IOS recordings in Whisper only gather the first few seconds

API
, ,
1

I am sending audio recordings to the OpenAI Whisper API and cannot get mobile recordings to accept past a few seconds of data, I have no idea why. Desktop audio recordings function perfectly fine but whenever I try on my phone the transcriptions only get a word or two

curl --request POST
–api.openai.com/v1/audio/transcriptions
–header ‘Authorization: Bearer TOKEN’
–header ‘Content-Type: multipart/form-data’
–form ‘file=@C:\Users\katra\Desktop\audio.wav’
–form model=whisper-1

If anyone can tell me what to do that would be great have been stumped and cannot find anyone with this issue at all

2

You have not shown your code recording your audio data but I suspect that it has MediaRecorder.start(1000) somewhere and this triggers it to record 1 sec of audio and the app sends the chunk to the backend where it is immediately sent to the Whisper API hence you only get a word of two. As to why it does not happen in the Desktop, I suspect the time slice is not supported. Again, these are all guesses without seeing your actual code.

3

It doesn’t have that sadly, I tested using postman and uploaded Audio files here

4

It doesn’t have that sadly, I tested using postman and uploaded Audio files here