Soniox releases STT v6 speech model with agent text input and echo cancellation

Soniox releases STT v6, a speech-to-text model for voice agents, dictation, and meeting tools, according to a post by Santiago (@svpino) that quotes Soniox's announcement.

Developers can send the agent's text along with the audio, which Santiago says helps the model recognize user replies better and filter out the agent's own playback, reducing false interruptions.

The announcement says STT v6 adds higher accuracy across 60+ languages and is available now for real-time streaming and asynchronous transcription. The accuracy claim comes from Soniox's announcement as quoted in the post; no independent benchmark is reported.