Hi everyone!
I’ve created a flow for transcribing files from Telegram.
When I send an audio file recorded on my phone’s voice recorder to the bot, everything works fine — the file is sent as m4a and transcribed successfully.
But when my friend sends a file from his phone, it somehow comes through as mp3, and Whisper OpenAI returns the error:
“Audio file might be corrupted or unsupported.”
Doesn’t Whisper support MP3?
Both phones are iPhones.
I’m using the gpt-4o-transcribe-diarize model.
Here is file data
File Name:
file_18.mp3
File Extension:
mp3
Mime Type:
audio/mpeg
File Size:
53.9 kB
