Pulse STT

Corrections to the Speech-to-Text response fields so the docs match what the API sends.

from_finalize does not depend on ITN. Every manual {"type":"finalize"} produces exactly one final with from_finalize: true, with itn_normalize on or off. Its transcript is empty if no audio was pending. The field is now on the /waves/v1/stt/live TranscriptionEvent schema and the realtime response format page too.

redacted_entities removed. Neither pre-recorded nor streaming responses include a redacted_entities array. Redacted values appear only as [ENTITYTYPE_N] placeholder tokens in the transcript text.

error_code on live error frames. The /waves/v1/stt/live error event now documents error_code with these values: LANGUAGE_NOT_ENABLED_IN_REGION, LANGUAGE_NOT_SUPPORTED_BY_MODEL, SERVICE_UNAVAILABLE, UPSTREAM_UNAVAILABLE (closes with 1012), and SERVICE_RESTART. Not every error frame carries an error_code. The event also documents error (a human-readable message on some frames, such as validation errors and errors from the model, which may appear alongside error_code), retryable, session_id, and region. It no longer lists a status field, which type: error frames do not carry.

from_finalize vs is_last. A final flushed by finalize is marked from_finalize: true. Only the terminal final after close_stream is marked is_last: true. Neither carries eot_finalized.

Redaction is Pulse only. On POST /waves/v1/stt/, redact_pii and redact_pci apply with ?model=pulse. Pulse Pro ignores them.

→ Redaction