be hard to produce a CD with the 16-bit depth the 24-bit recording produced. I really enjoy watching people hit each other over the head with science I scarcely understand, and mathematics which is, alas, incomprehensible to me. But it's more fun if everyone's in the same cage.
I never said a word about dither in my last post. Don't know why you insist on bringing that up.
But no matter what, low level signals dithered (with or without noise shaping) to 16 bit during recording and then normalizing through bit shifting will not in my opinion result in the same analog waveform with the same qualities of decay and timbre as those same low level signals recorded at the same level but quantized with 20+ bits (dithered or not), normalized non-destructively through bit shifting. This is especially true where the peak levels of the original recording are far away from 0dBFS.
Let's see, I could believe David Robinson, or I could believe you. No contest, David is right and you are wrong.
The rest gets the 48kHz treatment. Generally speaking, for mastered, fully produced music, I agree that 96kHz for playback is pretty much a waste of space and CPU power. Not having read Moran & Co.'s paper, but from the articles mentioned above, it would appear to me that the general focus of the test was to prove that higher sampling rates are a waste, and not greater wordlengths. Like I said, most of the time, and for the overwhelming majority of both listeners and listening material, I agree with their findings.