OpenRouter's audio integration lets an application combine transcription, a chat model, and synthesized speech using the same API key. In the October 7 announcement's walkthrough, Scribe v2 converts a recorded question into text, a chat model answers it, and Eleven v4 Turbo turns the answer into speech. The chat model can change without rewriting those audio calls.
The launch also supports transcription with speaker labels and word timestamps, useful for turning a recording into speaker-attributed notes. Its scope excludes Scribe v2 Realtime over WebSocket, so the recorded-audio workflow should not be read as an announcement of continuous streamingSending a model's response token by token as it is generated, so the reader sees text immediately instead of waiting for the whole answer.Full definition → transcription.
Terms in this piece · Glossary
streaming — Sending a model's response token by token as it is generated, so the reader sees text immediately instead of waiting for the whole answer.
