Back to search
DayBee
MCP serverGitHub
Desktop microphone recorder with real-time, on-device transcription, automatic audio segmentation, voice activity detection, and an MCP server for searching saved transcripts.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Checked today
Badge for your README
Markdown
[](/listings/nexmoe--daybee)HTML
<a href="/listings/nexmoe--daybee"><img src="/api/listings/nexmoe--daybee/badge.svg" alt="findsafeskills safety rating" /></a>Install
i
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 71 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- # DayBee
- English | [中文](README.zh.md)
- DayBee is a tray-first desktop recorder built with Go and mygo. It records microphone audio for long sessions, segments files automatically, and uses Apple SpeechAnalyzer or selectable `sherpa-onnx` ASR models plus Silero VAD for live transcription and offline reprocessing.
- ## Features
- - Long-running microphone recording with automatic segmentation
- - Live transcription with Apple SpeechAnalyzer, lightweight Whisper Tiny/Base/Small, SenseVoice Small, Qwen3 ASR 0.6B or FireRedASR2 AED/CTC (can be disabled)
- - VAD-only recording — only speech segments are saved to disk
- - Real-time waveform visualization
- - Manual microphone selection with automatic switch to the more active input
- - Native echo cancellation on macOS to reduce speaker playback in recordings and transcripts
- - Optional macOS system audio capture, saved and displayed separately from microphone speech
- - Automatic recording on launch and lossless FLAC archives
- - Batch transcription of existing WAV/FLAC recordings
- - Timeline of complete saved transcripts
- - Built-in MCP server for listing, searching and reading saved transcripts from AI assistants; copy the client configuration from General settings. See [MCP setup and tools](docs/subsystem/mcp.md).
- - Annotate transcripts with app names and icons from the same recording period
- - System tray app with launch-at-login, start-on-launch, and auto-update
- - Bilingual UI (English / 中文)
- - Light / Dark / System theme
- ## Desktop Preview
- 
- ## Download
- - Release page: [DayBee Releases](https://github.com/nexmoe/DayBee/releases)
- - macOS 13.3+: Apple Silicon / Intel `.dmg`
- - Windows: `.exe` (NSIS installer)
- The packaged app checks for signed updates on startup and every six hours, then installs and restarts when recording and transcription are idle. The first upgrade from Electron requires a manual installation; see the [migration verification and upgrade notes](docs/migration-verification.md). Releases require automated checks on all three platforms; outstanding manual migration acceptance remains recorded in the verification report.
- ## Runtime Notes
- - Compatible Macs running macOS 26 or later default to Apple SpeechAnalyzer for on-device transcription. Older Macs and other platforms default to Whisper Tiny (INT8). Existing model choices are preserved; all models remain selectable in Transcribe settings. SpeechAnalyzer resources are managed by macOS, and its System language option follows your Mac's preferred languages; select a language for other audio. Models can be switched while recording, and queued speech retains its original model and language. See [speech recognition](docs/subsystem/asr.md).
- - Silero VAD downloads when recording starts. The selected ASR model downloads when chosen or when live/batch transcription first needs it. Existing model caches are reused; failed downloads preserve the previous selection.
- - Native audio capture, sherpa-onnx 1.13.8 and ONNX Runtime 1.28.2 are bundled. Installed apps need no Node.js, Bun or Go runtime.
- - WAV/FLAC encoding, decoding, downmixing and resampling run directly in Go. Existing FLAC recordings remain readable; no FFmpeg executable or system installation is required.
- - On macOS, echo cancellation is enabled by default, including for existing settings. Turn it off in Input for unprocessed capture. Auto-switch input remains independently configurable and keeps echo cancellation enabled after switching; disconnect recovery remains available. See [echo cancellation](docs/subsystem/audio.md). System audio recording is opt-in on the same page; see [source separation](docs/subsystem/audio.md).
- - Microphone app detection runs independently while DayBee is open, checks once per second, and requires macOS 14.2+ on Mac. Daily archives are stored under `microphone-activity/` in the app data directory. See [microphone activity notes](docs/subsystem/microphone-activity.md).
- ## Output
- - Audio and transcript JSON files are archived under `recordings/YYYYMMDD/`
- - Filenames look like `daybee_YYYYMMDD_HHMMSS.wav` or `.flac`
- - With ASR enabled, each segment writes a matching `.json` with transcript text and metadata. ASR-disabled recording saves audio and capture timing metadata without transcript text.
- Example JSON:
- ```json
- {
- "audio_file": "daybee_20260201_120513.wav",
- "audio_path": "/path/to/recordings/20260201/daybee_20260201_120513.wav",
- "backend": "sherpa-onnx",
- "created_at": "2026-02-01T12:05:13.000Z",
- "input_device": "MacBook Air Microphone",
- "model": "Qwen3 ASR 0.6B",
- "segment_start_time": "2026-02-01T12:05:13.000Z",
- "status": "ok",
- "text": "Please stand up and stretch for a minute."
- }
- ```
- ## More Documentation
- - [Subsystem reference](docs/subsystem/README.md)
- - [Changelog](CHANGELOG.md)
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.