• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

What can be fixed with DSP?

Joined
Oct 31, 2020
Messages
27
Likes
16
Something has occured to me. We have smartphones, smart cars, smart fridges and what do I know, perhaps even smart toilets. But why don't we have smart headphones? Couldn't it be possible to have a smart headphone which would eliminate distortion by playing an inverted signal, a principle similar to the one noise cancelling headphones use? Of course, this would create distortion of its own, but of much lower level, which could then be again eliminated by the same method. You could go on and on until the distortion would be inaudible.

Or take CSD waterfall plots for instance. These measure the speed of sound decay. Couldn't the trick described above be used for increasing decay speed? If you are a reproducing a certain frequency, once you stop and the driver is resetting to its stationary position naturally, wouldn't it be possible to speed this up a bit by playing an inverted signal of a specifically calculated amplitude? Like quite literally taking the driver which is slowly finding its way back to its resting position and forcefully snapping it back in there quickly?

The algorithm could even look forward in the music track being played and evaluate to which extent to the incoming signals help reduce the decay time, and add its own little magic whenever necessary.

Am I completely crazy or do such ideas have some merit?
 

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,736
Likes
16,158
Such DSP based driver distortion is actually implemented in some current Samsung soundbars to help their very small drivers, am sure such could be implemented for headphones too, but distortion is not really a big problem on most decent headphones being quite lower compared to usual loudspeakers.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Some resonances inherent in horn designs using abrupt, sharp bends can be eliminated with FIR-based DSP correction -- e.g. Fulcrum Acoustic's coaxial horn designs.


You can't totally eliminate distortion, though.

*But I don't buy Dave's explanation for the built-in "house sound" in the cheaper Sceptre series line of studio monitors he's designed for Presonus -- it could have been EQ'd better out of the box! But what do I know, I'm no speaker designer... so I don't really know all the reasons behind the decisions.
 
Last edited:

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,736
Likes
16,158
You can't totally eliminate distortion, though.
But reduce it, its a similar principle that sensor based drivers use, i.e. compare and correct the actual driver position vs. the optimal one. Of course works only under the modal region of the driver where a driver shouldn't be optimally used anyway.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
But reduce it, its a similar principle that sensor based drivers use, i.e. compare and correct the actual driver position vs. the optimal one. Of course works only under the modal region of the driver where a driver shouldn't be optimally used anyway.

Reduce to low enough levels to be benign for a given range, I totally agree with. My thoughts are there are always absolute limits to what can be "fixed" after the case.
 

Hayabusa

Addicted to Fun and Learning
Joined
Oct 12, 2019
Messages
786
Likes
518
Location
Abu Dhabi
Headphone distortions are already much lower than speaker distortions...
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
But why don't we have smart headphones?
The AKG 700NC has such equalization. Measuring with or without it being on shows the stark difference.

For any equalization to be in a headphone it must be "active." Noise cancelling ones are so EQ can be implemented in them. With regular headphones with no source of power this is obviously not possible.
 

tvrgeek

Major Contributor
Joined
Aug 8, 2020
Messages
1,017
Likes
566
Location
North Carolinia
What you are talking about is basically just negative feedback. Fine, but you need a sensor accurate enough. I see no reason it is not possible in the analog domain, yet the failure of servo subs has shown this to be difficult.

DSP requires sampling, processing, and integrating. These take time. By definition, not real time. More time than you have to make a correction. Now, I suppose it is possible to characterize the driver. Send the signal to a processor, apply a reverse algorithm, integrate it and sent that signal to the cans. So you have a small delay. I suppose it is possible to make a generic processor and each pair of cans comes with downloadable correction algorithm. So, make a proposal on go-fund-me, hire some good engineers and coders and get busy.
 

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,736
Likes
16,158
DSP requires sampling, processing, and integrating. These take time. By definition, not real time. More time than you have to make a correction.
Correct, thats why the Samsung/Harman soundbar solution has no feedback, but send a pre-corrected signal to the driver since they have before measured its non-linearities. Of course this doesn't work for time-variant/drifting distortions.
 

Lunafag

Member
Joined
Apr 14, 2021
Messages
89
Likes
127
I believe the technique is called predistortion. EQ is predistortion for gain. Same for phase, but we don't care about that in audio. There're papers for predistortion for IMD. Haven't seen anything on harmonic distortion but I don't think it's that difficult. It would be really nice if all those corrections can be made using convolution, then Amir could generate complete headphone corrections from his measurements to share with people.
 

Hayabusa

Addicted to Fun and Learning
Joined
Oct 12, 2019
Messages
786
Likes
518
Location
Abu Dhabi
I believe the technique is called predistortion. EQ is predistortion for gain. Same for phase, but we don't care about that in audio. There're papers for predistortion for IMD. Haven't seen anything on harmonic distortion but I don't think it's that difficult. It would be really nice if all those corrections can be made using convolution, then Amir could generate complete headphone corrections from his measurements to share with people.

To make predistortion corrections you need non linear operations. I dont think (quite sure to be honest) it can be done with a convolution/fir filter
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,221
Likes
17,799
Location
Netherlands
To make predistortion corrections you need non linear operations. I dont think (quite sure to be honest) it can be done with a convolution/fir filter

Indeed! But what you can do is extract the linear components, and remove those. That can still lower distortion overall.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,221
Likes
17,799
Location
Netherlands
Some resonances inherent in horn designs using abrupt, sharp bends can be eliminated with FIR-based DSP correction -- e.g. Fulcrum Acoustic's coaxial horn designs.

This is basically only a good idea for correcting issues at the throat of the horn. The further you move to the mouth, the less effective it going to be. This because the corrections are rather directional. Off-axis you might be very much worse off.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
This is basically only a good idea for correcting issues at the throat of the horn. The further you move to the mouth, the less effective it going to be. This because the corrections are rather directional. Off-axis you might be very much worse off.

Indeed, I believe that was the major goal (mostly to eliminate coloration in the same flawed older designs not using DSP) since this kind of thing still required a relatively compact horn to fit in the middle of the midbass driver. It doesn't solve all problems of the older design, but fixes what can be addressed with signal processing.
 
Top Bottom