> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.kotoba.tech/overview/capabilities/speech-to-speech/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.kotoba.tech/_mcp/server.

# Speech to Speech Translation

Kotoba's Speech-to-Speech translation (S2ST) ingests audio in one
language and emits both an incremental transcript of the source and
synthesized speech in the target language — over a single WebSocket
connection. Use it for simultaneous translation, live captioning with
voice-over, and any scenario that needs sub-second turn-around without
splitting the pipeline into separate ASR + MT + TTS steps.

Supported languages: English (`en`), Japanese (`ja`), Korean (`ko`),
Chinese (`zh`), and Spanish (`es`).

## What you get back

* **`partial_transcript`** — incremental source-language transcript.
* **`audio_chunk`** — synthesized target-language audio, streamed as it
  is produced.
* **`done`** — emitted when the server has finished processing the
  committed audio.

## Where to go next

* [Python SDK for S2ST](/s2st/python-sdk) — `client.s2st.translate(...)`
  for a one-shot file, `client.s2st.stream(...)` for live audio.
* [API reference](/s2st/streaming) — the AsyncAPI spec for the S2ST WebSocket
  channel.

> Translate spoken audio into spoken audio in another language.