Skip to main content

ClovisSTT: from speech to text

Speech-to-text (STT) models convert spoken words into text. Use them to dictate a message, caption a conversation or transcribe a meeting before preparing its minutes.

The ClovisSTT range covers two workflows: receiving a transcript while someone speaks, or processing a completed recording.

Choose your model​

Beta
ClovisSTT/realtime

ClovisSTT/realtime

Transcribe live speech for dictation, captions and voice assistants.

Beta
ClovisSTT/transcription

ClovisSTT/transcription

Turn recordings into text: meetings, interviews, podcasts and voice messages.

What is the difference between Realtime and Transcription?​

With Realtime, your application streams audio and receives text fragments as the speaker talks. This mode suits interfaces that need to display or use the transcript immediately.

With Transcription, your application sends an existing recording and receives a JSON response containing the transcript in the text field. The HTTP route is available in beta. Set response_format="json"; this field must be supplied. See the HTTP parameters and examples.

Both modes produce a transcript. Your application can then send it to ClovisChat or ClovisLLM to summarize it, extract information or prepare a response.