Speech to text

Turn speech into
accurate, structured text.

Live microphone streaming and long-form file transcription with speaker labels, multilingual output modes, and export-ready captions—built for real conversations, accents, and noisy environments.

Live WebSocket streaming File upload up to hours of audio Speaker diarization TXT · SRT · VTT · JSON export
TypeScript
Ready
const job = await ebma.stt.transcribeFile({
  file: audioFile,
  language: "auto",
  diarize: true,
  speakers: 8,
});

const result = await ebma.stt.wait(job.id);
console.log(result.transcript);
Interactive preview

Test it before you sign in

Try a live browser preview on this page. Full production quality continues in the portal.

Try STT

Samples are free · Live mic uses EBMA ASR (5 free tries)

Two-speaker customer support snippet with diarized labels.

Press Run sample to stream a canned transcript.

Want diarization, SRT/VTT, and production ASR?

Capabilities

Built for real products

Click a capability to explore what it unlocks.

Spotlight

Live transcription

Stream microphone audio over WebSocket and watch phrases finalize as speakers pause—ideal for agents, captions, and assistive UX.

Input

Live mic PCM · file audio/video

Languages

Auto-detect + Indic & English codes

Modes

Native · Mixed · Romanized

Diarization

Optional speakers (auto or fixed count)

Use cases

Where teams use it

Contact centers

Transcribe calls, attribute speakers, and feed QA or CRM workflows with structured transcripts.

  • Start from a sample workflow in the portal
  • Connect STT / TTS / LLM as needed
  • Export or automate from your account

Spec snapshot

InputLive mic PCM · file audio/video
LanguagesAuto-detect + Indic & English codes
ModesNative · Mixed · Romanized
FAQ

Common questions

Yes. Use live microphone streaming for real-time phrases, or upload audio/video files for batch transcription with progress polling.

File transcription supports optional speaker identification. Choose automatic detection or set an expected speaker count when you know the roster.

Completed jobs support TXT, SRT, and VTT via the API, plus a client JSON export in the playground. Live sessions can export the same formats from finalized phrases.

Choose native script, mixed, or romanized so the transcript matches how your product displays Indic and multilingual text.

Ready to try speech to text?

Open the portal playground or create a free account to start building.