Speech 3.6 and Transcribe 2 now live. Hear the speech model →

Frontier research

#1-ranked realtime interaction models. Built on frontier AI research.

One API for speech generation, transcription, and production voice agents — fast enough for live conversation.

Play and Trust both go to team@sonacast.click. No public playground on this static site.

Live conversation · sub-90ms
Speech out44 languages
Transcribe inlive STT
Agentsproduction
Usage$0.06 / min
#1naturalness ranking
<90msspeech latency
44native languages
$0.014/min with our number

One API

Speech out. Transcribe in. Agents in production.

01 / Speech

Generation

The fastest, most natural TTS we ship. Sub-90ms. See the Speech page.

02 / Transcribe

Live conversation in

Speech-to-text on the same API as generation. Built for turn-taking, not batch files.

03 / Agents

Production voice

Prepaid agent minutes on every plan. UI-created agents are free for a limited time.

Voice agents for every industry

Finance. Healthcare. Government.

Illustrated floor-plan scenes from the source — realtime rooms, not a call-center overlay.

Fraud detectionReal-time outbound verification on suspicious transactions.
Customer supportListen-in rooms that stay on the same voice API.
CollectionsOutbound lines with prepaid agent minutes.
Loan assistanceGuided conversations, not a brittle IVR tree.
WealthLow-latency speech for desk and client lines.
ComplianceTrust notes on request. Same inbox.

Pricing for every stage.

Free $0 · Pro $5 · Startup $49 · Scale $299 · Enterprise custom.