"discord.ext.voice_recv.opus: x packets were lost being flushed in decoder" causing wav file to be output incorrectly

[CattoYT]

When packets are flushed, the output wav file seems to speed up and skip audio sections
Example files: https://drive.google.com/drive/folders/13TclkKZRA6_9nm_jJvg3-T3WSgENQeKJ?usp=drive_link

This also seems to happen with WaveSinks, so I'm not quite sure that its my code acting up.

What I'm currently doing is writing all of the received VoiceData to a bytes() and then using Wave_write.writeframes to create a wav file with the bytes.

Is this intended behavior, and if so, is there a way around it?

[imayhaveborkedit]

I still haven't fully hammered out how the packet flushing/teardown semantics are going to be. Since there's no equivalent function on the dpy voice send side of things I have to figure out if/how I want to expose it. There's a bit more design and testing to do on my end but at this point no, problems are not "intended", but for this issue in particular I don't think I have anything for you just yet.

[CattoYT]

no problem, I understand that this is probably a lot more complicated than what appears, so please take your time!

[realcoloride]

Hey there.. any update?

[imayhaveborkedit]

I've figured out what's causing this. The sequence rollover handling logic is a bit buggy causing it to erroneously drop packets. This will be fixed in due time.

[realcoloride]

About time haha! Good luck

[CattoYT]

Apologies for the necropost, but are there any updates on this yet?

[imayhaveborkedit]

Unfortunately not yet. I've been fairly busy lately and in the middle of working through the issues in this I ended up putting this aside for some time. At the moment im still a bit busy but I'll try to see if I can sort this issue out quickly or if I need to put more work into it.