Documentation / Transcriber
Transcriber
Voice Agents use STT (Speech to Text), to transcribe what the user speaks. This transcribed speech as text goes into an LLM to generate the response that gets played out to the user.
Tigy AI provides managed speech-to-text for every voice agent. The provider and model are selected and maintained by Tigy AI; users do not choose an external STT provider or model.
Warning
Transcriber providers receive the data needed to transcribe speech, such as call audio, language settings, keyterms, and request metadata. Review the provider's data processing, retention, model training, and regional hosting policies before using sensitive data.
Improve domain-specific recognition
The agent's dictionary supplies terms to boost speech recognition (STT) when used by the transcriber. The pipeline separates comma-delimited terms and sends them as keyterms.
Test product names, abbreviations and words relevant to your service. Compare transcripts before and after. This feature concerns recognition; it is not a phonetic dictionary changing TTS voice pronunciation.
For clipped speech or premature replies, review Conversation flow, not only the transcriber's language.