webrtc icon indicating copy to clipboard operation
webrtc copied to clipboard

oggwriter output fails validation with oggz-validate, rejected by AWS Transcribe

Open chrbsg opened this issue 5 months ago • 0 comments

Your environment.

  • Version: Pion v3.2.24

What did you do?

Use the oggwriter to create an ogg file. The file sounds fine when played with mplayer, but AWS Transcribe rejects it with BadRequestException: The data is corrupted.. The same PCM data encoded with opusenc is accepted.

Investigating this, I tried a validation tool on the Oggwriter output file:

$ oggz-validate pion_oggwriter.ogg 
pion_oggwriter.ogg: Error:
serialno 18446744071875297659: missing *** eos

I do not know if this is the reason Transcribe rejects the Ogg file, but it is possible that they validate it on the server side with some similar logic.

What did you expect?

Generated Ogg file should pass oggz-validate validation and be accepted by Amazon Transcribe.

What happened?

It fails validation.

Test ogg files

files.zip This zip file contains 1) opusenc.ogg - a valid file created by opusenc 2) pion_oggwriter.ogg - a rejected file created by pion oggwriter.

chrbsg avatar May 07 '25 13:05 chrbsg