 |
 |
 |
 |
 |
 |
 |
Forums - Computer music & technology
Subject: 16 bit - 24 bit ????
Viewing all 4 messages - View by pages of 10: 1
Original Message 1/4 09-Nov-98 @ 11:40 PM - 16 bit - 24 bit ????
just on the subject, here's an few articles worth a read..... just draw your won conclusions..... obviously Bob Lentini is talking about sawplus , his product... but it raises some interesting reading
Lately, there have been some growing concerns about 24-bit Digital Audio, Dithering, and the SAW product line. It seems that magazine articles and forum discussions about these subjects have many people stirred up about the current validity of their 16 Bit Digital Audio Workstations.First let me say BEWARE! I urge you to evaluate all the facts, many of which are not brought forth in these magazine articles, before condemning your 16-bit Workstation to death. It can provide you with many more years of faithful service before 24-bit Digital Audio becomes the industry norm. A quote from one paragraph of a recent article written by Bob Katz and published in Mix Magazine concerning the use of 16-bit Digital Audio Workstations... "Do not change gain (changing gain deteriorates sound by forcing truncation of extra wordlengths in a 16-bit workstation). Do not normalize (normalization is just changing gain). Do not equalize. Do not fade-in or fade-out. Just edit. By the way, every edit in a 16-bit workstation involves a gain change during the crossfade (mix) from one segment to another, which creates long wordlengths during the calculation period (usually a brief couple of milliseconds). You probably won't notice the brief deterioration if you keep your edits short. Leave the segues and fade-outs for the mastering house, where they can properly handle the long wordlengths necessary for smooth fades (so that's why your last fade-out sounded like it dropped off a cliff!). Follow these simple guidelines and your digital audio will immediately start sounding better." …paints a pretty dim picture if you take these words as gospel. I see two problems here, one is the unfortunate fact that many people claim themselves to be experts in this field, who in my opinion should really spend a little more time checking out the real facts before opening their mouths, and two is the unfortunate fact that the industry magazines the people look up to for information and guidance are not necessarily motivated by integrity and the pursuit of truth in what they print, and will allow highly sensationalized articles, such as this, to appear in their publications. In my opinion, this article would find a better home in the National Inquirer. Many happy SAW users, completely content and feeling good about the quality of work they have been outputting for years on their SAW Workstations, all of a sudden have voiced concern that they should shutdown their business, after reading this article. The fact that the magazines should hold that much power over the people, is a scary realization. The fact that this type of misinformation proliferates in the printed medium, makes me sick. A look at these concerns from a more realistic and educated perspective can shed quite a different light on the matter. While many of the concerns talked about in the article have a basis in concept, in reality, they can be shown to have no meaning whatsoever. Oh, I'm not arguing the fact that 24-bit digital audio can sound better than 16-bit audio... its a plain fact more bits, more detail in the digitized audio. But lets look at how this applies to reality. At what point can you really hear the difference... consider this... place two photographs on the wall in a well lit room.... One has been slightly enhanced and retouched, and all agree that it looks better. Dim the lights and see who can tell which is which. At what point can you really tell the difference. Sure, put on your headphones, crank up the volume to max, and listen to that lone reverb trail at the end of your music... with 16-bit audio, there will be a moment right as the music fades to nothingness that you might be able to hear the breakup of the last bits toggling on and off. With 24-bits, this will happen well beneath the human threshold of hearing. How many of us spend time listening to music in this manner with the volume cranked to the max inside a set of headphones... not many... not for long... you'll probably be deaf by the end of one CD. By the way, how do all those precious CD's you own sound on your killer stereo.... not bad huh... amazing that they are only 16-bits. The point is, at normal or even extremely loud playback volumes, the digital breakup of the last bits in a 16-bit system are well below the threshold of hearing, especially when adding the room noise and ambience of a real, even quiet listening environment. Now, lets address some of the concerns of the quoted paragraph. First of all, its obvious that Mr. Katz was not talking about the SAW editing environment. Those of us SAW users who have witnessed others doing those crossfade edits on some of those high priced fancy editors are left with one burning question... WHY? Yes, why would you do that for every edit... why not just zoom in instantly, mark the edit point in detail at a zero cross energy point, and do a perfectly quiet butt-splice edit, without the need for a mini crossfade that involves all those awful calculations. Could it be that its extremely difficult, if not impossible to perform a simple feat like that on many of those high priced editors? How about the wordlength truncation issue during calculations... this is a big point stressed over and over throughout articles of this type. Users write me asking "What is the internal wordlength of SAWPlus32?" While SAWPlus32 is currently a 16-bit DAW, its internal summing wordlength is a full 32-bits. All summing buffers used to mix the 24 stereo virtual tracks are 32-bits long. This translates into the ability to sum over 65000 tracks of 16-bit audio recorded at a 100% clipping level without overrunning the internal wordlength capacity. Do you think that can handle your 24 stereo tracks without destroying your audio? The concerns voiced about the internal wordlength during volume and eq calculations might be true for many of the high priced systems Mr. Katz is familiar with, but SAW addresses these issues a little differently. SAW uses high speed hand-coded 32-bit assembly language integer math, which bypasses many of the high level language limitations in controlling wordlength spillover during intermediate calculations. SAW takes advantage of the full 64-bit double register wordlength available in the 486 and Pentium processors in each of its calculations for processes like volume and eq and so forth. This results in a higher internal resolution than found in most DSP processors which are limited to 40-bits. As a result, SAW loses no data whatsoever during the internal manipulation of its audio. Adjust the volume to your heart's content... eq as much as you want... WE LOSE NO DATA! "But what about the end results?", a concerned user writes... "How does SAW squeeze all those tracks and calculations that overrun the 16-bit wordlength back down into the 16-bit mixdown?" A good question... how bout a good answer. The article goes on to explain the choices for doing so... truncating the data bits that don't fit.... or Dithering the data down.. Well, truncating the data is certainly a pretty bad choice, especially after doing so much work in the internal design to preserve all those precious bits... Not an option for SAW. Dithering... the incredible rocket science term that allows many experts to stand tall above you because they can wow you with all kinds of geek talk about how they can preserve your precious audio bits as you translate back from 24-bit audio to 16-bit audio in order to get your final project to a DAT or CD. Just what is dithering... well as I see it, after all is said and done and the smoke clears... its a way to add noise to your entire project just to mask the digital breakup of the last bits that you might be able to hear during the last reverb trail of your song if you listen to your music with headphones on at max volume. Beats me what all the fuss is about... if you want to add noise that sounds like tape hiss to your project, simply record an extra track of cassette hiss onto the SAW MultiTrack and mix the volume way down to -79 db. This will surely keep you from hearing the digital breakup of the last bits. Why all the geek stuff and the fancy algorithms and the long processing times required to dither... do it live in SAW. Bottom line, dithering was not a desirable option for SAW either. So how does SAW solve this mystery of mysteries. How about back to the basics... the simplest solution, I have found, is always the best solution. SAW simply lets you scale the data back down until it fits within the allotted 16-bits by using the output track faders. Adjust the fader in offset mode until any clipping distortion goes away... no audible data is thrown away, the volume is simply scaled back down till it fits... trust that your CD will be mastered at the highest level it can take. The thing that bothers me the most about the dithering discussions is the way the process is misrepresented into making the readers think they are getting back otherwise lost data... THIS IS AN OUT AND OUT LIE! You gain nothing by taking 16-bit data up to 24-bit data, processing and then dithering back down, except the noise that the dither process adds to mask the low level digital breakup. Bits that were not captured from the original digitizing process are not magically re-constructed during the 24-bit expansion process and bits that would otherwise be lost during the down conversion are still lost, blended into a bed of digital noise. The dithering process may certainly provide some noticeable benefit when down converting from 16-bits to 8-bits, because the last single bit level is quite audible under normal playback conditions. It is extremely questionable whether the same is true at the 16-bit last single bit level. And what about the argument for sampling at 24-bits and maintaining the resolution at 24-bits the whole time during editing and then dithering down at the very end of the project? Are the results better... or worse... or even detectable? There are many professionals who are responsible for mastering most of the Music Industry CD's you buy and keep in your collection, who would argue... if its going to end up in 16-bits, keep it 16-bits from the gitgo. In the next few years, we might start to see the industry change over to 24-bit DATs and CD players, but for now, buying into all this hype seems like a waste of time to me. Trust that when 24-bit is an industry reality all the way to the finished product, IQS and SAW will be there in full strength with a full 24 or 32-bit data path. By the way, all those who decide that you can no longer stomach working on a 16-bit platform like SAW to do your digital editing, please send all your un-listenable and useless 16-bit CD collections to me at IQS... I'll be happy to find new homes for them with those of us less discerning audiophiles. Bob Lentini
CEO/Owner Innovative Quality Software
9/15/97
Lately, there have been some growing concerns about 24-bit Digital Audio, Dithering, and the SAW product line. It seems that magazine articles and forum discussions about these subjects have many people stirred up about the current validity of their 16 Bit Digital Audio Workstations.First let me say BEWARE! I urge you to evaluate all the facts, many of which are not brought forth in these magazine articles, before condemning your 16-bit Workstation to death. It can provide you with many more years of faithful service before 24-bit Digital Audio becomes the industry norm. A quote from one paragraph of a recent article written by Bob Katz and published in Mix Magazine concerning the use of 16-bit Digital Audio Workstations... "Do not change gain (changing gain deteriorates sound by forcing truncation of extra wordlengths in a 16-bit workstation). Do not normalize (normalization is just changing gain). Do not equalize. Do not fade-in or fade-out. Just edit. By the way, every edit in a 16-bit workstation involves a gain change during the crossfade (mix) from one segment to another, which creates long wordlengths during the calculation period (usually a brief couple of milliseconds). You probably won't notice the brief deterioration if you keep your edits short. Leave the segues and fade-outs for the mastering house, where they can properly handle the long wordlengths necessary for smooth fades (so that's why your last fade-out sounded like it dropped off a cliff!). Follow these simple guidelines and your digital audio will immediately start sounding better." …paints a pretty dim picture if you take these words as gospel. I see two problems here, one is the unfortunate fact that many people claim themselves to be experts in this field, who in my opinion should really spend a little more time checking out the real facts before opening their mouths, and two is the unfortunate fact that the industry magazines the people look up to for information and guidance are not necessarily motivated by integrity and the pursuit of truth in what they print, and will allow highly sensationalized articles, such as this, to appear in their publications. In my opinion, this article would find a better home in the National Inquirer. Many happy SAW users, completely content and feeling good about the quality of work they have been outputting for years on their SAW Workstations, all of a sudden have voiced concern that they should shutdown their business, after reading this article. The fact that the magazines should hold that much power over the people, is a scary realization. The fact that this type of misinformation proliferates in the printed medium, makes me sick. A look at these concerns from a more realistic and educated perspective can shed quite a different light on the matter. While many of the concerns talked about in the article have a basis in concept, in reality, they can be shown to have no meaning whatsoever. Oh, I'm not arguing the fact that 24-bit digital audio can sound better than 16-bit audio... its a plain fact more bits, more detail in the digitized audio. But lets look at how this applies to reality. At what point can you really hear the difference... consider this... place two photographs on the wall in a well lit room.... One has been slightly enhanced and retouched, and all agree that it looks better. Dim the lights and see who can tell which is which. At what point can you really tell the difference. Sure, put on your headphones, crank up the volume to max, and listen to that lone reverb trail at the end of your music... with 16-bit audio, there will be a moment right as the music fades to nothingness that you might be able to hear the breakup of the last bits toggling on and off. With 24-bits, this will happen well beneath the human threshold of hearing. How many of us spend time listening to music in this manner with the volume cranked to the max inside a set of headphones... not many... not for long... you'll probably be deaf by the end of one CD. By the way, how do all those precious CD's you own sound on your killer stereo.... not bad huh... amazing that they are only 16-bits. The point is, at normal or even extremely loud playback volumes, the digital breakup of the last bits in a 16-bit system are well below the threshold of hearing, especially when adding the room noise and ambience of a real, even quiet listening environment. Now, lets address some of the concerns of the quoted paragraph. First of all, its obvious that Mr. Katz was not talking about the SAW editing environment. Those of us SAW users who have witnessed others doing those crossfade edits on some of those high priced fancy editors are left with one burning question... WHY? Yes, why would you do that for every edit... why not just zoom in instantly, mark the edit point in detail at a zero cross energy point, and do a perfectly quiet butt-splice edit, without the need for a mini crossfade that involves all those awful calculations. Could it be that its extremely difficult, if not impossible to perform a simple feat like that on many of those high priced editors? How about the wordlength truncation issue during calculations... this is a big point stressed over and over throughout articles of this type. Users write me asking "What is the internal wordlength of SAWPlus32?" While SAWPlus32 is currently a 16-bit DAW, its internal summing wordlength is a full 32-bits. All summing buffers used to mix the 24 stereo virtual tracks are 32-bits long. This translates into the ability to sum over 65000 tracks of 16-bit audio recorded at a 100% clipping level without overrunning the internal wordlength capacity. Do you think that can handle your 24 stereo tracks without destroying your audio? The concerns voiced about the internal wordlength during volume and eq calculations might be true for many of the high priced systems Mr. Katz is familiar with, but SAW addresses these issues a little differently. SAW uses high speed hand-coded 32-bit assembly language integer math, which bypasses many of the high level language limitations in controlling wordlength spillover during intermediate calculations. SAW takes advantage of the full 64-bit double register wordlength available in the 486 and Pentium processors in each of its calculations for processes like volume and eq and so forth. This results in a higher internal resolution than found in most DSP processors which are limited to 40-bits. As a result, SAW loses no data whatsoever during the internal manipulation of its audio. Adjust the volume to your heart's content... eq as much as you want... WE LOSE NO DATA! "But what about the end results?", a concerned user writes... "How does SAW squeeze all those tracks and calculations that overrun the 16-bit wordlength back down into the 16-bit mixdown?" A good question... how bout a good answer. The article goes on to explain the choices for doing so... truncating the data bits that don't fit.... or Dithering the data down.. Well, truncating the data is certainly a pretty bad choice, especially after doing so much work in the internal design to preserve all those precious bits... Not an option for SAW. Dithering... the incredible rocket science term that allows many experts to stand tall above you because they can wow you with all kinds of geek talk about how they can preserve your precious audio bits as you translate back from 24-bit audio to 16-bit audio in order to get your final project to a DAT or CD. Just what is dithering... well as I see it, after all is said and done and the smoke clears... its a way to add noise to your entire project just to mask the digital breakup of the last bits that you might be able to hear during the last reverb trail of your song if you listen to your music with headphones on at max volume. Beats me what all the fuss is about... if you want to add noise that sounds like tape hiss to your project, simply record an extra track of cassette hiss onto the SAW MultiTrack and mix the volume way down to -79 db. This will surely keep you from hearing the digital breakup of the last bits. Why all the geek stuff and the fancy algorithms and the long processing times required to dither... do it live in SAW. Bottom line, dithering was not a desirable option for SAW either. So how does SAW solve this mystery of mysteries. How about back to the basics... the simplest solution, I have found, is always the best solution. SAW simply lets you scale the data back down until it fits within the allotted 16-bits by using the output track faders. Adjust the fader in offset mode until any clipping distortion goes away... no audible data is thrown away, the volume is simply scaled back down till it fits... trust that your CD will be mastered at the highest level it can take. The thing that bothers me the most about the dithering discussions is the way the process is misrepresented into making the readers think they are getting back otherwise lost data... THIS IS AN OUT AND OUT LIE! You gain nothing by taking 16-bit data up to 24-bit data, processing and then dithering back down, except the noise that the dither process adds to mask the low level digital breakup. Bits that were not captured from the original digitizing process are not magically re-constructed during the 24-bit expansion process and bits that would otherwise be lost during the down conversion are still lost, blended into a bed of digital noise. The dithering process may certainly provide some noticeable benefit when down converting from 16-bits to 8-bits, because the last single bit level is quite audible under normal playback conditions. It is extremely questionable whether the same is true at the 16-bit last single bit level. And what about the argument for sampling at 24-bits and maintaining the resolution at 24-bits the whole time during editing and then dithering down at the very end of the project? Are the results better... or worse... or even detectable? There are many professionals who are responsible for mastering most of the Music Industry CD's you buy and keep in your collection, who would argue... if its going to end up in 16-bits, keep it 16-bits from the gitgo. In the next few years, we might start to see the industry change over to 24-bit DATs and CD players, but for now, buying into all this hype seems like a waste of time to me. Trust that when 24-bit is an industry reality all the way to the finished product, IQS and SAW will be there in full strength with a full 24 or 32-bit data path. By the way, all those who decide that you can no longer stomach working on a 16-bit platform like SAW to do your digital editing, please send all your un-listenable and useless 16-bit CD collections to me at IQS... I'll be happy to find new homes for them with those of us less discerning audiophiles. Bob Lentini
CEO/Owner Innovative Quality Software
9/15/97
Message 2/4 09-Nov-98 @ 11:42 PM - RE: 16 bit - 24 bit ????
here's some more...
First, let me say... this has been a fun one. Talk about controversy... my, my, did we strike a chord close to home on this topic. While I anticipated that taking a stand against one of the industries most preciously defended issues would result in some engaging rebuttals, the responses that arrived were, to say the least, interesting indeed! Some of the respected professionals and experts of the audio industry have rapidly responded to my original soapbox article, in an attempt to defend their position. What is unfortunate, is that the general tone of their messages, as you will see here, seems to have taken on a 'holier than thou attitude', one of a personal campaign to slander me and expose to the world my supposed lack of audio knowledge, and my devious attempts to lie and cover up the truth about my product’s capabilities. Not at all a very professional attitude, in my opinion. But yet, not a surprising one either. I'd like to point out that the original article was simply an attempt to shed a slightly different perspective on the issues involved in these controversial topics. Please note, that my original article never attacked Bob Katz's audio knowledge or abilities as an audio engineer, only stated that, in my opinion, in the quoted article, his case was overly one-sided, and potentially presented a message of doom for all audio enthusiasts, aspiring professionals, and practicing professionals who do not have access to a high end 24 bit workstation to do their audio editing. I do believe, if you re-read my article, that nowhere did I ever question the fact that 24 bit audio sounded better and nowhere did I ever say that I thought 16 bit was the same. I simply presented the notion that, 16 bit audio properly handled in a product like SAW, could result in excellent quality output that most people under normal listening conditions would have a hard time distinguishing from 24 bit audio which had been reduced to 16 bit output, dithered or not. In these responses, you will see me accused of all kinds of things... acting like a caged animal... ranting and raving... misleading the public with my ridiculous unsubstantiated ramblings, etc. etc. You will also see that, for reasons unknown, the entire focal point of my original article has been twisted and distorted beyond recognition and apparently missed altogether. I really don't think I was that vague about the issue at hand. It is also interesting to note the perceived fascination these people have with being right, and how much effort is put into finding fault after fault with my way of thinking and coding, again missing the original issue altogether. Read these writings of the professional audio community and then you come to your own conclusions; am I the ignorant criminal they make me out to be, or simply a professional in my own right, willing to question the ivory towers and find my own truth. All of the quoted responses, and more, are presented in full for your reading enjoyment, as I want there to be no misunderstandings about taking things out of context. Some of the more interesting segments, I will respond to here. Hopefully, after this, we can all step back, chuckle about the opinions and rebuttals presented here, and get on with life. Bob Katz replies... "I love controversy, especially when I know I'm right :-) I received the following letter from a Pro Audio subscriber. A company has been
mentioned by name, they have responded to one of my articles by name and in public,
so I see no problem in continuing this discussion by name and also in public. I regret
having to bring this response to the public, but since the question was made publicly,
I deserve to respond in like manner. I am cc'ing my response to three people: to IQS,
to my clients who came to me with their "SAW'ed" DAT, and to the Pro Audio list.
I suggest Mr. Lentini from IQS join the community of the Pro Audio list and I am
forwarding (by separate email) instructions to him on how to do so. The Pro Audio
Mailing list is a community of mastering engineers, recording engineers, dsp
designers and audiophiles who are concerned about good sound. Since the AES
Convention is coming up, I imagine this message will atrophy until around
October 3rd or 4th, when lots of responses will come up. I also urge anyone on
the pro audio list who wishes to respond to this message, in courtesy to the
parties involved, to please cc your message to iqsmag@iqsoft.com. That will
guarantee the gentleman (Bob Lentini) will receive a complete summary of the
opinions of those on the Pro Audio list about this important issue. And I hope
there will be some opinions! >Bob, >You don't know me personally, but I am on the Pro Audio Email list, and
>have read your Emails for quite some time. I have also read your articles
>both in Mix and on your web site, and have always had the greatest respect
>for your insights and your willingness to share your knowledge with others.
>Recently I was at the Innovative Quality Software web site
>(www.iqsoft.com), the company that creates and distributes the SAW+ and
>SAW+32 digital audio products for the PC, and saw the following article in
>a section called the IQS Magazine, under the heading "Bob's Soapbox". The
>author is Bob Lentini, who is also the primarly author of SAW+. He quotes
>one of your articles and responds to it. I thought it only fair that you
>should at least know about this, so that if you want you can respond to his
>comments. >Following is the text of his article. >Dean Richard ... Original soapbox article... ... There is more, and I will quote verbatim below, but let me begin to respond
to the top half of his "soapbox". First of all, as long as you asked, I was quite specifically talking about
SAW. I've heard the problems that SAW users have created in their
workstation, and they are not subtle. The gentleman's defense has little
relation to reality. I have received material on DAT that went through SAW
once or twice or even three times and which sounds exactly like it was cut
with a buzzsaw. I've had clients totally dismayed at the losses in their
material and the increased harshness and edginess in their material. It was
quite simple to A/B compare their original source to the DAT that was
processed through the SAW. The problems were intuitively obvious to the
most casual observer without turning up the gain, without examining tails,
simply by listening to the music. In fact, the music was hard rock and
levels rarely went below - 10 dBFS! If the room noise of a normal listening
environment were sufficient to mask these problems, if problems were simply
reverb tails and decays and other "subtleties", then I'm sure you could sit
right back and rest assured. Unfortunately, it is not true. The human
"threshholds" of hearing are amazing. Low level problems can be heard right
through the hottest material as.... FIRST: a loss of stereo separation
SECOND: a loss of clarity and solidity to the mix
THIRD: an increase in harshness and distortion, at all recorded levels and other obvious symptoms..." What I find interesting here is the immediate conclusion drawn about the tool (SAW) used to create a badly destroyed project tape. I do not question Mr. Katz's ability to judge whether the tape was destroyed or not. I do question, however the conclusion that was drawn with no further mention or apparent attention being paid to any of the other elements that may have played an important part in the destruction of that tape. What soundcard was used? (SAW does work with a SoundBlaster). How was SAW used... properly or improperly? Etc. Once again, he clearly states that the damage is directly due to the 16 bit editor. Worse, now he has gotten personal and directly blames the SAW program for the damage. Armed with this professional advise and newfound knowledge of deductive reasoning, I now feel my decision making abilities have been greatly enhanced: All Nikon cameras are bad because I once saw an out of focus and underexposed picture taken with one. or My neighbor owns a Sears Craftsman table saw. He builds lousy cabinets... splintered and crooked... therefore I should buy a Black and Decker. I am not saying he is wrong about 24 bit or dither... simply complaining about his approach to proving a point... very one-sided with little apparent regard for other possibilities and elements that can play an important part in the overall outcome. There are many elements that make or break an audio project... mic placement, proper use of eq, proper attention to phasing principles, good sense of space and instrument placement in the mix, good talent, good performance... etc... etc... etc... not just the difference between 16 and 24 bit digital or whether the project was dithered or not. Let's compare the final tape with the original, all other factors being equal... clearly the tape was damaged by SAW. Trust, that I'm sure I could set foot into any mastering facility, misuse the equipment and produce some very bad results, using this to misguide my clients into blaming the facility. Is this really conclusive evidence, or just plain BS?
This is an interesting series... Mr. Katz continues... " …A lot of very good scientists and psychoacousticians have spent years
developing 72 bit-accurate dithering algorithms to capture the highest
audio resolution, and they would be quite shocked to learn that their
efforts are unnecessary." I'll bet that many noted scientists of the day were also shocked to discover that the world was round. The boundaries of science have always been challenged and pushed to new limits since the beginning of science. "Mr. Lentini is restating a very old and specious
argument that the noise which is added by dither is more problematic than
the distortion which it fixes." Still an ongoing subjective debate that by no means determines the extent or legitimacy of an individual's knowledge or a product's worth. This argument is simply one of opinion and will rage on for a very long time. How can you be right about something like... Coke is better than Pepsi? " …In fact, I go over the issue of compromise quite well in my dither article,
with my (now famous?) quote: "Dither, you can't live with it, and you can't
live without it", which was designed to point out that there is an audible
veil added by dithering (even the best dithering), and that the best way to
avoid any compromise and the best way to maintain the highest quality audio
is to retain long wordlength results (high resolution), until the final
16-bit product is produced. The best way is to add the best quality dither
you have *once* and only *once* at the end of the project. Clearly, if you calculate to 24 or more bits, then add dither and store at
16 bits, then you recalculate and reprocess to a long wordlength 24 bits,
and then add dither again and store at 16 bits---the cumulation will add
quite a veil over your sound. Even one generation of dither is quite
audible, but to my ears (and those of a lot of audiophiles), that
generation of dither is far more pleasant to the ear than simple truncation
as is done by Mr. Lentini in his product (if I read his statement
correctly). There is no such thing as a free lunch. I think I understand
the need to use dither and the compromises involved in its use quite well,
and I wonder if Mr. Lentini has really thought as deeply about this issue." Mr. Mithat Konar, director of engineering of Biro Technology adds... "Judging from his writing, I believe Mr. Lentini has NOT thought very deeply
about the issue of dither as he seems to fundamentally misunderstand its
purpose and capabilities. Rather than take up the required bandwidth to set
him straight, I'll simply refer him to any of the JAES articles on the
subject--those by Lipshitz and/or Vanderkooy in particular being great
places to start. However, I WILL take up the bandwidth to restate the
following often-stated and well-proven conclusions: (1) Dither is NOT simply a means "to mask the low level digital breakup".
(2) Dither extends the resolution of a digital system below the LSB threshold.
(3) Dither decorrelates truncation errors from low-level input signals. In
layman's terms this means dither eliminates a particularly nasty form of
distortion in low-level signals. (Implicit in the above is the assumption that the signal being subject to
the dithering operation has resolution below the LSB threshold. Virtually
any gain change, filtering, or mixing operation will result in "new"
information below the LSB of the input.) I submit it is the most profound professional irresponsibility for anyone
to make claims that contradict the above until s/he has better researched
the topic and can DISPROVE these PROVEN findings. Or, to use the same
language that Mr. Lentini has used, to assert anything contrary to the
above statements "IS AN OUT AND OUT LIE!"" Gentleman #1 adds... "You really should put a muzzle on Lentini. As much as he may insist the
opposite is true, he is DEAD wrong on a number of issues and points out that
he has some really grotesque misconceptions about how things are really done
in the real mastering world. I won't even enumerate these mistakes right
now since I'm leaving in 10 minutes for New York for the AES show (I'll be
doing a workshop on 96 kHz mastering with Daniel Weiss, Bob Stuart, Mike
Storey, Andy Moorer, and Ian Dennis. Mr. Lentini might really benefit from
being there.) I really don't think you want Lentini representing your
company, for a number of reasons. I assure you he looks quite foolish to
those of us who really do know what we're doing. Furthermore, he's
disseminating blatant nonsense to a confused public who trust him. When I
get back, I would be glad to enumerate each mistake he made in a posting to
the Pro Audio Mailing List at pgm.com, point by point, in a civil manner,
and let him know where he went wrong. I'm sure he's probably getting
bombarded by mail like this now, but there's no conspiracy here. He just
screwed up BIG time, plain and simple, and it'll soon be time for a bit of
explanation and/or back-pedaling." And in another message adds... "To whom it may concern, I URGE YOU TO READ ALL OF THE FOLLOWING I feel the need to express my profound concerns over the rantings of
your CEO and owner, Mr. Richard Lentini. (Did I miss something or was this Dick's Soapbox?) I will be brief here, but I would be glad to expound in detail any of the points
I touch upon. It's easy to see Lentini's diatribe for what it is: he's obviously (and
understandably) frustrated by a fickle customer base that has found
itself, all of a sudden, concerned that IQS's product isn't good
enough. Join the club. As manufacturers, we all have to deal with this
problem from time to time. What I object to is the GROTESQUE
misinformation he spews forth. Unsettling as it may be, I hope he
didn't believe what he was saying and that this was just a misguided
application of marketing hyperbole. At the very least, Lentini displays a thorough and frightening
misunderstanding of the most basic concepts of dynamic range, noise
shaping, and dither. Additionally, he doesn't seem to have a clue about
how compact discs are mastered or how other products work. Furthermore,
his ridiculous comments about dither and "rocket science" clearly
indicate he has never listened to music in a real mastering environment
or on a properly calibrated monitoring system.... What's worst of all, Lentini is misleading droves of uneducated
customers who count on him to provide guidance and the best product
possible. Rather than lashing out like a cornered animal, he'd have
been better off quietly spending a little time and money on R&D,
admitting to himself that he doesn't know everything about DSP and audio
signal processing, and making the product better and more up-to-date.
If he's succeeded in convincing the legions of SAW users that he's right
and the real experts are wrong, he's created a monster: a diminution of
the state of the art and the mistaken belief among a wide community of
users that their recordings sound as good as they can be. Not to be overly didactic, but Lentini is so wrong on so many counts
here that it's hard to keep track. He can scream and rant all he wants,
but there's a community of experts who will be glad to set the record
straight. I'm talking about people like Dr. Stanley Lipschitz and Dr.
John Vanderkooy (Univ. of Waterloo); Julian Dunn, Ian Dennis, and Graham
Boswell (PrismSound); Jim Johnston (AT&T); Dr. James Moorer (Sonic
Solutions); Mike Storey (dCS); Dan Lavry and Bruce Hemingway (dB
Technologies); Bob Adams (Analog Devices); Daniel Weiss (Weiss
Engineering); Bob Stuart and Rhonda Wilson (Meridian); John LaGrou and
Fred Forssell (Millennia Media); and me, to name but a few. I've never been moved to write a letter to another manufacturer, but
Lentini's rant was just so out-of-line I felt I had to say something.
He should admit he's wrong, apologize, join the world of civilized
discourse, and learn a thing or two along the way. Everyone would
benefit." Whew!!! What seems most noteworthy about these excerpts is the fact that these letters were received from professionals proudly stating their names and titles, and the fact that many of them were on their way to important seminars and meetings at AES. In fact it was even recommended that I should attend so I might learn some of the real truth about audio. Did you notice the professionalism and debating skills demonstrated by such profound responses like "You really should put a muzzle on Lentini". Why the shift to personal assault? This whole thing simply began as a conversation about the usefulness of 16 bit workstations. Varying opinions are one thing, but the stoop to direct personal assault and the unqualified judgements about one's knowledge on a subject, hardly qualify as a professional attitude and do little to win anyone's respect. First of all, I must say I am truly amazed by the clairvoyance and the psychic abilities of these professionals, to be so self-assured about what I do or don't know about digital audio, based on an opinion I published about one small facet of it. In fact, its clear they must know all about my 28 year audio career, down to every detail of my circuitry design, construction and application in both the pristine studio environment and also the live theatre environment. But this is all irrelevant... I challenged one of their beliefs... that challenge nulls and voids any other possible knowledge or skills I might have... and now I must be punished! "And I wonder if Mr. Lentini has really thought as deeply about this issue...". And I wonder how much code these professionals have written, and how many dither and DSP algorithms they designed and experimented with before forming their opinion. And I wonder how many bit patterns they have studied in hex and audio editors looking at the effects of dither on the LSB of a 16 bit signal, before making their decision about this issue. Don't give me a list of articles or books to read. I'm sure I can double or triple your list with the books sitting right here on my shelf. Perhaps what they are missing here, is that maybe I have given this subject much thought, and maybe I even experimented with dozens of dither algorithms, and maybe I even wrote code, and then more code, and applied it to my 16 bit audio data, and maybe I read every article I could find on the subject and maybe, just maybe, I wasn't afraid to think for myself and just happened to conclude that maybe, in the end, a lot of this hoopla is just that... hoopla! Not to worry... just the ramblings of an insignificant idiot who obviously knows nothing about digital audio... it's okay though... I feel better now. Yes I know all about the effects of pulse width modulation and the psycho-acoustic affects attributed to dither and the results shown in the FFT's with dithered data.... I can read too, but thank you all for trying to enlighten me. I never said anything to discount all that research... simply that maybe 16 bit audio, properly processed and handled might be okay without it. Amazing that a simple statement like that would condemn me to death!
Here's one that I really like: "To Whom It May Concern: I was just getting ready to evaluate the latest PC digital audio systems
for myself and the companies I consult for. Please thank Bob Lentini for
saving me the trouble of evaluating the SAW system. A company that has a
CEO that can ramble on with such obvious lack of actual clue could not
possibly design a product that I would be interested in. See you at AES? Sincerely, Tom Garneau
Audioactive Productions" My concern here is for the companies he consults for. It's obvious he is qualified to make the best choices for their needs. Now that I know his powerful secret for picking the best products, maybe I too can become a consultant. Let's see... This company's logo is pink... I hate pink... therefore this company's products are no good. or The owner of this company eats meat... therefore this company's products are not healthy to be around. or This product is the most expensive in its category... therefore it must be the best.
This next one represents some of the side-effects of becoming too righteous about how much you know. Herbie Robinson of HR Music Systems writes... "Let me correct some more technical inaccuracies here (In addition to
what Bob said). I'm not going to comment on the obvious...
If Bob L. really believed that BS diatribe about 16 bits being enough,
then why does he devote the next half of his artical explaining why all of
the processing in SAWxx is done in more than 32 bits?" Uh oh! ... has he got me? Keep digging for dirt... come on... you can nail me on something! Because in a virtual mixing environment, which SAW is, a real and most noticeable concern, in my opinion, is the truncation of bits at the MSB location. Without the 32 bit summing buffers, a few tracks of normally recorded audio would overrun the 16 bit storage capacity and truncate data at the Most Significant Bit, which is unquestionably audible to all that listen, not only the pristine golden ears of the audiophile. In answer to one of my paragraphs... "> >This translates into the ability to sum over 65000 tracks of 16-bit
> >audio recorded at a 100% clipping level without overrunning
> >the internal wordlength capacity. Do you think that can handle your 24
> >stereo tracks without destroying your audio? This is only true if no FADERS are allowed for. In other words, this is
pure BS from a practical standpoint." Maybe I'm missing something here, but I fail to see what FADERS have to do with it. Summing this way allows for 65535 tracks at max 16 bit level to be summed without overload clipping distortion being introduced. So... what's the pure BS from a practical standpoint mean? This is extremely important in practical applications as it allows you to record the number of tracks you need for your project at a good hot level and be confident that you can mix them together without internally clipping the data. Without this capacity, you would have to record each track at a significantly lower level, truly then running into lesser audio quality due to not maximizing the use of the 16 bits you do have to work with. "> >...hand-coded 32-bit assembly language integer math, which bypasses many
> >of the high level language limitations in controlling
> >wordlength spillover during intermediate calculations. SAW takes
> >advantage of the full 64-bit double register wordlength
> >available in the 486 and Pentium processors in each of its calculations
> >for processes like volume and eq and
First, let me say... this has been a fun one. Talk about controversy... my, my, did we strike a chord close to home on this topic. While I anticipated that taking a stand against one of the industries most preciously defended issues would result in some engaging rebuttals, the responses that arrived were, to say the least, interesting indeed! Some of the respected professionals and experts of the audio industry have rapidly responded to my original soapbox article, in an attempt to defend their position. What is unfortunate, is that the general tone of their messages, as you will see here, seems to have taken on a 'holier than thou attitude', one of a personal campaign to slander me and expose to the world my supposed lack of audio knowledge, and my devious attempts to lie and cover up the truth about my product’s capabilities. Not at all a very professional attitude, in my opinion. But yet, not a surprising one either. I'd like to point out that the original article was simply an attempt to shed a slightly different perspective on the issues involved in these controversial topics. Please note, that my original article never attacked Bob Katz's audio knowledge or abilities as an audio engineer, only stated that, in my opinion, in the quoted article, his case was overly one-sided, and potentially presented a message of doom for all audio enthusiasts, aspiring professionals, and practicing professionals who do not have access to a high end 24 bit workstation to do their audio editing. I do believe, if you re-read my article, that nowhere did I ever question the fact that 24 bit audio sounded better and nowhere did I ever say that I thought 16 bit was the same. I simply presented the notion that, 16 bit audio properly handled in a product like SAW, could result in excellent quality output that most people under normal listening conditions would have a hard time distinguishing from 24 bit audio which had been reduced to 16 bit output, dithered or not. In these responses, you will see me accused of all kinds of things... acting like a caged animal... ranting and raving... misleading the public with my ridiculous unsubstantiated ramblings, etc. etc. You will also see that, for reasons unknown, the entire focal point of my original article has been twisted and distorted beyond recognition and apparently missed altogether. I really don't think I was that vague about the issue at hand. It is also interesting to note the perceived fascination these people have with being right, and how much effort is put into finding fault after fault with my way of thinking and coding, again missing the original issue altogether. Read these writings of the professional audio community and then you come to your own conclusions; am I the ignorant criminal they make me out to be, or simply a professional in my own right, willing to question the ivory towers and find my own truth. All of the quoted responses, and more, are presented in full for your reading enjoyment, as I want there to be no misunderstandings about taking things out of context. Some of the more interesting segments, I will respond to here. Hopefully, after this, we can all step back, chuckle about the opinions and rebuttals presented here, and get on with life. Bob Katz replies... "I love controversy, especially when I know I'm right :-) I received the following letter from a Pro Audio subscriber. A company has been
mentioned by name, they have responded to one of my articles by name and in public,
so I see no problem in continuing this discussion by name and also in public. I regret
having to bring this response to the public, but since the question was made publicly,
I deserve to respond in like manner. I am cc'ing my response to three people: to IQS,
to my clients who came to me with their "SAW'ed" DAT, and to the Pro Audio list.
I suggest Mr. Lentini from IQS join the community of the Pro Audio list and I am
forwarding (by separate email) instructions to him on how to do so. The Pro Audio
Mailing list is a community of mastering engineers, recording engineers, dsp
designers and audiophiles who are concerned about good sound. Since the AES
Convention is coming up, I imagine this message will atrophy until around
October 3rd or 4th, when lots of responses will come up. I also urge anyone on
the pro audio list who wishes to respond to this message, in courtesy to the
parties involved, to please cc your message to iqsmag@iqsoft.com. That will
guarantee the gentleman (Bob Lentini) will receive a complete summary of the
opinions of those on the Pro Audio list about this important issue. And I hope
there will be some opinions! >Bob, >You don't know me personally, but I am on the Pro Audio Email list, and
>have read your Emails for quite some time. I have also read your articles
>both in Mix and on your web site, and have always had the greatest respect
>for your insights and your willingness to share your knowledge with others.
>Recently I was at the Innovative Quality Software web site
>(www.iqsoft.com), the company that creates and distributes the SAW+ and
>SAW+32 digital audio products for the PC, and saw the following article in
>a section called the IQS Magazine, under the heading "Bob's Soapbox". The
>author is Bob Lentini, who is also the primarly author of SAW+. He quotes
>one of your articles and responds to it. I thought it only fair that you
>should at least know about this, so that if you want you can respond to his
>comments. >Following is the text of his article. >Dean Richard ... Original soapbox article... ... There is more, and I will quote verbatim below, but let me begin to respond
to the top half of his "soapbox". First of all, as long as you asked, I was quite specifically talking about
SAW. I've heard the problems that SAW users have created in their
workstation, and they are not subtle. The gentleman's defense has little
relation to reality. I have received material on DAT that went through SAW
once or twice or even three times and which sounds exactly like it was cut
with a buzzsaw. I've had clients totally dismayed at the losses in their
material and the increased harshness and edginess in their material. It was
quite simple to A/B compare their original source to the DAT that was
processed through the SAW. The problems were intuitively obvious to the
most casual observer without turning up the gain, without examining tails,
simply by listening to the music. In fact, the music was hard rock and
levels rarely went below - 10 dBFS! If the room noise of a normal listening
environment were sufficient to mask these problems, if problems were simply
reverb tails and decays and other "subtleties", then I'm sure you could sit
right back and rest assured. Unfortunately, it is not true. The human
"threshholds" of hearing are amazing. Low level problems can be heard right
through the hottest material as.... FIRST: a loss of stereo separation
SECOND: a loss of clarity and solidity to the mix
THIRD: an increase in harshness and distortion, at all recorded levels and other obvious symptoms..." What I find interesting here is the immediate conclusion drawn about the tool (SAW) used to create a badly destroyed project tape. I do not question Mr. Katz's ability to judge whether the tape was destroyed or not. I do question, however the conclusion that was drawn with no further mention or apparent attention being paid to any of the other elements that may have played an important part in the destruction of that tape. What soundcard was used? (SAW does work with a SoundBlaster). How was SAW used... properly or improperly? Etc. Once again, he clearly states that the damage is directly due to the 16 bit editor. Worse, now he has gotten personal and directly blames the SAW program for the damage. Armed with this professional advise and newfound knowledge of deductive reasoning, I now feel my decision making abilities have been greatly enhanced: All Nikon cameras are bad because I once saw an out of focus and underexposed picture taken with one. or My neighbor owns a Sears Craftsman table saw. He builds lousy cabinets... splintered and crooked... therefore I should buy a Black and Decker. I am not saying he is wrong about 24 bit or dither... simply complaining about his approach to proving a point... very one-sided with little apparent regard for other possibilities and elements that can play an important part in the overall outcome. There are many elements that make or break an audio project... mic placement, proper use of eq, proper attention to phasing principles, good sense of space and instrument placement in the mix, good talent, good performance... etc... etc... etc... not just the difference between 16 and 24 bit digital or whether the project was dithered or not. Let's compare the final tape with the original, all other factors being equal... clearly the tape was damaged by SAW. Trust, that I'm sure I could set foot into any mastering facility, misuse the equipment and produce some very bad results, using this to misguide my clients into blaming the facility. Is this really conclusive evidence, or just plain BS?
This is an interesting series... Mr. Katz continues... " …A lot of very good scientists and psychoacousticians have spent years
developing 72 bit-accurate dithering algorithms to capture the highest
audio resolution, and they would be quite shocked to learn that their
efforts are unnecessary." I'll bet that many noted scientists of the day were also shocked to discover that the world was round. The boundaries of science have always been challenged and pushed to new limits since the beginning of science. "Mr. Lentini is restating a very old and specious
argument that the noise which is added by dither is more problematic than
the distortion which it fixes." Still an ongoing subjective debate that by no means determines the extent or legitimacy of an individual's knowledge or a product's worth. This argument is simply one of opinion and will rage on for a very long time. How can you be right about something like... Coke is better than Pepsi? " …In fact, I go over the issue of compromise quite well in my dither article,
with my (now famous?) quote: "Dither, you can't live with it, and you can't
live without it", which was designed to point out that there is an audible
veil added by dithering (even the best dithering), and that the best way to
avoid any compromise and the best way to maintain the highest quality audio
is to retain long wordlength results (high resolution), until the final
16-bit product is produced. The best way is to add the best quality dither
you have *once* and only *once* at the end of the project. Clearly, if you calculate to 24 or more bits, then add dither and store at
16 bits, then you recalculate and reprocess to a long wordlength 24 bits,
and then add dither again and store at 16 bits---the cumulation will add
quite a veil over your sound. Even one generation of dither is quite
audible, but to my ears (and those of a lot of audiophiles), that
generation of dither is far more pleasant to the ear than simple truncation
as is done by Mr. Lentini in his product (if I read his statement
correctly). There is no such thing as a free lunch. I think I understand
the need to use dither and the compromises involved in its use quite well,
and I wonder if Mr. Lentini has really thought as deeply about this issue." Mr. Mithat Konar, director of engineering of Biro Technology adds... "Judging from his writing, I believe Mr. Lentini has NOT thought very deeply
about the issue of dither as he seems to fundamentally misunderstand its
purpose and capabilities. Rather than take up the required bandwidth to set
him straight, I'll simply refer him to any of the JAES articles on the
subject--those by Lipshitz and/or Vanderkooy in particular being great
places to start. However, I WILL take up the bandwidth to restate the
following often-stated and well-proven conclusions: (1) Dither is NOT simply a means "to mask the low level digital breakup".
(2) Dither extends the resolution of a digital system below the LSB threshold.
(3) Dither decorrelates truncation errors from low-level input signals. In
layman's terms this means dither eliminates a particularly nasty form of
distortion in low-level signals. (Implicit in the above is the assumption that the signal being subject to
the dithering operation has resolution below the LSB threshold. Virtually
any gain change, filtering, or mixing operation will result in "new"
information below the LSB of the input.) I submit it is the most profound professional irresponsibility for anyone
to make claims that contradict the above until s/he has better researched
the topic and can DISPROVE these PROVEN findings. Or, to use the same
language that Mr. Lentini has used, to assert anything contrary to the
above statements "IS AN OUT AND OUT LIE!"" Gentleman #1 adds... "You really should put a muzzle on Lentini. As much as he may insist the
opposite is true, he is DEAD wrong on a number of issues and points out that
he has some really grotesque misconceptions about how things are really done
in the real mastering world. I won't even enumerate these mistakes right
now since I'm leaving in 10 minutes for New York for the AES show (I'll be
doing a workshop on 96 kHz mastering with Daniel Weiss, Bob Stuart, Mike
Storey, Andy Moorer, and Ian Dennis. Mr. Lentini might really benefit from
being there.) I really don't think you want Lentini representing your
company, for a number of reasons. I assure you he looks quite foolish to
those of us who really do know what we're doing. Furthermore, he's
disseminating blatant nonsense to a confused public who trust him. When I
get back, I would be glad to enumerate each mistake he made in a posting to
the Pro Audio Mailing List at pgm.com, point by point, in a civil manner,
and let him know where he went wrong. I'm sure he's probably getting
bombarded by mail like this now, but there's no conspiracy here. He just
screwed up BIG time, plain and simple, and it'll soon be time for a bit of
explanation and/or back-pedaling." And in another message adds... "To whom it may concern, I URGE YOU TO READ ALL OF THE FOLLOWING I feel the need to express my profound concerns over the rantings of
your CEO and owner, Mr. Richard Lentini. (Did I miss something or was this Dick's Soapbox?) I will be brief here, but I would be glad to expound in detail any of the points
I touch upon. It's easy to see Lentini's diatribe for what it is: he's obviously (and
understandably) frustrated by a fickle customer base that has found
itself, all of a sudden, concerned that IQS's product isn't good
enough. Join the club. As manufacturers, we all have to deal with this
problem from time to time. What I object to is the GROTESQUE
misinformation he spews forth. Unsettling as it may be, I hope he
didn't believe what he was saying and that this was just a misguided
application of marketing hyperbole. At the very least, Lentini displays a thorough and frightening
misunderstanding of the most basic concepts of dynamic range, noise
shaping, and dither. Additionally, he doesn't seem to have a clue about
how compact discs are mastered or how other products work. Furthermore,
his ridiculous comments about dither and "rocket science" clearly
indicate he has never listened to music in a real mastering environment
or on a properly calibrated monitoring system.... What's worst of all, Lentini is misleading droves of uneducated
customers who count on him to provide guidance and the best product
possible. Rather than lashing out like a cornered animal, he'd have
been better off quietly spending a little time and money on R&D,
admitting to himself that he doesn't know everything about DSP and audio
signal processing, and making the product better and more up-to-date.
If he's succeeded in convincing the legions of SAW users that he's right
and the real experts are wrong, he's created a monster: a diminution of
the state of the art and the mistaken belief among a wide community of
users that their recordings sound as good as they can be. Not to be overly didactic, but Lentini is so wrong on so many counts
here that it's hard to keep track. He can scream and rant all he wants,
but there's a community of experts who will be glad to set the record
straight. I'm talking about people like Dr. Stanley Lipschitz and Dr.
John Vanderkooy (Univ. of Waterloo); Julian Dunn, Ian Dennis, and Graham
Boswell (PrismSound); Jim Johnston (AT&T); Dr. James Moorer (Sonic
Solutions); Mike Storey (dCS); Dan Lavry and Bruce Hemingway (dB
Technologies); Bob Adams (Analog Devices); Daniel Weiss (Weiss
Engineering); Bob Stuart and Rhonda Wilson (Meridian); John LaGrou and
Fred Forssell (Millennia Media); and me, to name but a few. I've never been moved to write a letter to another manufacturer, but
Lentini's rant was just so out-of-line I felt I had to say something.
He should admit he's wrong, apologize, join the world of civilized
discourse, and learn a thing or two along the way. Everyone would
benefit." Whew!!! What seems most noteworthy about these excerpts is the fact that these letters were received from professionals proudly stating their names and titles, and the fact that many of them were on their way to important seminars and meetings at AES. In fact it was even recommended that I should attend so I might learn some of the real truth about audio. Did you notice the professionalism and debating skills demonstrated by such profound responses like "You really should put a muzzle on Lentini". Why the shift to personal assault? This whole thing simply began as a conversation about the usefulness of 16 bit workstations. Varying opinions are one thing, but the stoop to direct personal assault and the unqualified judgements about one's knowledge on a subject, hardly qualify as a professional attitude and do little to win anyone's respect. First of all, I must say I am truly amazed by the clairvoyance and the psychic abilities of these professionals, to be so self-assured about what I do or don't know about digital audio, based on an opinion I published about one small facet of it. In fact, its clear they must know all about my 28 year audio career, down to every detail of my circuitry design, construction and application in both the pristine studio environment and also the live theatre environment. But this is all irrelevant... I challenged one of their beliefs... that challenge nulls and voids any other possible knowledge or skills I might have... and now I must be punished! "And I wonder if Mr. Lentini has really thought as deeply about this issue...". And I wonder how much code these professionals have written, and how many dither and DSP algorithms they designed and experimented with before forming their opinion. And I wonder how many bit patterns they have studied in hex and audio editors looking at the effects of dither on the LSB of a 16 bit signal, before making their decision about this issue. Don't give me a list of articles or books to read. I'm sure I can double or triple your list with the books sitting right here on my shelf. Perhaps what they are missing here, is that maybe I have given this subject much thought, and maybe I even experimented with dozens of dither algorithms, and maybe I even wrote code, and then more code, and applied it to my 16 bit audio data, and maybe I read every article I could find on the subject and maybe, just maybe, I wasn't afraid to think for myself and just happened to conclude that maybe, in the end, a lot of this hoopla is just that... hoopla! Not to worry... just the ramblings of an insignificant idiot who obviously knows nothing about digital audio... it's okay though... I feel better now. Yes I know all about the effects of pulse width modulation and the psycho-acoustic affects attributed to dither and the results shown in the FFT's with dithered data.... I can read too, but thank you all for trying to enlighten me. I never said anything to discount all that research... simply that maybe 16 bit audio, properly processed and handled might be okay without it. Amazing that a simple statement like that would condemn me to death!
Here's one that I really like: "To Whom It May Concern: I was just getting ready to evaluate the latest PC digital audio systems
for myself and the companies I consult for. Please thank Bob Lentini for
saving me the trouble of evaluating the SAW system. A company that has a
CEO that can ramble on with such obvious lack of actual clue could not
possibly design a product that I would be interested in. See you at AES? Sincerely, Tom Garneau
Audioactive Productions" My concern here is for the companies he consults for. It's obvious he is qualified to make the best choices for their needs. Now that I know his powerful secret for picking the best products, maybe I too can become a consultant. Let's see... This company's logo is pink... I hate pink... therefore this company's products are no good. or The owner of this company eats meat... therefore this company's products are not healthy to be around. or This product is the most expensive in its category... therefore it must be the best.
This next one represents some of the side-effects of becoming too righteous about how much you know. Herbie Robinson of HR Music Systems writes... "Let me correct some more technical inaccuracies here (In addition to
what Bob said). I'm not going to comment on the obvious...
If Bob L. really believed that BS diatribe about 16 bits being enough,
then why does he devote the next half of his artical explaining why all of
the processing in SAWxx is done in more than 32 bits?" Uh oh! ... has he got me? Keep digging for dirt... come on... you can nail me on something! Because in a virtual mixing environment, which SAW is, a real and most noticeable concern, in my opinion, is the truncation of bits at the MSB location. Without the 32 bit summing buffers, a few tracks of normally recorded audio would overrun the 16 bit storage capacity and truncate data at the Most Significant Bit, which is unquestionably audible to all that listen, not only the pristine golden ears of the audiophile. In answer to one of my paragraphs... "> >This translates into the ability to sum over 65000 tracks of 16-bit
> >audio recorded at a 100% clipping level without overrunning
> >the internal wordlength capacity. Do you think that can handle your 24
> >stereo tracks without destroying your audio? This is only true if no FADERS are allowed for. In other words, this is
pure BS from a practical standpoint." Maybe I'm missing something here, but I fail to see what FADERS have to do with it. Summing this way allows for 65535 tracks at max 16 bit level to be summed without overload clipping distortion being introduced. So... what's the pure BS from a practical standpoint mean? This is extremely important in practical applications as it allows you to record the number of tracks you need for your project at a good hot level and be confident that you can mix them together without internally clipping the data. Without this capacity, you would have to record each track at a significantly lower level, truly then running into lesser audio quality due to not maximizing the use of the 16 bits you do have to work with. "> >...hand-coded 32-bit assembly language integer math, which bypasses many
> >of the high level language limitations in controlling
> >wordlength spillover during intermediate calculations. SAW takes
> >advantage of the full 64-bit double register wordlength
> >available in the 486 and Pentium processors in each of its calculations
> >for processes like volume and eq and
Message 3/4 09-Nov-98 @ 11:42 PM - RE: 16 bit - 24 bit ????
and... related but intersting
and her's the WIN OS article....
and her's the WIN OS article....
| Ah yes...Windows 95... we've all heard the story... 32 bits... faster... better... true multitasking... etc.Fact or Fiction? If you are like I was, I believed these stories... I went out and re-bought all my favorite software re-written for the 32 bit environment, expecting incredible speed increases and performance stability beyond the likes of Windows 3.1. As a software developer, I invested thousands of dollars in books and new 32 bit development software, and countless hours studying all the new functions involved in the Windows 32 bit API. I spent months translating my software over to the 32 bit platform completely re-designing (rather than just translating) the SAWPlus engine to take full advantage of all the new thread technology and priority capabilities of this new environment. My results.... well.... not quite as promised. First, I needed to up my machines power considerably. Okay, I figure that's good for all my software usage... but where's all the speed increase I was supposed to get? I now have a faster machine, with larger hard drives, but my performance is about the same, and my hard drive freespace is the same or less because the operating system now takes up 10 times as much room... what's going on here.... Sound familiar? You bet. Making the switch to 95 was not necessarily all it was cracked up to be. A very nasty realization came to me about two days before the NAB 97 trade show opened in Las Vegas. I had been developing SAWPlus32 on Windows NT for a period of about 7 months, and had everything ready to go, or so I thought. My trade show demos make extensive use of video and MIDI, and as of yet, there were no MIDI drivers available for NT and our AVI Viewer had not been moved to 32 bits yet, so I was forced to use Windows 95 on all of my demo machines. No problem, maybe I'll take a slight performance hit, but all in all, things should be the same as NT Right?? WRONG!!!! After all, 95 is a 32 bit operating system just like NT, Right?? WRONG!!!! Well, at least the thread technology, that I worked so hard on will be a performance enhancement Right?? WRONG!!!! Was I in for a surprise. Ok, call me naive, but I read all the trade magazines. No one warned me about these findings. Everyone seemed to keep pushing me into the advantages of this new all powerful 32 bit operating system. Every company was quick to get their software moved over to the 32 bit API... what is all this commotion about.... Better Performance, or More Profit? It seems, once again, that the magazines did not print the whole story. It's hard for me to believe that out of all the hundreds and hundreds of articles that were published about Windows 95, not one that I read ever mentioned the facts that I tripped over two days before the convention. Maybe I just wasn't paying attention. My question is... Why all the Hype and Sensationalism... why misrepresent the facts... why not just print the Truth! If Windows 95 was just advertised for what it is... a 16 bit operating system over top of DOS 7 with Win32 extensions built in... I, for one, would have been spared a lot of grief and disappointment. I'd be willing to bet many other developers have been mislead down the 32 bit development path expecting different results than what they achieved also. If I'd have known ahead of time that the handling of threads in Windows 95 has nothing to do with the way threads are handled in Windows NT, I might have designed the SAWPlus32 engine quite differently or might have originally considered two different versions, one for each operating system. Unfortunately, this knowledge came to me way too late to really do anything constructive about it. Meanwhile I had a show to do and a product to release. As it turns out, I delayed the release of SAWPlus32 for another four weeks while I tried to get some compromise between the two operating systems. Of course all the public sees is the fact that the product is late... they have no idea of why... and do they really want to know the truth or do most want to keep their rose colored glasses on about the wonderful new operating system they just upgraded to. The truth is... Windows 95 is mostly a 16 bit operating system running on top of DOS, just like Windows 3.1. It appears that all multimedia calls (Audio, MIDI and VIDEO) are 16 bit code... all GDI calls (Drawing to the screen) are 16 bit code. Need we go further? No. A program like SAWPlus32 is already completely crippled... all my tedious work writing 32 bit function calls is being thunked down to 16 bit translated calls. Every time 16 bit code is being exercised, all other threads appear to be locked. Well, sheesh, lets buy a Ferrari, pull out all the spark plug wires but one, and then complain to Ferrari about their lousy design. This is not a pretty picture. Just about everything SAWPlus32 needs to do forces it into 16 bit code. Things get worse... it also seems that the 16 bit apps actually appear to be getting priority, cutting off some of my time critical threads designed specifically so they will not be cutoff by other activity. As an example, we wrote a test program that runs many separate threads at the same time, each drawing a different colored shape, like a square, a circle, and so on. Running this on Windows NT delivers results as promised... what appears to be simultaneous drawing of different colored shapes, all evenly happening at the same time. Running this on Windows 95 displays quite different results... a bunch of blue squares first, then a bunch of red circles next, and so on... quite different from the original design concept. Now apply those results to a program design like SAWPlus32 which is designed to do multiple things simultaneously in order to accomplish its task, and break up the simultaneous into separate chunks of one thing at a time, and maybe you begin to see the problem. Am I a happy camper... what do you think? Now, what does all this mean? Well, hopefully being armed with this information makes it easier for you, the public, to understand the delays involved in getting high quality software to be compatible between two highly different and incompatible system designs. Also maybe it helps you to understand our published statements about obtaining better performance in the NT platform. This is not meant to say that Windows 95 is useless... only to deliver some facts about how it was misrepresented. To say the least, performance expectations might be less than anticipated when running code designed to take full advantage of the 32 bit Windows API. In fact, SAWPlus32 in 95 does obtain a slight performance increase over the 16 bit SAWPlus, but running it in NT displays the real performance capability of the re-designed engine. Unfortunately, switching to NT is not without its problems also. First, it requires an even healthier machine than 95, with plenty of RAM at its disposal so it can breath. Next, and possibly the most important thing that keeps NT from being the complete solution at the time of this writing, is the lack of drivers, specifically MIDI drivers, making it hard to run a full fledged SAWPlus32 rig. These NT problems will probably be corrected soon, at least there is a chance of a solution around the corner. Windows 95 being fixed... oh well... don't hold your breath! Bob Lentini CEO/Owner - Innovative Quality Software | ||
Addendum - Added 9/15/97 Since the original release of this article, in learning more about the intricacies of the Windows 95 platform, I’ve discovered ways around some of the limitations brought forth in the article as far as the SAW32/SAWPlus32 interaction with Windows 95 is concerned. By removing some of the multimedia function calls from inside critical timing loops and replacing them with Assembly language routines that perform the same functions, I was able to bypass the thunking mechanism that was interfering with the multithreaded operation of the program. This results in much smoother SAW32/SAWPlus32 performance in Windows 95 than was possible in the earlier versions. So whether Windows 95 or NT is your choice of operating systems, feel confident that the newer versions of SAW32/SAWPlus32 will deliver the enhanced performance that we promise. Bob Lentini CEO/Owner - Innovative Quality Software | ||
Message 4/4 10-Nov-98 @ 08:53 AM - RE: 16 bit - 24 bit ????
puh.....so where does that leave us then??.....not in 16 bit land.....
Dunder
Dunder
Viewing all 4 messages - View by pages of 10: 1
There are 4 total messages for this topic
Reply to Thread
You need to register/login to use the forum.
Click here to Signup or Login !
[you'll be brought right back to this point after signing up]
Back to Forum
