torchaudio.info.num_frames can give wrong results so it can provide false exceptions

[grazder]

Tested versions

current master

System information

linux

Issue description

I get here wrong num_frames on m4a opus-files

pyannote-audio/pyannote/audio/core/io.py

Line 280 in cd3f550

frames = info.num_frames

Repro and described error can be founded here: pytorch/audio#3731

It's better not to trust this method. For example, I've got minumum dataset item length of ~ 121s, but the minimum torchaudio_info(...).num_frames is 5674 which is obviously fake.

Minimal reproduction example (MRE)

pytorch/audio#3731

[grazder]

pytorch/audio#3573

[hbredin]

Thanks for the report.
What solution do you suggest?