why don't we use predistortion more in the audio world?

Such honesty - kudos :) Its rather odd that the people who don't like digital sound blame it on the theory, or the 'flawed' CD format rather than on the implementations they actually hear don't you think? What might be behind this misunderstanding in your view?


In think Gilbert Briggs said a friend of his seemed to have outstanding hearing . He could without any information as to design say how filters sounded . Briggs said his observations matched the emerging theories . He especially wanted his opinion of the Baxendale circuit . On testing this friend he was found to be very impaired . It seemed to make him more critical and accurate . Michael Gerzon was said to be more than averagely sensitive about distortion . He claimed to be able to hear 23 kHz . I have my doubts about that . Perhaps he was the opposite and too proud to say ? I remember Michael laughing at me when I denounced digital ( not at me , at what I was saying ) . He confessed to understanding my concerns . It was funny , he went almost into a trance and said how it would be . He was right . Michael was the first to attempt real explanations of filter design . Friends make money from his work . FM is a mystery , a true silk purse .

It can all be explained , it is a lifetimes work doing it . Then something new comes along ........
 
The problem is you're trying to do a best fit minimization (akin to least squares) of multiple positions, which in essence means you're trading off full minimization of distortion in one specific spot in order to expand the area within which the listener may be positioned while getting sort-of-OK results. Of course, this is not a knock against Audissey; it's a problem with all speaker systems vs the use of headphones and binaural recordings. The only way around it using speaker systems and users that are not in a specific fixed location is active user tracking, similar to advanced 3D displays that track viewer position and render the correct view for that particular viewer. It's quite difficult to implement, but one would be able to get away with using only two speakers and still get fully positional 3D audio. Such a system would need a sufficiently-finely sampled HRTF measurements of the listening space (think moving around the microphones and making measurements in a grid pattern) so that interpolation between the grid points will give sufficiently precise HRTF coefficients for the DSP to do its predistortion+crosstalk-cancellation+inverse HRTF convolution while maintaining good 3D results.

Nevertheless, Audissey manages well to ignore position-depended frequency response in the room equalizing almost speakers only. I could not do such equalization manually using 1/3 octave LC-filters, it was too complex. Software does that better. However, it is good for reproduction only, due to necessary latency, for live sound I still use good old manual approach, with analog EQ and frequency analyzer.
 
abraxalito said:
Where are the flaws in this article? - they're saying DAB is 20X greener than FM.
One flaw is that the author appears to be a DAB fan, probably for non-technical reasons. There seem to be lots of pro-DAB 'gurus' cropping up. Some are media 'experts' i.e. they seem to earn their living from writing about the media. Given the track record of the digital radio industry, I suspect that some of these 'gurus' are funded by the industry.

Decoding DAB will always use more power than FM, assuming similar generation technology, simply because there is more to do. In the article he appears to compare current FM energy with predicted future DAB energy - a classic marketing trick. Sadly, such distortion can easily fool a politician and in the end it will be politicians who decide.

In the UK many people find DAB to be unreliable because the transmitter power is far too low, and Band 3 VHF does not penetrate buildings particularly well. The poor sound quality is another issue. He is correct in saying that DAB can be better than FM under multipath conditions such as mountains and canyon-like city centres.
 
I listen to LP because I have LPs and do not wish to rebuy them in another format, even where they are available (not all are, or may have been 'remastered').
I'm sure that's true but... I need to know that there's someone who really can laugh in the face of all things that defy 'conventional analysis' (and all your writings so far seem to claim this, including your fairly explicit denunciation of anything mysterious in audio the other day!) so could you just confirm that, in a hypothetical situation, you could transfer one of your vinyl LPs to digital 'transparently' without loss of audible niceness? Why would you want to do this, you would protest? Just to humour me..? Or perhaps one of your copyright-free records has a bad warp, and you know that it cannot be played with your conventional arm settings so you decide to make a CD copy of it. Or maybe in one particular copyright-free record, leaping out of your chair to turn it over half way through destroys the mood, so you want to make a continuous disc of it. I just need to know that you believe it is a straightforward, scientific or engineering matter to do it to an arbitrary degree of exactness. i.e. there is no 'magic' in vinyl that 'defies conventional analysis'..!

There is an exception: predistortion cannot remove noise, while feedback can. Predistortion relies on accurately predicting what the amp will do, but you can't predict noise. Feedback can reduce noise in real time because it simply compares input and output and removes what should not be there. If you tried to make an amp with no feedback, just predistortion, then noise could be a problem. With BJTs you can't have a Vbe signal of more than a few mV because of the exponential characteristic distortion - so poor signal-noise. You would have to allow yourself local feedback, such as degeneration, to get round this but then you no longer have a feedback-free amp.

This is an excellent point. Do we have to use BJTs? Does a delta sigma DAC source for the amp invalidate the feedback-free criterion, anyway? Or is it reasonable to think that it is only the power amp stage driving a difficult load (a speaker) that becomes 'controversial' when feedback is applied, and therefore could justify such an experiment?
 
Last edited:
Nevertheless, Audissey manages well to ignore position-depended frequency response in the room equalizing almost speakers only. I could not do such equalization manually using 1/3 octave LC-filters, it was too complex. Software does that better. However, it is good for reproduction only, due to necessary latency, for live sound I still use good old manual approach, with analog EQ and frequency analyzer.

However, DSP-controlled analog EQ that uses motorized potentiometers, inductors, relays, would be a real dream-machine...
 
However, DSP-controlled analog EQ that uses motorized potentiometers, inductors, relays, would be a real dream-machine...

Forgive me if I'm stating something obvious that everyone knows about, but reading these forums I'm never sure...

Could we just clarify something about room correction and EQ? A room has a 'frequency response', but am I correct in thinking that, this only applies to steady state signals? If I send out a short transient from a speaker, it can reach my ears before any other reflections i.e. there will not be any cancellation or addition due to multiple paths so it will be 'flat'. In the case of most real music the phase/frequency/spectrum is changing rapidly, so millisecond-scale reflections don't give me those permanent, fixed notches and peaks in the response - which is why I fondly imagine that we can all 'hear through' the room when we listen to a system in any surroundings.

If I try to EQ the room to flat using conventional (e.g. analogue) EQ won't I therefore fail on anything but test tones? It may be OK for low frequencies, but really the only solution is to do a proper impulse response correction..?

(again, apologies if this is just stating the obvious to everyone here)
 
CopperTop said:
could you just confirm that, in a hypothetical situation, you could transfer one of your vinyl LPs to digital 'transparently' without loss of audible niceness?
Copying from one format to another will mean that the result is some hybrid of the two e.g. LP surface noise with CD convenience. I have no idea what will happen to 'niceness'. I don't attach any magical significance to LP. It is conceivable that its pleasant sound arises from a gentle HF rolloff (due to vinyl elasticity) and some low level low order distortion, but I don't know.

Audio electronics is a branch of applied science. We don't know all the answers, but we do know that some questions are silly. The unknowns tend to be in psychoacoustics (e.g. setting the requirements for our circuits) rather than circuit theory (which has been rather settled for many decades). As I said, people who don't understand something often like to pretend that nobody understands it.
 
there are "room correction" threads, books, comments from researchers in the subject like Dr Geddes

well established psychoacoustics including precedence, temporal, amplitude, frequency masking perceptual effects
Hi jcx

Thanks, but my point is that people often talk about correcting the room with a 1/3 octave graphic equaliser, and I'm never quite sure what it is they hope to achieve. I'm just trying to work out whether they know something I don't, or vice versa.
 
Copying from one format to another will mean that the result is some hybrid of the two e.g. LP surface noise with CD convenience. I have no idea what will happen to 'niceness'.
Yes, but... you said

I use both CD and LP, and recognise that they both sound nice but different.

If the only drawback with CD is that it is more convenient, then you seem to be implying that digital can be made to sound exactly like LP (by making a digital recording of an LP), so there's no reason why the niceness should not be left intact. If that's not the case, where does the digital fail?

Edit: basically I need to know what the limits of digital are, from someone who really knows about this stuff.
 
Last edited:
Could we just clarify something about room correction and EQ?

It is obvious: room correction is not the best employment of the EQ. Whet it is best employment, is correction of frequency response of speakers in this room. It is quite different paradigm that leads to different approach, and to totally different results.
Also, you may see that I don't participate in discussion of predistortion of non-linearities, especially when they are mixed with phase shifts, delays, and resonances.
 
I was told by a digital designer whose name I never asked that he uses a Garrard 401 to test digital recording devices . He said the strong character of the 401 was often lost . He also said a repeatable test . I heard Led Zeppelin 4 from master tape and CD transfer through the same amp and speakers . I have to say the acetate sounded more like the master tape . This like FM is strange as cutting a LP involves so many twists and turns of EQ and avoiding resonances . I would love to try this with FM BBC style ( 13 bit ) . I suspect it would be the best of all if ignoring the slight bandwidth restriction . How crazy ?
 
It is obvious: room correction is not the best employment of the EQ. Whet it is best employment, is correction of frequency response of speakers in this room. It is quite different paradigm that leads to different approach, and to totally different results.

Hi Wavebourn

Does this mean that while you might aim for a flat frequency response from your speakers in an anechoic chamber, you aim for something different in a normal room? Is this a low frequency thing, only?
 
Hi Wavebourn

Does this mean that while you might aim for a flat frequency response from your speakers in an anechoic chamber, you aim for something different in a normal room? Is this a low frequency thing, only?

Obviously yes. The trick is, using multiple position of microphone in the room eliminate as many as possible variations of the sound field dependent on listening position, and equalize speakers only, as best as it is possible. As I said before, Audissey does that quite well.

Of course, room treatment is a good tool as well; you can't predistort signals for room treatment.
 
I was told by a digital designer whose name I never asked that he uses a Garrard 401 to test digital recording devices . He said the strong character of the 401 was often lost . He also said a repeatable test . I heard Led Zeppelin 4 from master tape and CD transfer through the same amp and speakers . I have to say the acetate sounded more like the master tape . This like FM is strange as cutting a LP involves so many twists and turns of EQ and avoiding resonances . I would love to try this with FM BBC style ( 13 bit ) . I suspect it would be the best of all if ignoring the slight bandwidth restriction . How crazy ?

Knowing my own inability to not hear what I expect to hear when it comes to subtle differences, I am doubtful whether I would be able to meaningfully try this myself. To me, a CD recording of an LP sounds exactly like the LP, but maybe that's just because that is what I expect to hear. I am looking for confirmation of digital's transparency (to an arbitrary degree) from an arch-objectivist who actually understands fully the characteristics of these systems..! Otherwise, there must always be an audiophile question mark over any correction system based on digital.
 
Obviously yes. The trick is, using multiple position of microphone in the room eliminate as many as possible variations of the sound field dependent on listening position, and equalize speakers only, as best as it is possible. As I said before, Audissey does that quite well.

Of course, room treatment is a good tool as well; you can't predistort signals for room treatment.

Not suggesting that you would pre-distort for room treatment, at all. Just curious about your system for speaker/room correction because I'd like to try it myself.

Is it a low frequency thing, only? And when you measure and eliminate the variations of sound field, what are your test signals?
 
revolutionizing commercial music recording practice, distribution formats is decade time scale project that has no visible momentum today

we might want to listen to existing recordings in the meantime
The way neural prostheses are progressing (see for example in the even more complex case of vision http://physiology.med.cornell.edu/faculty/nirenberg/lab/papers/PNAS-2012-Nirenberg-1207035109.pdf), there won't be physical sound reproduction being done at all in a couple of decades. It will be just signals feeding into your nervous system. Note the link I've added is an especially good example as they bypass the first two layers of visual neurons and perform the transcoding normally done by them in silicon. In 20 years this technology won't be just for the disabled but fully mainstream with permanent human-artificial sensor interfaces commonplace--I bet money on that.
 
No, it is a whole audio band thingy only. The one I use is patented by Denon and implemented in their latest receivers. It uses fast sweeps.

You don't happen to know the patent number do you?

In the meantime, is my point about a short, isolated transient reaching my ears directly without interference from any reflections, and therefore 'flat' in frequency response, a valid one? Presumably if the Denon system recommends any EQ of the signal at medium to high frequencies, that short transient will be rendered non-flat in frequency response. I would imagine that the effectiveness would be highly-dependent on the representativeness (is that a word?) of the test signal to real music.
 
The way neural prostheses are progressing (see for example in the even more complex case of vision http://physiology.med.cornell.edu/faculty/nirenberg/lab/papers/PNAS-2012-Nirenberg-1207035109.pdf), there won't be physical sound reproduction being done at all in a couple of decades. It will be just signals feeding into your nervous system. Note the link I've added is an especially good example as they bypass the first two layers of visual neurons and perform the transcoding normally done by them in silicon. In 20 years this technology won't be just for the disabled but fully mainstream with permanent human-artificial sensor interfaces commonplace--I bet money on that.

I hope you are right! Could they also synthesise the physical sensation of the bass drum sound hitting your chest etc.? I imagine the 'trouser flap' might be a difficult one to do, though.