All the configuration properties described in the feature pages are defined in the POST /v2/live endpoint.
Endpointing
Control how long to wait for silence before closing an utterance.
Partial transcripts
Stream incremental hypotheses while audio is being processed.
Multiple channels
Handle up to 8 channels with channel-aware transcription. Set
channels in the init request.Custom vocabulary
Boost recognition accuracy for brand, product, and domain terms by matching how they sound (phoneme-based).
Custom spelling
Replace specific text variants with your preferred spelling.
Custom metadata
Attach metadata to sessions for organization and auditability.