Kurt Kartaltepe

This looks like an issue in the simde project that we use for arm/non-x86. However we dont support ppc/arm on linux, users of these platforms would need to see if...

And an example bad audio encoder frame generation series is https://gist.github.com/kkartaltepe/d9bc973b03eda915829d4df10f289439 At the first attempted sync (`Pruning! 0`) we have ``` debug: packet: audio 1, ts: 149505163945, pruned = false,...

> Frame duration comes from 1024 samples for each frame; 1024 / 48kHz = 21.3 ms or 1024 / 44.1kHz = 23.2 ms. Did you intentionally put the magic number...

> Instead of having a hard coded constant, cann't we use 1000000 / output->audio->info->samples_per_sec? We could but this is a per encoder thing, hardcoding (hopefully) makes it clear its not...

> Is it really a good idea to hardcode with an AAC value when in the future (while ignoring FTL) using an Opus encoder will be possible for recording, SRT...

> > It possible this is too short for 44.1kHz Ill test and see maybe we need to be more conservative. > > Instead of having a hard coded constant,...

> Would there be any issues with enabling 96 kHz at some future date with the new sync logic? The current version is no longer hardcoded this should be fine...

i cannot replicate this with intel (old/new) on sway. Please provide an obs log if you are affected instead of trying to give a description of your hardware.

Also just a note, for the continual renegotiation that presents as: ``` error: Cannot create EGLImage: EGL failed to allocate resources for the requested operation. info: [pipewire] Renegotiating stream error:...

Rebased on top of the warnings to errors changes.