openai-whisper-api

To transcribe m4a audio to text, run the transcribe.sh script with your audio file path. It performs an authenticated multipart upload to the OpenAI /audio/transcriptions endpoint and outputs a text file next to the input.

You can convert m4a and ogg audio files to JSON output using curl. The skill sends a multipart upload to the OpenAI API, letting you specify the whisper-1 model and JSON as the response_format parameter.

Yes, you need an OPENAI_API_KEY to perform speech-to-text conversion. The script uses this key for authenticated requests to the /v1/audio/transcriptions endpoint, and optionally accepts an OPENAI_BASE_URL for proxy support.

Yes, you can specify a language when transcribing audio with the whisper-1 model. The skill supports adding language parameters and prompt hints to improve transcription accuracy for your speech-to-text conversion tasks.

The best way to get structured JSON from speech-to-text transcription is to set the response_format parameter to JSON. This skill performs the authenticated upload via curl and returns structured output directly from the Whisper API.

A multipart upload to the audio transcriptions endpoint sends your audio file along with model, language, and response_format parameters via curl. The Whisper API processes the speech-to-text request and returns the converted text or JSON output.