Naar navigatie springen

WebSocket

The Realtime Translation WebSocket API streams translation audio continuously to a target language given source audio. Currently for enterprise workspaces only.

Handshake

WSS
/v1/translation/realtime

Headers

xi-api-keystringOptioneel

Queryparameters

authorizationstringOptioneel
Your authorization bearer token.
target_languagestringVerplicht

Target language as an ISO 639 code (e.g. de, es, fr, zh).

source_languagestringOptioneel

Source language as an ISO 639 code. When omitted the server auto-detects the spoken language.

voice_idstringVerplicht
Voice ID used for the translated speech output.
input_formatanyOptioneel
Audio encoding format for realtime translation input.
output_formatanyOptioneel
Audio encoding format for realtime translation output.
tts_model_idenumOptioneelStandaard eleven_v4_turbo
TTS model used to synthesize the translated speech.
Toegestane waarden:
apply_text_normalizationenumOptioneelStandaard off

This parameter controls text normalization with three modes: ‘auto’, ‘on’, and ‘off’. When set to ‘auto’, the system will automatically decide whether to apply text normalization (e.g., spelling out numbers). With ‘on’, text normalization will always be applied, while with ‘off’, it will be skipped.

Toegestane waarden:
eagernessenumOptioneelStandaard medium
How eagerly translation starts while the speaker is still talking. 'high' starts translating soonest for the lowest latency. 'low' waits for more of the sentence, giving higher quality translations with more delay.
Toegestane waarden:
client_session_idstringOptioneel

Optional client-defined session identifier for tracking purposes.

enable_zrmbooleanOptioneelStandaard false

When enable_zrm is set to true, zero retention mode will be used for the request. Sensitive content (transcripts, translations, and TTS text) is not written to application logs or session history. Zero retention mode may only be used by enterprise customers.

Versturen

inputAudioChunkobjectVerplicht
Send a chunk of audio for translation.
OR
endOfStreamobjectVerplicht
Signal that the audio stream has ended.

Ontvangen

sessionStartedobjectVerplicht
Sent by the server when the session is initialized.
OR
statusobjectVerplicht
Session lifecycle update indicating the session has started or stopped.
OR
partialTranscriptobjectVerplicht
Interim transcription of the source audio.
OR
finalTranscriptobjectVerplicht
Final transcription of a segment of source audio.
OR
translationobjectVerplicht
Translated text in the target language.
OR
audioobjectVerplicht

Base64-encoded audio of the translated speech.

OR
errorobjectVerplicht
Error encountered during translation.
OR
Translate Auth ErrorobjectVerplicht
Authentication failed for the session.
OR
Translate Rate Limited ErrorobjectVerplicht
The request rate limit was reached.
OR
Translate Queue Overflow ErrorobjectVerplicht
Audio was sent too frequently and the session was terminated.
OR
Translate Input ErrorobjectVerplicht

An input message could not be handled; check the message format.

OR
Translate Transcriber ErrorobjectVerplicht
The transcriber failed to process the audio.
OR
Translate Translation ErrorobjectVerplicht

Translation failed for a segment; the session continues.

OR
Translate Audio Output ErrorobjectVerplicht

Audio output for a segment failed or was dropped; the session continues.

OR
Translate Quota Exceeded ErrorobjectVerplicht

The workspace is out of credits; the session is closed.

OR
Translate Payment Issue ErrorobjectVerplicht

The workspace has a payment issue or is suspended; the session is closed.

OR
Translate Unaccepted Terms ErrorobjectVerplicht

The workspace has not accepted terms required for this feature; the session is closed.