Half a croissant, on a plate, with a sign in front of it saying '50c'
h a l f b a k e r y
If you need to ask, you can't afford it.

idea: add, search, annotate, link, view, overview, recent, by name, random

meta: news, help, about, links, report a problem

account: browse anonymously, or get an account and write.

user:
pass:
register,


                                     

Hifi Audio from High-Precision, Lofi Equipment

Compensate for distortion instead of trying so hard to eliminate it
 
(+2, -2)
  [vote for,
against]

Every component of an audio system, no matter how well made it may be, will distort its input measurably and often audibly. Rather than pursue the inevitably losing battle of trying to perfect audio components, one can instead compensate for the distortion they cause. The compensations are made by a computer program which calculates the precise amplitude each sound sample needs to have to create waves which correspond precisely to the audio data. This means pre-correcting for distortions such as: transistor and speaker nonlinearities; frequency response; speaker inertia; and doppler effects.

There are several approaches to applying this concept for public consumption.
* Every audio system is profiled individually.
* Every model of audio system is profiled.
* Audio systems are designed, and adjusted as needed to match one of several standard profiles (for different mobilities, price ranges, headphones versus speakers and so on...).

Whichever approach is chosen, the sound is processed to precompensate for the known distortions expected from the particular system it is to be played through. One important thing to note, is that the sound has to be processed differently for each volume setting on a system, so one couldn't just distribute preprocessed audio unless the public could agree on a standard volume setting.
Processing could be simplified by having volume controlled with linear attenuators between the amplifier output and the speakers. This practice would allow the amplifier to be driven at a constant level regardless of the desired sound output, but the speakers would still have unique nonlinearities and doppler effects for each power level it is driven at.

Since audio systems would be expected to have their distortions corrected, their designs can focus more on higher gain, lower noise and higher energy efficiency, and relax somewhat about fidelity. Transistor preamp stages could be made as sensitive as possible without regard to distortion (less negative feedback), and thereby get the signal farther above the noise floor.

1/f noise could be greatly reduced by a circuit which rapidly alternates between:
(1) Disconnecting the signal to the preamp and sampling the current noise level.
(2) Reconnecting the signal to the preamp and subtracting the sampled noise level from the input.
This cycle would be repeated at perhaps around 1,000,000 times per second.

If systems are to match a standard profile, then their designers would need to incorporate means of adjusting every component very accurately to ensure sound processed for it would sound perfect when played through it. Electronic components, including transistors, would need to be adjusted at the factory (laser trimming, for example). Do-it-yourselfers could try improving their systems by such things as adjustments to speakers, such as cone mass, cone alignment, impedance and linearity.

For extreme audiophiles:
This technology could be used to remove the last remaining traces of distortion from their expensive, high-end sound systems. Profiles for individual sound systems could be made, and pre-profiled listening-room designs could be used to avoid the expense of a technician's visit.
They could also have themselves, their families and their friends acoustically profiled so they can get together in any number and combination without marring the sound quality as long as they sit in predesignated places and with predesignated postures.
Even the reverberations from a profiled listening room could be compensated for and effectively removed.
There would still be a so-called "sweet-spot" which would have the most perfect sound in the entire room.
Alvin, Jan 10 2012

3d sound / speaker arrays http://www.cambridg...om/technology/audio
[webfishrune, Jan 10 2012]

Class D amplifiers http://en.m.wikiped...i/Class-D_amplifier
[Ling, Jan 10 2012]

An example similar to what I meant, being pursued. http://www.klippel....how/literature.html
Look at all of the non-linear speaker response curves to see what happens to good audio signals. [Alvin, Dec 02 2015, last modified Apr 08 2016]

In the oven! https://hackaday.io...good-sound-from-bad
Ongoing DIY project. Looks stalled; you should ask the developer to get it going again. [notexactly, Dec 02 2015]


Please log in.
If you're not logged in, you can see what this page looks like, but you will not be able to add anything.



Annotation:







       You can't mend a signal if the information in the signal has been lost. E.g. if a component in your audio system is dropping a range of frequencies then no amount of processing will put those frequencies back properly, because you don't know how they vary, what amplitude they were at in the original signal, etc. Your approach works when the signal is very slightly changed by a component in the signal chain. What most often happens is that information in the signal is lost (or not picked up in the first place) and no amount of processing or knowledge of the properties of your amplifier, etc. will get this information back.
hippo, Jan 10 2012
  

       Well I understood that care would be taken to reproduce all frequencies, but not minding if some were reproduced louder, out of phase, or otherwise differently?   

       //There would still be a so-called "sweet-spot" which would have the most perfect sound in the entire room.// If you are having this much analysis and compensation, presumably a camera and image recognition system could track the locations of soft bodies and heads in the room, and adjust the output to keep the sweet spots focussed on the ears?
pocmloc, Jan 10 2012
  

       I think you should already know about class D amplifiers. If not, see link.   

       I think these can be corrected by having a feedback loop. However, if they are designed with your idea in mind, they wouldn't need a feedback loop (which has non linear characteristics with frequency). Since digital switching and integration are easier to model, maybe this would be a worthwhile approach.*   

       *Personally, I think the speakers are by far the biggest problem, and most high end amplifiers are indistiguishable.
Ling, Jan 10 2012
  

       Signal compensation is well-baked I'm afraid: quite a few pro speakers have an option to purchase of a digital-based filter specific to that speaker, and there's devices to shift frequencies around to avoid the doppler effect caused by HF's riding on top of LF's.... digital de-essers, Dolby, the list does go on.
FlyingToaster, Jan 10 2012
  

       hippo,
Yes, the systems would need to have adequate response to all of the frequencies intended to be reproduced, but the response curve could be very irregular without reducing the fidelity of the final output.

pocmloc,
Part of the concern with positions and postures of listeners in a room was that one person could alter the room's acoustics with the slightest movement and "spoil" the experience for everyone. Remember, the topic of that section was "extreme audiophiles", the kind that would spend thousands of dollars on speaker cable alone.

Ling,
Part of the reason for this approach was to reduce feedback, since it always reduces gain and it can't ever remove all of the distortion no matter how much you increase it. Class D sounds appealing, but even class AB would be better than the very inefficient class A one is normally stuck with if he wants maximum linearity.
Alvin, Jan 10 2012
  

       No, you missed the point I was making - in poor audio systems, information is lost and cannot be recovered, no matter what post-processing you do. This is like Information Theory, as described by Claude Shannon in the 1940's. It's like looking through a window made from rippled glass - the image is all wobbly, and there's no way of reconstructing the original image, no matter how much you know about the properties of the glass, because information from the original image has been lost.
hippo, Jan 10 2012
  

       The whole idea ultimately falls foul of Heisenberg's Uncertainty Principle, since it isn't possible to truly "measure" anything with absolute precision, only to observe (imperfectly) the results of interactions.   

       Maybe you should start by improving your ears ...
8th of 7, Jan 10 2012
  

       In this article, "low fidelity" does not mean severly distorted sound such as fuzz guitars, but rather audio that one might get from the amplifier in a $100 stereo: good, but not great. Audio of that quality could be "repaired" through calculation.

8th of 7
I didn't mean to suggest that the fidelity of these systems would be literally perfect. It should be closer than anything achievable without this technology though.

Corrective feedback from speakers can't reduce doppler effects. It also relies too much on a positional sensor which is not likely to be very linear.
This technology is about predicting how a system will respond to each sample of a signal by calculations based on detailed knowledge of all its components. That's very different from using feedback.
I really hope this idea is baked, but I still haven't seen an example of it anywhere. The links don't lead to anything equivalent to this.
Alvin, Jan 10 2012
  

       Quite a bit of the stuff you're looking for is available in a pro-audio shop, or for that matter online.   

       I'd suggest starting off easy: a de-esser to get rid of some of the noise, a stereo enhancer to do what it says, then a 12 channel stereo equalizer to compensate for the room and the speakers response curve, and BBL's on each channel to dedoppler.   

       Should sound okay.   

       But at each stage you'll be losing some definition.   

       And of course an 8" "subwoofer" isn't a subwoofer... the only way you could even call it a "woofer" is if you consider Chihuahuas to be dogs... which they aren't.
FlyingToaster, Jan 11 2012
  

       [pocmloc] may laugh about the 'sweet spot' being focused on listener's ears, but I attended a lecture from a acoustic engineering PhD student at Queen Mary University of London who had been looking into arrays of speakers adding and cancelling so that the sound different people in different places in a room could listen to different music at the same time.
TomP, Jan 11 2012
  

       I never laughed; my suggestion was a serious contribution to the subject.
pocmloc, Jan 12 2012
  

       I think the extreme(ly deluded) audiophiles would ask you for a method of removing all distortion... But keeping the "warmth", also, could you do all this processing and correction without any of that nasty digital stuff? Very "cold" digital....
bs0u0155, Dec 02 2015
  

       // that nasty digital stuff? Very "cold" digital....//   

       The solution, on which MaxCo. engineers are working around the analog clock, is the Digital Valve Music Engine - essentially an iPod using 1950's computational hardware. Every bit is lovingly generated, and carries with it all the warmth and glow of the cathodes.
MaxwellBuchanan, Dec 02 2015
  

       Seems like feedback is the critical missing component in all digital signal processing. A microphone of exceptional quality for establishing baseband response at different frequencies and volume levels. The computing power required to do this effectively is going to be immense.
WcW, Dec 02 2015
  


 

back: main index

business  computer  culture  fashion  food  halfbakery  home  other  product  public  science  sport  vehicle