Models and Languages#

We offer speech recognition models for various languages and domains. However, we offer only high accuracy speech recognition solutions, meaning that you should be able to transcribe any audio in supported languages/domains and obtain a highly accurate transcription. For comparison with other providers, please see Benchmark Data.

All models also support real-time and low-latency speech recognition.

You should use the General domain model unless your use case comes from Medical or IVR domain. At the moment, Medical and IVR models are supported only with the English language.

Language Domain Model name Speaker AI Supported
English General en_precision Yes
  Medical en_precision_medical Yes
IVR en_precision_ivr Yes
Spanish General es_precision No
German General de_precision No

You can configure which model to use by setting the model TranscriptionConfig field to one of the supported values (e.g. en_precision_medical). Refer to Configure Requests.