Speech to text for India
Scriba turns speech into accurate transcripts in 22 Indian languages and English — native script, code-mixed or verbatim — with speaker labels, timestamps and separate call channels. One API, from a short voice note to a two-hour recording.
QuickstartYour first transcript in five minutes.
Long recordingsUpload, transcribe, get a webhook.
API referenceEvery endpoint and field.
What you can build
- Meeting notes — speaker-labelled transcripts of calls and meetings in the language people actually spoke. See the meeting notes recipe.
- Call-centre analytics — stereo calls transcribed one channel at a time, so agent and customer never blur. See call centre.
- Subtitles and records — timed segments for captions, hearings and lectures. See subtitles.
- Voice features in apps — short audio in one request with the synchronous endpoint.
How it works
- Create a project in the console. Each project lives in one region, where its uploads and transcripts are stored.
- Create an API key for the project. Requests go to the project's own address.
- Send audio. Short clips in one request; anything longer as an upload followed by a transcription.
- Get the transcript by polling or with a signed webhook.
Languages
All 22 scheduled languages of India plus English, each in three output modes. The full list, with codes, is on Languages.