Auto gain control (AGC) effect #14617
Conversation
|
@fonsargo Just from the PR description, to me it looks like the same Replay Gain does. What's the difference between Replay Gain and your goal? |
The difference is very big, because AGC dynamically adjusts the gain of a signal every second or even millisecond, while ReplayGain just calculates the average audio volume and add/subtract the constant amount of gain for the whole track. You can try it using my MR and you will hear the difference (choose some track with big amplitude, the good example is Adele - Skyfall. You can barely hear anything in the beginning, but it's quite loud in the middle). |
|
So, it's similar to an Auto Make-up Gain of a compressor but without the additional compression, right? |
|
No, it's not similar. Auto make up adjusts gain, so the output level is as close as possible to the input level. It's useful when you want to compress audio, but do not want to change the amplitude of the whole track (in my example with Adele - Skyfall the begging will still be quiet and the middle will still be loud after compression). |
|
Just a friendly reminder if someone can take a look on this, please. |
|
Could you please merge the main branch in, to trigger CI. |
There was a problem hiding this comment.
haven't tested this locally yet, but code LGTM. Waiting for @JoergAtGithub
| double calculateBallistics(double paramMs, const mixxx::EngineParameters& engineParameters) { | ||
| return exp(-1000.0 / (paramMs * engineParameters.sampleRate())); |
There was a problem hiding this comment.
If I were pedantic, this should also just take std::chrono::milliseconds and cast to double internally, but that really doesn't matter. In theory you could create your own double-based chrono milliseconds using std::chrono::duration<std::milli, double>, but thats just a neat trick and not necessary.
There was a problem hiding this comment.
The reason why I didn't use milliseconds is that attack and release could be fractional, for example fast attack could be 0.05ms. Also it comes from attack and release knob, which returns double: m_pAttack->value(). I didn't consider creating my own double-based chrono ms, but I'm not sure that will increase readability to be honest :)
There was a problem hiding this comment.
yeah, it probably wouldn't improve readability, just make it less likely to accidentally assign unrelated quantities/units to each other.
|
LGTM! Thank you for this new builtin effect! |
Here is a new AGC (Auto Gain Control) effect for Mixxx. AGC automatically adjusts the gain of an audio signal to maintain a consistent output level.
I couldn't find an existing issue or feature request for this, but I had a prototype from last year when I was working on the compressor effect. So I thought it would be great to turn it into a proper effect as well. It's based on the compressor effect, but with different gain calculations.
Feel free to share any suggestions or ideas for improvement!