Voice-to-Memory
Voice-to-Memory lets you speak a thought and turn it into searchable knowledge inside MindStore.
Voice-to-Memory lets you speak a thought and turn it into searchable knowledge inside MindStore.
What It Does
- records audio in the browser
- transcribes it with OpenAI Whisper or Gemini
- stores the transcript as a voice recording
- lets you save that recording into your main memory base
How To Use It
- Open
/app/voice. - Start recording.
- Stop when you are done speaking.
- Review the transcript and edit the title if needed.
- Save it to your knowledge base.
Requirements
- a browser with microphone access
- an OpenAI or Gemini API key configured in MindStore settings
What Gets Stored
- recording title
- transcript text
- provider and model metadata
- duration, language, and word count
- whether the transcript has been saved as a memory
Why This Port Matters
Voice-to-Memory is the second major frain feature ported into the codex architecture.
It is the reference example for media-aware AI plugins because it combines:
- browser capture
- transcription provider selection
- reusable server-side business logic
- save-to-memory integration
- codex docs and tests