Back to catalog / Agent workflows
Build a live spoken-translation room
Stream speaker audio through Realtime WebSockets into a target-language listener app while retaining tone, pace, and emotion.
Goal
Stream speaker audio through Realtime WebSockets into a target-language listener app while retaining tone, pace, and emotion.
Steps
- Start the speaker client, WebSocket server, and listener client. [src-openai-build-live-spoken-translation-room]
- Send raw audio into a Realtime session and broadcast translated audio to the chosen language channel. [src-openai-build-live-spoken-translation-room]
- Test a known phrase for translation, listener delivery, and session shutdown. [src-openai-build-live-spoken-translation-room]
How to verify
Listener audio must convey the known phrase in the target language/channel; confirm disconnect releases the server session.
Troubleshooting
If listeners hear nothing, inspect WebSocket relay events and client connections; if prosody is flat, verify raw audio input rather than transcript-only text.
Permissions
Prerequisites
- OpenAI Realtime API credentials, WebSocket server, speaker/listener clients, test audio, and target-language configuration.
Permissions
- Audio leaves the client and API usage may be billed; the demo is turn-based, so speakers need brief pauses.
Limitations
This is not zero-latency simultaneous interpretation; disclose recording and audio processing first.
Pinned upstream source
- Repository
- openai/openai-cookbook
- File path
- examples/voice_solutions/one_way_translation_using_realtime_api.mdx
- Pinned commit
5bd6cb40b6b84b69b2032d810610aa85064102df- Source checked
- 2026-10-14 · Reviewed against the pinned upstream source
Repository metrics
These values describe the upstream repository, not popularity or likes for an individual skill or playbook.
- Stars
- Not collected
- Forks
- Not collected
- Observed at
- Not collected
No metrics snapshot is available yet.
License: MIT
Main topic
Turn ecommerce campaign materials into a reviewable weekly brief