Audio buffer support in filter graph #389
Conversation
|
This is looking quite lovely! Thank you. This line lies, but that is easy to change or add that feature like the My only lingering feeling is: since the planar audio comes back as 2D, should the non-planar audio do the same (except the axis be reversed)? |
- either layout or channel is required (see C code) - time_base - as Fraction number Ported Ffmpeg audio filtering C example - see https://github.com/FFmpeg/FFmpeg/blob/master/doc/examples/filter_audio.c
|
I've added a ported filter_audio example. The filter chain looks usable now. Please review. |
|
I'm not familiar with the libav terminology but what is "abuffer"? Unless this is the exact name of something in libav, "audio buffer" would be more explicit |
|
@jlaine Unfortunately, it really is called "abuffer". |
|
It looks like appveyor 3.5 build is failing on random dependency download error. Is it possible to trigger rebuild? |
|
Could someone please re-review this one so it could be merged. |
|
I think this is very cool, but as there are zero tests for this new code, I'm not comfortable with merging it. Doubly so as there are a lot of conditional branches.. |
|
Hi Jeremy, the Graph part is currently undertested anyway so I'll just add a few tests for audio filters. |
|
Right the current filter graph code is untested, otherwise the "elif" error would have been caught. We are currently trying to raise the test coverage so any new code must come with tests. |
|
Any updates on this? |
|
Hi @morrolinux, I'm currently lack free time to write proper set of unit tests for this one. I'll fix the conflicts though. |
|
Thanks @egao1980 for trying to do that. My apologies for rarely having free time either; paying the bills has to come first, unfortunately. |
…o-filters-orig # Conflicts: # av/audio/frame.pyx
|
Just bumping this up, hoping it could be merged in. I've been using this branch for a while with no problems. |
|
Please re-review as I've added error codes, fixed example, and most importantly - added a few unit tests covering audio filters. Please note that several Travis configurations fail to initialise even before actually building the project. |
|
I'm interested in working through this, but it really turned into a monolith. Anything that isn't related to the specific purpose of this has to be dealt with elsewhere (e.g. build and test fixes). The error codes stuff sits particularly badly with me, as it stretches so much further than audio buffers and was complete unnecessary AFAICT. If someone wants to champion this, it will need to be based off of the current develop (not with the current develop merged into it!), and be isolated to the audio buffer support. I'll try to take a look at it myself, as I'm on a bit of a PyAV binge, but I make no promises. I honestly don't know the scope of it because the history as-is is way too convoluted. If I do it, I might start from scratch and lose the authorship information. |
|
Closing in favour of #562. |
No description provided.