The SDK shipped steady API surface updates tracked through automated Fern regenerations, with targeted feature work on multimodal and music capabilities. A multimodal_message WebSocket event landed to support real-time multimodal interactions, while music endpoint parsing was refined to handle CRLF line endings correctly. Speech-to-text wrapper signatures received overload support to improve developer ergonomics, and the Scribe WebSocket URI now respects audio_format parameters when specified.
Multimodal WebSocket support entered the SDK this month. The team added a multimodal_message event for real-time bidirectional communication, while also fixing music endpoint parsing to handle CRLF line endings correctly and regenerating client bindings to track API changes.