refactor: python implementation of audioop.mul

[davidhozic]

Summary

This PR replaces the PCMVolumeTransformer.read() method with a Python equivalent, that does not use the audioop module, which is deprecated since Python 3.11 and will be removed in Python 3.13.

This implements #2177.

Information

• This PR fixes an issue.
• This PR adds something new (e.g. new method or parameters).
• This PR is a breaking change (e.g. methods or parameters removed/renamed).
• This PR is not a code change (e.g. documentation, README, typehinting,
  examples, ...).

Checklist

• I have searched the open pull requests for duplicates.
• If code changes were made then they have been tested.
  • I have updated the documentation to reflect the changes.
• If type: ignore comments were used, a comment is also left explaining why.
• I have updated the changelog to include these changes.

[VincentRPS]

LGTM

[OmLanke]

A python equivalent will not be nearly as close as the C one in terms of performance. Live audio is a heavy load and has to be low latency.

[VincentRPS]

A python equivalent will not be nearly as close as the C one in terms of performance. Live audio is a heavy load and has to be low latency.

I mean we can always rewrite it in Rust.

[VincentRPS]

and also this largely uses python stdlib modules which from the seems of things are written in C in many places.

[OmLanke]

A python equivalent will not be nearly as close as the C one in terms of performance. Live audio is a heavy load and has to be low latency.

I mean we can always rewrite it in Rust.

The whole voice stuff could be rewritten to Rust 😬

[OmLanke]

and also this largely uses python stdlib modules which from the seems of things are written in C in many places.

Possible to do actual performance tests?

[VincentRPS]

A python equivalent will not be nearly as close as the C one in terms of performance. Live audio is a heavy load and has to be low latency.

I mean we can always rewrite it in Rust.

The whole voice stuff could be rewritten to Rust 😬

only if :)

[davidhozic]

audioop:

Without audioop:

Hardware:
OS Name Microsoft Windows 10 Pro
Processor AMD Ryzen 5 5600X 6-Core Processor, 3701 Mhz, 6 Core(s), 12 Logical Processor(s)

[OmLanke]

Hmm that sure is a difference.

Use time.perf_counter_ns() instead of time.time(). The result will be more accurate and will return ns instead of s

[OmLanke]

Can we maintain and ship our own audioop C file? Dpy plans to do something like that

Or we could try Cython or maybe Rust?

[davidhozic]

R5 5600x 4.7 GHz

Elapsed: 31819500 ns
Elapsed: 1016100 ns
Elapsed: 868800 ns
Elapsed: 739800 ns
Elapsed: 897400 ns
Elapsed: 743300 ns
Elapsed: 732200 ns
Elapsed: 784500 ns
Elapsed: 840000 ns

[OmLanke]

Wow that is some real improvement 👍

LGTM

X