feat: add audio peak limiting with configurable settings #751
Conversation
src/source/amplify.rs
Outdated
| } | ||
|
|
||
| /// Internal function that converts decibels to linear | ||
| pub(super) fn to_linear(decibels: f32) -> f32 { |
There was a problem hiding this comment.
Should we move these methods to mod.rs or some utility module?
There was a problem hiding this comment.
We have src/math it could fit there. Feel free to move it and rename it to decibels_to_linear(). But its also fine where it is right? this is always amplifying, sometimes a number sometimes a source.
There was a problem hiding this comment.
I moved and renamed them to src/math.rs in b1497f8. It doesn't feel right to me that we were adding a utility function to_db to a module that didn't use it. I also made them public, because they are useful as general-purpose audio functions.
There was a problem hiding this comment.
they are useful as general-purpose audio functions
I don't follow exactly. We have different amplify methods on source for the different scales. How does converting a single float to another float help audio? Do you mean they are useful as parts for custom sources?
There was a problem hiding this comment.
As a user of this library, I frequently run into the need to convert one into the other and display it to the end user. For example: Deezer, Qobuz and Spotify all send linear values and I want to print their dB equivalents on screen. Easy peasy when Rodio has them public.
|
I love the settings example showing some analysis! It shows a different application of rodio then playback with effects. Could we augment that example with 2 of the examples from #750? I would say the "compressor_wav" and "compressor_wav_alternate.rs". We could rename those to "compressor_basic" and "compressor_changing". |
There was a problem hiding this comment.
The documentation is spot on, its a nice balance between explaining audio and being to the point.
I fear we need to talk about our nemesis spans again. Specifically what happens when the channel count changes. Do we need to switch from limit::Mono to Limit::Stereo then?
The rest is minor stuff: language preference (you can ignore if you disagree), a test that's not needed, comments that are superfluous etc.
|
Thank you for your review.
I did not add anything like |
Provides a new LimitSettings API and limiter filter to prevent audio peaks from exceeding a threshold. Supports soft-knee limiting, per-channel detection, configurable attack/release, and efficient processing for mono, stereo, and multi-channel audio. Includes comprehensive tests and usage examples.
…dule - Add LimitSettings presets for common use cases (dynamic_content, broadcast, mastering, gaming, live_performance) - Expand and clarify LimitSettings documentation and dBFS reference - Move dB/linear conversion functions and tests from amplify.rs to new math.rs module - Refactor limiter implementation for clearer internal structure and documentation - Add new example (limit_wav) and integration tests for limiter behavior - Clean up and modernize limit_settings.rs example - Update module imports to use new math API
roderickvd
force-pushed
the
feat/feedforward-limiter
branch
from
e47684a to
e61dce9
Compare
Then lets not do that 👍 |
|
ready to merge now right? |
|
👍 I will squash merge now. |
|
I may have some confusion regarding terminology, how such limiter is different to a compressor? Compressor also has attack/release time parameters, with perhaps only difference that below-the-knee-threshold samples are also affected. |
|
A limiter is also a compressor, a particular type of it. Any compressor has a threshold (amplitude) beyond which it starts reducing dynamic range with a certain ratio. For a limiter, that ratio is very high or infinite. |
Provides a new LimitSettings API and limiter filter to prevent audio peaks from exceeding a threshold. Supports soft-knee limiting, per-channel detection, configurable attack/release, and efficient processing for mono, stereo, and multi-channel audio. Includes comprehensive tests and usage examples.