A question may involve reverse engineering of sounds.

[TuneInSoul]

I am not an expert and don't know how to describe this appropriately.
For example,If I EQ a sound,this can change frequency response and distortion of sounds.
Let's consider this,if I just have two or more sound samples which are results of an EQ effect,we can call them altered samples,and an original sound sample.But I don't know the sound effects the altered samples used.I just know the frequency response and the distortion of the altered samples.And I have some information about altered samples,like some values of the effects used in altered samples.
For example,
Sample 1 : value1=1
Sample 2 : value1=2
Sample 3 : value1=5
The question is:
If there is just one effect used in altered samples and the author of altered samples want to change ony one "dimension" of sound ,but it has caused two or more "dimensions" of sound varied,is there a way to know what the author want to change?

 

[GXAlan]

Long and short answer.

Short answer. Yes.

Longer answer:

If I understand you correctly, what you are looking to explain is the idea of a transfer function. You know what the recording is and you know what the output is from a system, whether it is non-linear electrical or non-linear acoustic/mechanical.

At this point it’s all math and you can create an equation that replicates the transfer function for a given parameter set.

In graphics, you can have matrix LUTs to translate colors. For audio, you have digital pianos, virtual tube amplifiers, and various plugins, etc.

Doing this is sort of what advanced room correction does.
Doing this as a hobbyist is very complex…

 

[Blumlein 88]

If the only change is FR, and you have the unaltered version that is easy to find out. Paul's Deltawave will compare two files and figure that out for you.

Deltawave can subtract one from the other and you can hear the difference from more than just frequency response. Which can be from all sorts of effects. However, undoing distortion, compression and most other DSP is not possible in general in the way one can reverse EQ.

 

[TuneInSoul]

Long and short answer.

Short answer. Yes.

Longer answer:

If I understand you correctly, what you are looking to explain is the idea of a transfer function. You know what the recording is and you know what the output is from a system, whether it is non-linear electrical or non-linear acoustic/mechanical.

At this point it’s all math and you can create an equation that replicates the transfer function for a given parameter set.

In graphics, you can have matrix LUTs to translate colors. For audio, you have digital pianos, virtual tube amplifiers, and various plugins, etc.

Doing this is sort of what advanced room correction does.
Doing this as a hobbyist is very complex…

Appreciate.

 

[TuneInSoul]

If the only change is FR, and you have the unaltered version that is easy to find out. Paul's Deltawave will compare two files and figure that out for you.

Deltawave can subtract one from the other and you can hear the difference from more than just frequency response. Which can be from all sorts of effects. However, undoing distortion, compression and most other DSP is not possible in general in the way one can reverse EQ.

Thanks.

 

[TuneInSoul]

Long and short answer.

Short answer. Yes.

Longer answer:

If I understand you correctly, what you are looking to explain is the idea of a transfer function. You know what the recording is and you know what the output is from a system, whether it is non-linear electrical or non-linear acoustic/mechanical.

At this point it’s all math and you can create an equation that replicates the transfer function for a given parameter set.

In graphics, you can have matrix LUTs to translate colors. For audio, you have digital pianos, virtual tube amplifiers, and various plugins, etc.

Doing this is sort of what advanced room correction does.
Doing this as a hobbyist is very complex…

Does AI can help?
I mean one can build a database to store data of results of all kinds of effects,and use the data to train an AI to recognize the pattern of transfer function.

 

[antcollinet]

Does AI can help?
I mean one can build a database to store data of results of all kinds of effects,and use the data to train an AI to recognize the pattern of transfer function.

It might help people answer you better if you explain what it is you are trying to achieve.

 

[Guermantes]

It's possible to capture an effects chain or an acoustic response as a convolution/impulse response. In fact, this is the current trend in guitar amp and effect rig simulations:

TONEX

AmpliTube TONEX for Mac/PC uses breakthrough AI Machine Modeling™ technology to model the sound of any amp, cabinet, combo and most pedals and turn it into a plug-in.

www.ikmultimedia.com

I suppose what you are asking, though, is if the effects chain could be identified (reverse engineered) from the convolution. Yes, possibly, with a large corpus of IR profiles of known equipment/acoustics.

Edit: added more terminology stuff

 

[TuneInSoul]

It might help people answer you better if you explain what it is you are trying to achieve.

Nothing I am trying to achieve in reality.
I just want to know an anwser.
Sorry,I can't explain it clearly.
Just like a puzzle about sound someone asked me.
The one changed a sound by someway,and ask me "what did you hear?"
And I do know something,the sound changed in many aspects,I want to know what happend,but the one who asked me doesn't give me the answer about what has changed.
So I asked.
In the future I may know I think.
This may be hard to understand,it is really complicated.

 

[TuneInSoul]

It's possible to capture an effects chain or an acoustic response as a convolution/impulse response. In fact, this is the current trend in guitar amp and effect rig simulations:

TONEX

AmpliTube TONEX for Mac/PC uses breakthrough AI Machine Modeling™ technology to model the sound of any amp, cabinet, combo and most pedals and turn it into a plug-in.

www.ikmultimedia.com

I suppose what you are asking, though, is if the effects chain could be identified (reverse engineered) from the convolution. Yes, possibly, with a large corpus of IR profiles of known equipment/acoustics.

Edit: added more terminology stuff

Thank you.