The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Q for Paul Frindle Dynamics Plugins
Old 2nd March 2007
  #151
Lives for gear
 

Quote:
Originally Posted by innesireinar View Post
Hi Paul,
Happy to see you here.
After tons of posts about analog summing vs. ITB summing around various fora, which is your finally word about this war?
I think you, who have designed the best analog consoles and the most famous plugs for PT, could be the highest authority for saying something about that.
I remember the thread on PSW but now after two years is there something changed about that?

Cheeres

ranieri senni

No, nothing has changed. The idea that computers are somehow unable to add up is as absurd as ever.

As the PSW thread points out, the reason why summing OTB sounds different is that you are going through extra stages of filtering and reconstruction. Apart from adding extra potential errors to the signal (depending on how good the converters actually are), this also has the effect of behaving differently (and often unpredictably) to the over-hot signals that are currently fashionable - amongst people wrongly persuaded that bits are somehow 'lost' at lower levels.

The bottom line is, that it maybe the case that all this 'experimentation' might end up producing sounds you like artistically (and why not), but from the point of view of accuracy, repeatability and brute signal quality (particularly at the users end) you are better off remaining in the digital domain and just modulating at sensible levels instead :-)

Forget the term 'resolution' - it is a marketing term that has no meaning in digital audio. A digital audio PCM signal is fundamentally different from an LCD TV or a digital camera and therefore does NOT behave in any similar way!

A 24 bit signal has 140dB or so of dynamic range - plently more than you will ever need and quite enough to accomodate sensible levels without audible quality loss :-)
Old 2nd March 2007
  #152
Gear Nut
 
innesireinar's Avatar
 

Thank you Paul.
Which could be the future next steps (if any) for improving the quality of the digital sys in the next future?

And apart from the fact of having a big thing under our fingers why there are still people who spend 1 million for analog boards like a 9000j or a 88R?
Old 3rd March 2007
  #153
Lives for gear
 

Quote:
Originally Posted by innesireinar View Post
Thank you Paul.
Which could be the future next steps (if any) for improving the quality of the digital sys in the next future?
Well the peripherals like conversions could still be improved - (if there is a will in the market to pay for it anymore) and of course processing will continue to become more powerful and/or cheaper. But in my honest opinion the biggest factor to improvement in audio is still dependant on people using it correctly. This means both the designers and kit manus making stuff that is intuitive, artistically capable and disseminating real understanding about the technology - and lastly but most importantly users need to become much more canny and lose the existing fads and misconceptions. To do this they must realise that this is now largely a consumer type market with similar ethics and marketing practices. So it is now up to users wanting to use this stuff seriously to themselves find out the truths from the myths. IMO this is the biggest single challenge to overcome - not the technology itself.

Quote:
And apart from the fact of having a big thing under our fingers why there are still people who spend 1 million for analog boards like a 9000j or a 88R?
Thats a complex question involving preferences, perspectives and the type of work one is engaged in. It is difficult to see how the monster analogue desks can survive when even the era of really capable monster digital desks has passed for most of us. Some applications like film and broadcast and other immediate work flow situations will always require (possibly large) manual working surfaces. There will remain (at least for a while) those artists and engineers so historically successful that they can continue to demand their environments of choice. But for most of us, the devaluation of the recording industry is such that it can no longer begin to support such costs and we are consigned to working with workstations and controllers.
As these become more and more powerful they start to challenge the power of even the largest and most capable of the (now obsolete) digital consoles and far surpass the brute capability of many analogue systems from the past in terms of facilites. But whilst workstations have opened up completely new dimensions in artistic capability and associated art, IMVHO they could still learn from the experience and accumulated knowledge that existed in the era of the analogue console designers :-)
Workstations seem to have grown independently into mixers, from their origins of HD storage and manipulation systems, as the processing cost allowed it in that market place - without the full benefit of the knowledge that mixer designers had amassed previously? At this late stage in the game with so much market penetration and so much kit and associated industry practices concreted into place, the biggest challenge is to take on board these issues and actually affect some useful (if painful) changes? With the attitudes that currently prevail - IMHO it is still more practical for the user to understand these issues and circumvent them where possible, rather than waiting for any fundamental conceptual changes..
Old 4th March 2007
  #154
Gear Nut
 
innesireinar's Avatar
 

Thank you for your answers. You have been very clear.
And about the 44.1 - 96 - 192 KHz war?

Sometimes I hear people who state to hear differences between 96 and 192.
Is it true that processing audio (eq, dyn, etc) at high SR the SWs (plugins, etc.) work better and more comfortable?

About the use of big surfaces (controllers or large analog consoles) I would like to add that every time I worked on them I've always been in the trouble of having to return in the middle of th desk after each settings done on the channels on the boundary of the desk, especially when panning. Therefore I have been always a fun of small controller and staying in the middle of the monitors as much as possible.
Old 6th March 2007
  #155
Lives for gear
 

Quote:
Originally Posted by innesireinar View Post
Thank you for your answers. You have been very clear.
And about the 44.1 - 96 - 192 KHz war?

Sometimes I hear people who state to hear differences between 96 and 192.
Is it true that processing audio (eq, dyn, etc) at high SR the SWs (plugins, etc.) work better and more comfortable? .
The best way to make sense of all this is to separate the notions of processing and delivery rates.

For delivery:
There is no particular quality based reason to sample higher than needed to reconstruct the signals we can hear + a sensible bit extra to allow the filtering to be done in reasonable latency.

IMO for playout 44.1K is a little close to the mark. 48K is absolutely fine. Higher rates (and higher passbands) can actually reduce performance, as the freqs involved can stress reproduction system and cannot be heard. More is definitely NOT always better :-(

For processing:
Firstly for non-linear processes (like distortion which can benefit from higher rates) it must be understood that increasing sample rates only reduces some problems - it never entirely fixes them - it's therefore a losing battle. So in situaions where non-linear processing is used increasing sample rates may reduce the errors, but never completely lose them. If you need to process at very high rates for such things, it would be silly (and expensive) to impose this waste of power on the whole system! So you keep this processing internally, only for the situations it is needed.

Secondly for some processes that operate in the freq domain (EQ etc) we only have delay to work on (i.e there are no freq sensitive components such as capacitors and inductors like in analogue). Since delay is an integral part of the process - increasing sample rates can actually make things worse because the delays reduce. For instance, for an EQ as sample rate increases we need higher and higher math precision (bits) to maintain equivalent performance. Notionally at infinite sample rates no precision however high would allow the function to work and it would be impossible! There really isn't any point in sampling them at faster rates than required - there is nothing to be gained - but loads to be lost if you are not careful in the design :-(

So the simple and sensible answer is that input and output signals and internal processing should as far as possible optimally reflect the required range we can actually hear. Upsampling should be used only when needed internally within specific processes (to reduce aliassing errors). There is no benefit to the user from oversampling his whole system - it is a colossal waste of processing that halves the power of your H/W each time rates are doubled - basically for no gain to you at all!!

Quote:
About the use of big surfaces (controllers or large analog consoles) I would like to add that every time I worked on them I've always been in the trouble of having to return in the middle of th desk after each settings done on the channels on the boundary of the desk, especially when panning. Therefore I have been always a fun of small controller and staying in the middle of the monitors as much as possible.

It's definitely true that mixing on a big console can be difficult when you have to keep getting out of the sweet spot etc.. I have fought this in the old days with frustration - you do get used it, but it's never perfect.

There's no doubt that smaller control surfaces (that don't compromise operational ease) would always be the goal. The assignable surfaces that we concocted in the late 80's and early 90's were for this specific purpose - they were potentially cheaper and much kinder to control room acoustics generally. The problem though is deciding how to assign control banks and facilities to preserve workflow. In those days people pretty much played their desks a bit like instruments, mixing often instinctively. The slightest extra intrusion to control things was pretty unwelcome. In fact many people predicted that the assignable surface would never be acceptable! Now of course many of us mix one handed on a screen with a mouse! How times and concepts change :-)
Old 6th March 2007
  #156
Gear Guru
 
lucey's Avatar
 

Verified Member
Nice stuff Paul. The myth of "ooh, aaah, higher sampling rate" is totally out of control.

If only the CD were 24 bit 48k we'd have a much simpler and better world. Even if it was only 40 minutes long ...
Old 6th March 2007
  #157
Lives for gear
Quote:
Originally Posted by Paul Frindle View Post
Upsampling should be used only when needed internally within specific processes (to reduce aliassing errors). There is no benefit to the user from oversampling his whole system - it is a colossal waste of processing that halves the power of your H/W each time rates are doubled - basically for no gain to you at all!!
i've always wished for multiple sampling rates in the same session. do you think this is something that could be implemented practically by DAW manufacturers?
Old 6th March 2007
  #158
Gear Maniac
 

Thank you Paul for your excellent posts. It's a great support for those who are still trying to defend lower (= reasonable?) sampling rates

Quote:
Originally Posted by lucey View Post
If only the CD were 24 bit 48k we'd have a much simpler and better world. Even if it was only 40 minutes long ...
Are you sure the end product needs more than 16 bits ? I've recently asked a few major mastering engineers for a sample where the 24/16 bit difference is clearly audible. Two seemed interested but so far they didn't send anything.
It seems pretty hard to ABX (or other double-blind test) 16 against 24 bit under normal listening conditions. Needless to say that for recording and intermediate processing 24 bit are very comfortable.
Old 6th March 2007
  #159
Lives for gear
 

Quote:
Originally Posted by Kees de Visser View Post
Thank you Paul for your excellent posts. It's a great support for those who are still trying to defend lower (= reasonable?) sampling rates

Are you sure the end product needs more than 16 bits ? I've recently asked a few major mastering engineers for a sample where the 24/16 bit difference is clearly audible. Two seemed interested but so far they didn't send anything.
It seems pretty hard to ABX (or other double-blind test) 16 against 24 bit under normal listening conditions. Needless to say that for recording and intermediate processing 24 bit are very comfortable.
I think 16 bits is perfectly adequate for play out purposes :-)
Old 6th March 2007
  #160
Lives for gear
 

Quote:
Originally Posted by raal View Post
i've always wished for multiple sampling rates in the same session. do you think this is something that could be implemented practically by DAW manufacturers?
No, it's not multiple sample rates in the same session that you need, it's simply the internals of processing that can benefit from it. You don't have to see that or invoke it yourself - it's simply a plug-in design issue :-)

If what you are saying is that certain plugs sound better at higher overall DAW rates - it's simply that these plugs were not optimally designed for the base sample rates, or the designers did not think the differences would be important to you.

P.S. I should add to this that because of current trends in production and the market, there are other reasons why a plug might be designed to always run at the sample rate of the host system and avoid over/undersampling of any kind. As designers we must be extremely sensitive to what people perceive and expect to see - whether strictly correct or not. We are not going to change the perceptions of the whole market place with one product :-(
Old 6th March 2007
  #161
Lives for gear
 
Empty Planet's Avatar
 

Many thanks, Paul. I thought I'd grokked these concepts from your posts in various threads, but it's good to see a lot of it consolidated here.

A huge service for those of us whom these data leave a bit


Cheers.

Old 7th March 2007
  #162
Gear Nut
 
innesireinar's Avatar
 

[QUOTE=Paul Frindle;1164198]
Secondly for some processes that operate in the freq domain (EQ etc) we only have delay to work on (i.e there are no freq sensitive components such as capacitors and inductors like in analogue). Since delay is an integral part of the process - increasing sample rates can actually make things worse because the delays reduce. For instance, for an EQ as sample rate increases we need higher and higher math precision (bits) to maintain equivalent performance. Notionally at infinite sample rates no precision however high would allow the function to work and it would be impossible! There really isn't any point in sampling them at faster rates than required - there is nothing to be gained - but loads to be lost if you are not careful in the design :-(

This means that the delay in the plugs is sample/based (same amount of samples at any SR) and therefore at higher rate the delay (time/based) is halved? And the plug itself has less time for processing?
If so it could not be possible to design plugs that when run at high SR the amount samples of delay is twice for having the same time for the plug for processing?
Maybe the plugs could work more comfortable at high SR with the benefit of this high SR for FX plugs that you have referred before?
Old 7th March 2007
  #163
Lives for gear
Quote:
Originally Posted by Paul Frindle View Post
If what you are saying is that certain plugs sound better at higher overall DAW rates - it's simply that these plugs were not optimally designed for the base sample rates, or the designers did not think the differences would be important to you.
so what to do about this? imho some plugs do sound better at 96kHz, and some converters do too but that's probably not the case for all converters or all plugs.

a) do you think a safe bet could generally be 96kHz, or is that too much?

b) if DXD comes around eventually do you think it will be worth pursuing? i've read there should be no aliasing problems at that speed.

from your comments i gather the speed thing is not only a matter of diminishing returns, but eventually a self defeating process. so my question is what you think a good average could be, and if you think most plug manufacturers are on the same page as to what the end user will need.

thank you paul.
Old 7th March 2007
  #164
Gear Guru
 
lucey's Avatar
 

Verified Member
Quote:
Originally Posted by Kees de Visser View Post
I've recently asked a few major mastering engineers for a sample where the 24/16 bit difference is clearly audible. Two seemed interested but so far they didn't send anything.
That's weird (of course it does take an extra 5 minutes)

Quote:
It seems pretty hard to ABX (or other double-blind test) 16 against 24 bit under normal listening conditions.
Just line the files up, mute one and hit it's solo button a few times with your eyes closed. Yes?

Quote:
Originally Posted by Paul Frindle View Post
I think 16 bits is perfectly adequate for play out purposes :-)
I have to disagree, but only a little. 16 bits is of course sufficient, and 16 vs. 24 is very subtle in most material ... but in anything big and well recorded in the first place, it's a real difference of depth and detail. If we were at 24/48 in all video and audio it would not only be sufficient but uniform. We're used to 16 bits and we record in a way that makes it seem sufficient, but lets assume top converters or top analog tape ... and a great mix of dynamic material ... you'd hear the lost bits everytime. Of course that's all very rare, so we say 16 is sufficient.

In heavily compressed material it's also audible, but only on a good playback system, else the distortions overrule everything about the lost sonics.




Sorry to derail ...
Old 7th March 2007
  #165
Gear Maniac
 

Quote:
Originally Posted by lucey View Post
Just line the files up, mute one and hit it's solo button a few times with your eyes closed. Yes?
Ehmm, no. Although that's a typical way of listening in the recording industry, it's not a double blind test and therefore not reliable. There is quite some info about ABX testing on the internet, like here. It's amazing how easy it is to deceive yourself in non-double-blind tests. That doesn't have to be a problem, as long as you realize that any conclusions based on such tests will be highly subjective.
Let my try again here: is there anyone who has a (<60sec) sample where the difference between 24 and (properly dithered) 16 bit is clearly audible, and is willing to share a copy for testing purposes? I'd be very grateful.
Old 7th March 2007
  #166
arf
Gear Addict
 
arf's Avatar
 

Quote:
Originally Posted by Kees de Visser View Post
Ehmm, no. Although that's a typical way of listening in the recording industry, it's not a double blind test and therefore not reliable. There is quite some info about ABX testing on the internet, like here. It's amazing how easy it is to deceive yourself in non-double-blind tests. That doesn't have to be a problem, as long as you realize that any conclusions based on such tests will be highly subjective.
Let my try again here: is there anyone who has a (<60sec) sample where the difference between 24 and (properly dithered) 16 bit is clearly audible, and is willing to share a copy for testing purposes? I'd be very grateful.
Such a sample is easily made if you set the level low enough. Weiss Engineering put out a sampler CD a while back with a cello recorded in 24-bit at around -50 transferred to CD at 16 bit with dithered and truncated examples. Undithered truncation to 16 bit sounded like typical robot voice effect, different dithers had different tonalities, but at least sounded like a cello with noise added.

24 vs 16 dithered differences are most audible with spacious acoustic music with wide dynamic range and a low averager recording level. You're not going to hear any difference listening to a highly limited full scale track at 24 vs 16dithered. With proper dithering the difference is subtle, except for the noise!
Old 7th March 2007
  #167
Gear Maniac
 

Quote:
Originally Posted by arf View Post
Such a sample is easily made if you set the level low enough. Weiss Engineering put out a sampler CD a while back with a cello recorded in 24-bit at around -50 transferred to CD at 16 bit with dithered and truncated examples.
Many dither demonstration samples have been made like that and that's how I've been testing dither myself for years (low level music and high monitor gain). But IMO it's not correct to listen to the dither at such a high level. Noise shaped versions of dither e.g. have been optimized for levels around the threshold of hearing. By increasing the monitor level, you are violating the very purpose of the shaping by moving the noise above the hearing threshold.
With music that peaks around -50 dBFS, aren't you in fact ignoring 8 MSB's and effectively listening to an 8-bit signal ?
I'm interested in the audible effects of dither (if any) at normal listening levels, in the presence of music at normal recording levels (say peaking between -3 and 0 dBFS).
Old 8th March 2007
  #168
Gear Guru
 
lucey's Avatar
 

Verified Member
Quote:
Originally Posted by Kees de Visser View Post
Ehmm, no. Although that's a typical way of listening in the recording industry, it's not a double blind test and therefore not reliable.
Exactly what's unreliable about aligning two files to sample accuracy, muting one, and hitting solo enough times to confuse yourself as to which is which? That is not only the typical way to compare, it's the only way to do it in a real world session.

rant on:
I'm so amused by these "double blind" scientific posters. When was the last time you folks bought a piano? You'll need to drive 10 minutes to 5 hours between showrooms. There are different acoustics in each showroom. Your home is different yet. And still, with focused listening and a keen ear you can pick the right tone for your space and also reliably predict it's tone in your room! Same goes for auditioning acoustic guitars, or changing tube brands in an amplifyer. Just because you dont trust your ears for more than 1 second dont get all holier than thou on the rest of us and start in with claims of "unreliability".

The alignment procedure I noted is plenty reliable for A/B in a session.

rant off:


Sorry, again ... to derail.
Old 10th March 2007
  #169
Lives for gear
 

[QUOTE=innesireinar;1166437]
Quote:
Originally Posted by Paul Frindle View Post

This means that the delay in the plugs is sample/based (same amount of samples at any SR) and therefore at higher rate the delay (time/based) is halved? And the plug itself has less time for processing?
If so it could not be possible to design plugs that when run at high SR the amount samples of delay is twice for having the same time for the plug for processing?
Maybe the plugs could work more comfortable at high SR with the benefit of this high SR for FX plugs that you have referred before?
Yes - this of course is sub-sampling, which of course is undoing any notional 'good' the higher sampling rate 'might' have had - and simply wasting data in the time domain.

You see, there really is a good technical reason to select a sample rate which is correct for the signal bandwidth you are trying to process. This really is a situation where more is not necessarily better. In engineering there still exists a valid notion of optimum. It is only in marketing that notions of the 'optimal' have been replaced by the 'more is always better' mantra. We must (try at least to) remember that the laws of physics and the natural world do not conform the the principles of marketing and capitalism. BTW - this was also true of analogue engineering, for instance there was never a good technical argument for say a mic amp that also amplified radio signals and other extraneous junk and interferences. The fact that some did by bad design was nothing more that a darned niusance!!
Old 10th March 2007
  #170
Lives for gear
 

Quote:
Originally Posted by Kees de Visser View Post
But IMO it's not correct to listen to the dither at such a high level. Noise shaped versions of dither e.g. have been optimized for levels around the threshold of hearing. By increasing the monitor level, you are violating the very purpose of the shaping by moving the noise above the hearing threshold.
dBFS).
Yes precisely :-)
Old 11th March 2007
  #171
Gear Nut
 
innesireinar's Avatar
 

Thank you, Paul for sharing you knowledge with us...
Talking about bit rate does the same rule work?
Or in this case more is always better?

Some weeks ago I and a my collab. did a listening test by importing in PT some commercial tracks in PT (Brian McKnight and the S John Passion intro) in their original cd format and in MP3 192 format (converted with iTune) and we were not able to listen to a difference between the two format. How it was possible?
And if we could not listen to a difference between 16 bit and a compressed file how is it possible that someone can listen to a difference between 16 and 24 bit?

We did the test by switching two stereo tracks without watching on the screen which file was playing.

I'd like to know if was a our ears problem or if there is not as much difference.
Old 11th March 2007
  #172
Lives for gear
 

Just wanted to chime in and say that this kind of thread is why I scan the forums. Truly a gem here.
Thanks for sharing your insights Mr. frindle , your one smart cookie!
Old 11th March 2007
  #173
Lives for gear
 

Quote:
Originally Posted by innesireinar View Post
Thank you, Paul for sharing you knowledge with us...
Talking about bit rate does the same rule work?
Or in this case more is always better?

Some weeks ago I and a my collab. did a listening test by importing in PT some commercial tracks in PT (Brian McKnight and the S John Passion intro) in their original cd format and in MP3 192 format (converted with iTune) and we were not able to listen to a difference between the two format. How it was possible?
And if we could not listen to a difference between 16 bit and a compressed file how is it possible that someone can listen to a difference between 16 and 24 bit?

We did the test by switching two stereo tracks without watching on the screen which file was playing.

I'd like to know if was a our ears problem or if there is not as much difference.

I don't know why you could not hear it? I have never yet found a complex programme track that is not changed audibly by bit rate compression. Some very simple signals (like a few sine waves) are of course fairly unchanged to the ear as the coder does not need to work hard.

But I suppose one possibility might be that the track had already been compressed and decoded again before the CD was made? In this case the further coding your end would change it very little? But this would seem very unlikely?

It has to be said that the artefacts of coding at high rates can be a bit subtle (and unusual) and when first meeting them it's necessary to learn what they are - after which of course you will hear them all the time and they may begin to get on your nerves :-( ALthough for some very compressed and saturated popular music the encoding actually improves it's sound IMHO..
Old 11th March 2007
  #174
Lives for gear
 

Quote:
Originally Posted by raal View Post
so what to do about this? imho some plugs do sound better at 96kHz, and some converters do too but that's probably not the case for all converters or all plugs.

a) do you think a safe bet could generally be 96kHz, or is that too much?
If it were up to me I would aim from around 60-64KHz, just to give the converter filters an easier time. Beyond that I can't see any point.

Quote:
b) if DXD comes around eventually do you think it will be worth pursuing? i've read there should be no aliasing problems at that speed.
Hmm.. DXD is a 24bit PCM signal at 352.8KHz, 8 times more data than is strictly necessary to reconstruct the audio signal!! The irony of the association with DSD 1bit format should not be lost on people - amongst those who think and remember :-( Of course from a technical aspect it's pointless (beyond trying to process DSD signals that consist of full level out of band noise) - but since when has that fact damaged a good marketing strategy.. Some will appreciate the potential financial advantage of managing to fill new generation Bluray discs with a conventional 45 minute album and convincing people this is worth paying a premium for both H/W and media? I will say no more - except to stress that you must remember that the final quality you will actually hear will (as always) be restricted to fidelity only up to 20KHz..

Quote:
from your comments i gather the speed thing is not only a matter of diminishing returns, but eventually a self defeating process. so my question is what you think a good average could be, and if you think most plug manufacturers are on the same page as to what the end user will need.
Yes it is not only a situation of diminishing returns, it is potentially also a situation of actually worsening the final quality you may get to hear.

The point is that increasing sampling rate is a compromise (like everything else). If a process within a plug-in produces fewer audible artefacts at higher rates than it otherwise might - and the compromise is therefore worth it for that particular effect, then the upsampling should be restricted to that process only - NOT applied to the whole darned system!! :-(
Old 11th March 2007
  #175
Gear Maniac
 

Quote:
Originally Posted by innesireinar View Post
Some weeks ago I and a my collab. did a listening test by importing in PT some commercial tracks in PT (Brian McKnight and the S John Passion intro) in their original cd format and in MP3 192 format (converted with iTune) and we were not able to listen to a difference between the two format. How it was possible?
And if we could not listen to a difference between 16 bit and a compressed file how is it possible that someone can listen to a difference between 16 and 24 bit?
The differences between a good 192 kbs mp3 and 16/44.1 cd audio can indeed be very small. So small that most people won't hear any difference with music they listen to most of the time. Double blind tests confirm this. It's possible to find "problem samples" where coding artefacts are audible, but they are not very common. Not all mp3 (and other lossy formats) encoders give equally good results, especially for lower bitrates. AAC e.g. seems to sound better than mp3 at the same bitrate.
Lossless compression formats prove that it's possible to reduce the bitrate to about 50% without any audible difference (lossless). Reducing bitrate doesn't automatically result in inferior sound.
I too have doubts if it's easy to hear a difference between 16 and 24 bit at 44.1 kHz. So far, I haven't found samples nor tests using music at reasonable levels that confirm the audible difference.
Old 12th March 2007
  #176
Lives for gear
 

Quote:
Originally Posted by Kees de Visser View Post
The differences between a good 192 kbs mp3 and 16/44.1 cd audio can indeed be very small. So small that most people won't hear any difference with music they listen to most of the time. Double blind tests confirm this. It's possible to find "problem samples" where coding artefacts are audible, but they are not very common. Not all mp3 (and other lossy formats) encoders give equally good results, especially for lower bitrates. AAC e.g. seems to sound better than mp3 at the same bitrate.
Lossless compression formats prove that it's possible to reduce the bitrate to about 50% without any audible difference (lossless). Reducing bitrate doesn't automatically result in inferior sound.
I too have doubts if it's easy to hear a difference between 16 and 24 bit at 44.1 kHz. So far, I haven't found samples nor tests using music at reasonable levels that confirm the audible difference.
I agree, lossless bit rate reduction cannot change the signal because it is - lossless. But the programme type that is most obviously changed with lossy reduction (MP3, WMA etc.) is material where there is significant simultaneous dynamic range where something quite loud plays on top of complex recognisable sounds in the background. Solo guitar with orchestral accompaniment is a good example of such material. What happens is the timbre of the orchestra changes dynamcally throughout the guitar expressions..
I find this kind of effect on timbral changes most disturbing, this is perhaps why I am completely turned off by multi-band compressors - which are of course hundreds of times worse than lossy coding :-(
Old 12th March 2007
  #177
Gear Guru
 
lucey's Avatar
 

Verified Member
Quote:
Originally Posted by Paul Frindle View Post
... this is perhaps why I am completely turned off by multi-band compressors - which are of course hundreds of times worse than lossy coding :-(
I agree with you, but why is it so? What is it about the way MB works?

Quote:
Originally Posted by Kees de Visser View Post
The differences between a good 192 kbs mp3 and 16/44.1 cd audio can indeed be very small. So small that most people won't hear any difference with music they listen to most of the time. Double blind tests confirm this. ... (ed) ... Reducing bitrate doesn't automatically result in inferior sound.I too have doubts if it's easy to hear a difference between 16 and 24 bit at 44.1 kHz. So far, I haven't found samples nor tests using music at reasonable levels that confirm the audible difference.
You might not want to post this opinion if you're working in engineering. It's like saying "what goes in to an AD DA is exactly what comes out."

Quality recording is a game of inches, not double blind tests on the general public.
Old 12th March 2007
  #178
Gear Nut
 
innesireinar's Avatar
 

Quote:
Originally Posted by Kees de Visser View Post
The differences between a good 192 kbs mp3 and 16/44.1 cd audio can indeed be very small. So small that most people won't hear any difference with music they listen to most of the time. Double blind tests confirm this. It's possible to find "problem samples" where coding artefacts are audible, but they are not very common. Not all mp3 (and other lossy formats) encoders give equally good results, especially for lower bitrates. AAC e.g. seems to sound better than mp3 at the same bitrate.
Lossless compression formats prove that it's possible to reduce the bitrate to about 50% without any audible difference (lossless). Reducing bitrate doesn't automatically result in inferior sound.
I too have doubts if it's easy to hear a difference between 16 and 24 bit at 44.1 kHz. So far, I haven't found samples nor tests using music at reasonable levels that confirm the audible difference.
Thank you for confirming this, this means that my ears still work.
Old 12th March 2007
  #179
Gear Nut
 
innesireinar's Avatar
 

Quote:
Originally Posted by Paul Frindle View Post
I agree, lossless bit rate reduction cannot change the signal because it is - lossless. But the programme type that is most obviously changed with lossy reduction (MP3, WMA etc.) is material where there is significant simultaneous dynamic range where something quite loud plays on top of complex recognisable sounds in the background. Solo guitar with orchestral accompaniment is a good example of such material. What happens is the timbre of the orchestra changes dynamcally throughout the guitar expressions..
I find this kind of effect on timbral changes most disturbing, this is perhaps why I am completely turned off by multi-band compressors - which are of course hundreds of times worse than lossy coding :-(
I will do a test with this kind material. BTW during my A/B I've taken care to the rev tails that could simulate something like the material you are referring to, especially when comparing the S John Passion files but even in this case I was not able to hear any difference.
Old 12th March 2007
  #180
Gear Nut
 

Hi Paul

thanks to you this is one of my favorite threads! A truly great read :-)

So you would agree that either 48kHz or 88.2kHz fs would be the best choices for practical work in the digital domain?

Which one is to prefer
- if I work with 48kHz then I have to sample rate convert to 44.1 in the end and that's something I don't really like... maybe you can give us some insight into that as well...
- if I work with 88.2kHz then of course it also has to be SRC'd but it seems easier "just" to divide by 2, right????

Thanks for you sharing your knowledge ;-=

cheerio
Roger
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
Elsteve9 / So much gear, so little time
5
djui5 / Gear free zone - shoot the breeze
3
slantbar / Q+A with Mike Shipley
2
AudioGaff / The Moan Zone
1

Forum Jump
Forum Jump