Rime launches Prism, a speech-to-speech voice model that writes replies as text first

Rime has launched Prism, a speech-to-speech model for AI voice agents.

Unlike other speech-to-speech models, Prism writes each reply as text before speaking it, which lets companies edit the transcript and block content the agent should not say. Rime says no other speech-to-speech model works this way, and presents Prism as a safer fit for banks and other regulated businesses.

Rime's launch post offers the first 100 people who sign up a free month. The report does not describe pricing beyond that offer.