Whisper API: How Can I Transcribe Audio Using File Uploads or a Direct URL?

Below is some information regarding the transcription endpoint and file handling:

Current File Upload Process

Right now, the transcription API requires you to send the file as part of a multipart/form-data POST request. For example:

BASHCopied
curl https://api.openai.com/v1/audio/transcriptions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: multipart/form-data" \
  -F file="@/path/to/file/audio.mp3" \
  -F model="whisper-1"

This means that the API expects the raw file to be uploaded directly rather than providing a URL (such as one from Google Cloud Storage or S3) for retrieval.

Feature Suggestion Feedback

We understand that handling file storage externally can be crucial for high-scale applications. Your suggestion—to allow either a file or a direct URL—matches the challenges many developers face when they wish to avoid the additional overhead of downloading and re-uploading large files.

While this functionality isn’t available at the moment, feedback like yours helps guide future improvements. I recommend sharing your feature request through the official OpenAI feedback channels. This helps the team prioritize new features that can better accommodate workflows involving cloud-based file storage.

Summary

• Current Process: Requires direct file upload via a multipart/form-data request.
• Direct URL Feature: Not available at this time.
• Next Steps: Consider submitting your feedback through the appropriate feedback channels to help guide future API enhancements.

Thank you for your kind words and for sharing your use case. Your input is valuable and is taken into consideration when evolving the API.