Whisper API From Google drive file

I am trying to use the Whisper API to transcribe an audio file from Google Drive. I need to send the Whisper API a readable audio file, but I cannot seem to get it from Google Drive at this time.

How can I download the file in the flow and share the file with Whisper? I know that with other tools like Pipedream, I can store the file in a temporary storage and then share the URL with the Whisper API. Is there any way to do something similar using AP?

Hi David,

We added a Read File action to Google Drive.

However, I don’t think there is yet a practical way to pass a file output to another input that is not file.

We currently give files the db:// representation which internally reference the file and can be passed to another file input.

We’re currently discussing how we can make these files more versatile as in:

  • Can be converted to long term shareable links, but this requires us to incorporate file storage in users’ accounts.

  • Or at least allow them to be passable to other services with a temporary link.

What are your thoughts on this?

Although tools like Pipedream offer users temporary storage, it has its limits. I can understand why AP would want to avoid giving every user temporary storage, but it would be nice to be able to set up my own data server to create a temporary link.

I was thinking of passing the data through my public Bunny.net account, publit.io, or any other FTP file storage I have. Either way, I need a way to pass the data from Google Drive to another server.

Will the new Read File action allow me to do this?

The new Google Drive action lets you read the file and pass it to any other piece that accepts files in its inputs. For example, you can read a file from Google Drive and upload it to S3:

The file icon in the input means it accepts files from a previous step.

The change we need to make is to let these files be passable to services that don’t explicitly accept files (i.e in the form of link instead, or as file data in an HTTP request). Both of the options are still not supported and I can’t think of workarounds at this point yet.

Ok, that is a good start :slight_smile: Do you have any limit of file type or file size?

@abuaboud Can you provide information about the limitations on type/size of the files that can be passed through the flow?

@David_M whisper actions are now released in OpenAI.

1 Like

This topic was automatically closed 24 hours after the last reply. New replies are no longer allowed.