augain: add audio volume filter module #3166
Conversation
modules/augain/augain.c
Outdated
| if (!st || !af || !af->sampv || !af->sampc) | ||
| return EINVAL; | ||
|
|
||
| if (af->fmt != AUFMT_S16LE) { |
There was a problem hiding this comment.
This check should be moved to encode_update (prm->fmt), not needed to execute for every frame.
There was a problem hiding this comment.
I moved af->fmt check but was not sure if also if (!st || !af || !af->sampv || !af->sampc) could have been moved.
modules/augain/augain.c
Outdated
| gained_sample = (int16_t)(sample * encode_gain); | ||
| /* | ||
| info("augain: sample=%d/gained=%d\n", sample, gained_sample); | ||
| */ |
|
Does it make any sense to have call in |
|
|
||
| highest_possible_gain = 32767.0 / highest_abs_sample; | ||
| if (encode_gain > highest_possible_gain) | ||
| encode_gain = highest_possible_gain; |
There was a problem hiding this comment.
It looks like the real gain is different for each frame, depending on the frame energy.
Does this algorithm work, is there any audible noise or artefacts ?
There was a problem hiding this comment.
Yes. I thought that it is better than cutting those sample values that that would not fit 32768 - 32767 range.
I have made tests and haven't noticed anything very bad. It is trade of between not hearing anything and hearing, but with some artifacts.
There was a problem hiding this comment.
I made several tests using both adaptive and fixed algorithm and noticed less artifacts with the adaptive one. Thus I don't think that it is worth to add fixed algorithm to this PR. It can be added later if someone needs it.
From my point of view, this PR is done.
|
Here is another implementation, for inspiration: https://github.com/FFmpeg/FFmpeg/blob/master/libavfilter/af_volume.c |
|
@alfredh If I interpreted the quoted ffmpeg code correctly, it is after applying the gain cutting the resulting value to -32768 - 32767 range, whereas the code in this PR lowers the gain so that the the resulting value stays within the range. To me the ffmpeg version (if I understood it correctly), could result to lots of min or max sample values if gain is large. |
|
One possibility might be to introduce two algorithms, say |
|
thanks, can you please also update this page with the new module and your name ? https://github.com/baresip/baresip/wiki/Modules |
|
Thanks for the merge. Added the module to Wiki. |
This module can be used to increase volume of audio source, for example, microphone.