Murmur is an open voice modality for wearables, phones, headsets, computers, applications, and agents. It deliberately separates three concerns that products usually fuse: the source of speech, the transcription provider, and the software that consumes the text. This keeps voice input portable across devices and applications.
A voice source on one device can feed a transcription provider of your choice, whose output any consuming application or agent can use. Nothing is locked to a single vendor's capture-to-text pipeline, which keeps the voice layer reusable across the tools you already run.
Murmur is in early development. Interfaces and scope are still forming, and the Murmur repository is the place to follow the design and contribute.