openai-whisper-api

You can transcribe audio to text by sending an audio file to the OpenAI Whisper API endpoint. The process targets /v1/audio/transcriptions and returns a plain text transcript by default, requiring an OPENAI_API_KEY.

OpenAI speech-to-text transcription supports model selection, language hints, and optional prompts. It produces a plain text transcript by default and can emit JSON output for downstream processing.

You can use a proxy for audio transcription by setting the OPENAI_BASE_URL environment variable. This allows the transcription process to target OpenAI-compatible proxies instead of the default endpoint.

Yes, an OPENAI_API_KEY is required to convert recorded meetings into searchable written content. This key authenticates your requests to the Whisper speech-to-text API for generating accurate transcripts.

The best way to generate transcripts from voice notes is using the Whisper API with JSON output formatting. This produces structured text transcripts from your audio files suitable for direct downstream processing.

Audio transcription output formatting depends on your selected output controls. By default it produces a .txt transcript, but you can configure the Whisper API to emit JSON if structured formatting is required.