openAI Whisper Transcript API Call

Hi everyone.

I’m trying to make an API call in openAI Whisper Transcription. Basically, the endpoint is this one: /v1/audio/transcriptions.
My goal is to get a transcription with TimeStamp and I cannot really do that using the available Whisper Transcription module so I want to replicate the API call and simply add necessary parameters.

However, I’m getting this error:
image

The file “value” parameter is directly coming from a download file module and data is basically an IMTBuffer.

Any idea how to fix this? Thank you

Hi @Redante_Techflow.ai
It seems the json is not valid.
you can use online tools to check if the json is valid or not
tool:https://jsonlint.com/

If you require additional assistance, please don’t hesitate to reach out to us.
MSquare Support
Visit us here
Youtube Channel

2 Likes

I’m guessing it has something to do with the {{data}} value.

But the value is directly mapped from a download module and it has IMTBuffer binary data. So it should work fine.

Trying the same thing and came across this post… I think the issue is that this isn’t possible. Putting the encoded binary data into json is not (as I’ve learned from my research) going to work, and the api call option doesn’t offer up a true upload that would transfer with the post request, so I don’t believe this can be accomplished currently.

Unless, of course, you got it to work, so any progress share would be appreciated.

The openAI Make an API Call won’t be able to handle this as the encoded binary data is not possible in json.

Quick workaround was to use an HTTP module and set content type to “form data” to allow mapping of actual file.

1 Like