Notely

Audio Recording & Speech-to-Text ​

Notely includes a dual-stream audio recording engine and speech-to-text pipeline for voice notes and conference call transcriptions.

1. Recording Modes ​

Trigger audio recording via Alt + V, the 🎙️ Record Audio toolbar button in the Markdown editor, or the Quick Capture Bar on the landing dashboard.

  1. Meeting Mode (Mic + System Loopback):
    • Captures microphone input and Windows system audio (Zoom, Google Meet, Microsoft Teams, YouTube) concurrently.
    • Merges both streams into a synchronized stereo track via Web Audio API.
  2. Microphone Only:
    • Voice memo and dictation capture from your primary recording device.
  3. System Audio Only:
    • Records desktop/speaker audio output without microphone commentary.

2. Speech-to-Text (STT) Engines ​

Configure STT preferences in AI -> AI Settings -> Speech-to-Text:

  • Local ONNX Whisper (Offline):
    • Runs on-device via WebAssembly/ONNX Runtime.
    • Models: whisper-tiny.en (~40MB), whisper-base.en (~140MB), whisper-small (~460MB).
    • Private, zero external network traffic.
  • Cloud Whisper:
    • High-throughput cloud transcription via Groq (whisper-large-v3) or OpenAI (whisper-1).
    • Sub-second turnaround using your configured API keys.

3. Storage & Markdown Integration ​

  • Asset Bundles: Audio and video recordings are stored inside isolated bundle directories:
    • Audio: {workspace}/media/audio/<recording_id>/<recording_id>.webm
    • Companion Transcripts: {workspace}/media/audio/<recording_id>/transcript.json
  • Editor Markdown Preview:
    • Audio files render interactive player cards with waveform visualizers.
    • Companion transcripts render rich .markdown-transcript-card elements with View, Copy Text/Markdown, and Download actions.
  • Synchronized Transcript Drawer: Clicking any audio or video recording with companion transcript data opens the media player alongside an interactive, timestamp-synchronized side drawer.

4. Diagrams & Media Asset Manager Integration ​

  • Filter by Audio, Videos, and Transcripts categories in Workspace -> Diagrams & Media (Ctrl/Cmd + Alt + M).
  • On-Demand AI Transcript Generation (✨): Run speech-to-text on any existing recording without re-recording.
  • Reference Tracking: View exactly which notes and lines reference the audio/video recordings and transcripts.