AbhiMeet - Gravador de Reuniões com IA Local
abhishekjoshi-eng/AbhiMeet
Grave reuniões localmente, transcreva com Whisper, relatorios com Claude MCP
Ver no GitHubGravador de reuniões open-source que captura áudio e tela, transcreve localmente com Whisper AI em 90+ idiomas e gera relatórios via Claude MCP. Totalmente gratuito, sem nuvem, sem inscrição — alternativa local ao Fireflies.ai.
Instalação
npx convez-ai abhishekjoshi-eng/AbhiMeet1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Gravar e transcrever reuniões de trabalho sem depender de serviços cloud
- Gerar automaticamente atas, action items e sumários via Claude
- Analisar conteúdo visual de reuniões com extração de frames de vídeo
- Documentar reuniões em idiomas não-ingleses (90+ suportados)
Veredito

Ferramenta sólida para equipes que valorizam privacidade e soberania de dados. A integração com Claude MCP é robusta (15 tools), mas requer Node.js + Python local; ideal para desenvolvedores e não-técnicos que aceitam setup inicial. Preenche lacuna real vs. Fireflies.ai.
README
A tradução para o português deste README ainda está na fila.
Why AbhiMeet?
| Feature | Fireflies.ai | AbhiMeet |
|---|---|---|
| Monthly cost | $19-39/user | Free forever |
| Data storage | Their cloud | Your local disk |
| Privacy | Uploads to cloud | 100% offline |
| Languages | Limited | 90+ via Whisper |
| AI Reports | Built-in (limited) | Claude AI (unlimited) |
| Screen recording | No | Yes |
| Custom reports | No | Yes — MOM, action items, decisions, anything |
| Open source | No | Yes — MIT License |
Demo
You: "Transcribe my latest recording"
Claude: Transcribing Budget-Review_160426_0230PM... Detected Hindi (92% confidence)
Transcribed 47 segments in 12 seconds. Saved.
You: "Generate meeting minutes with action items"
Claude: ## Meeting Minutes — Budget Review (16 Apr 2026)
### Action Items:
1. Abhishek to finalize vendor quotes by Friday
2. Send revised budget to accounts team
...
Features
- One-Click Recording — Single button to record mic + system audio + screen simultaneously
- 3 Output Files Per Recording:
Subject_Audio_DDMMYY_HHMMam.mp3— Audio only (for Claude transcription)Subject_Video_DDMMYY_HHMMam.webm— Screen only (for visual analysis)Subject_AV_DDMMYY_HHMMam.mp4— Combined audio + video (for sharing)
- Live Audio Waveform — Real-time visualization with frequency bars so you know recording is working
- Recording Feedback — Animated indicators: "Audio capturing", "Screen capturing", "Saving to disk"
- Built-in Media Player — Audio, Video, and Combined tabs with native playback
- Whisper AI Transcription — Local speech-to-text (no API key, no internet required)
- Video Frame Extraction — Claude can "see" what was on screen during the meeting
- 15 MCP Tools — Full Claude integration for transcription, summaries, reports
- Multi-Language — English, Hindi, Gujarati, Marathi, Bengali, Kutchi + 90 more
- Configurable Storage — Browse and set your own recording folder
- Dark Theme UI — Clean, professional, distraction-free interface
- Zero Subscription — No cloud, no account, no recurring payments
Installation
Prerequisites
Quick Setup
# 1. Clone
git clone https://github.com/abhishekjoshi-eng/AbhiMeet.git
cd AbhiMeet
# 2. Install the Electron app
cd app
npm install
cd ..
# 3. Install the MCP server + Whisper
cd mcp-server
uv sync
cd ..
# 4. Download FFmpeg (Windows)
# Download from https://www.gyan.dev/ffmpeg/builds/
# Place ffmpeg.exe and ffprobe.exe in app/ffmpeg/
# 5. Run!
cd app
npx electron .
Connect to Claude Desktop (Cowork)
Add this to %APPDATA%/Claude/claude_desktop_config.json:
{
"mcpServers": {
"abhimeet": {
"command": "uv",
"args": ["--directory", "C:/path/to/AbhiMeet/mcp-server", "run", "main.py"]
}
}
}
Connect to Claude Code
Add this to .claude/.mcp.json:
{
"mcpServers": {
"abhimeet": {
"command": "uv",
"args": ["--directory", "/path/to/AbhiMeet/mcp-server", "run", "main.py"]
}
}
}
Restart Claude. You'll see 15 AbhiMeet tools available.
MCP Tools (15)
| Tool | What It Does |
|---|---|
list_recordings | List all meeting recordings with metadata |
get_recording_info | Get full details about a specific recording |
get_audio_file_path | Get path to the audio MP3 for analysis |
get_video_file_path | Get path to the screen recording WebM |
get_combined_file_path | Get path to the combined A+V MP4 |
transcribe_recording | Transcribe audio using Whisper AI (runs locally, no API key) |
extract_video_frames | Extract screen frames so Claude can see what was shown |
save_transcription | Save transcription text to the recording folder |
read_transcription | Read back a saved transcription |
save_summary | Save a meeting summary |
read_summary | Read a saved summary |
save_report | Save reports — MOM, action items, decisions, follow-ups |
search_recordings | Full-text search across all recordings |
get_storage_stats | Storage usage and disk space stats |
delete_recording | Delete a recording and all its files |
How It Works
Record Meeting Whisper Transcribes Claude Analyzes
| | |
v v v
[AbhiMeet App] ---> [MCP Server] ---> [Claude Desktop/Code]
| | |
v v v
3 files saved Speech -> Text Reports, MOM,
(MP3+WebM+MP4) (90+ languages) Action Items, Summary
Usage Examples
After connecting to Claude, just talk naturally:
| What You Say | What Happens |
|---|---|
| "List my recordings" | Shows all recordings with dates, durations, file types |
| "Transcribe my latest recording" | Whisper runs locally, saves transcription |
| "Generate meeting minutes" | Claude reads transcription, creates structured MOM |
| "What action items came up?" | Extracts and lists all commitments and deadlines |
| "Summarize in Hindi" | Generates summary in Hindi from English recording |
| "What was shown on screen?" | Extracts video frames, Claude describes visual content |
| "Search recordings for 'budget'" | Finds all recordings mentioning budget |
File Naming Convention
Every recording creates a folder with 3 properly named files:
Budget-Review_160426_0230PM/
Budget-Review_Audio_160426_0230PM.mp3 (Audio only)
Budget-Review_Video_160426_0230PM.webm (Screen only)
Budget-Review_AV_160426_0230PM.mp4 (Combined)
metadata.json (Recording metadata)
Format: Subject_Type_DDMMYY_HHMMam.ext
Supported Languages
Powered by Whisper AI — supports 90+ languages including:
| Language | Code | Language | Code |
|---|---|---|---|
| English | en | Hindi | hi |
| Gujarati | gu | Marathi | mr |
| Bengali | bn | Tamil | ta |
| Telugu | te | Kannada | kn |
| Urdu | ur | Punjabi | pa |
| Arabic | ar | Chinese | zh |
| Japanese | ja | Korean | ko |
| French | fr | German | de |
| Spanish | es | Portuguese | pt |
Project Structure
AbhiMeet/
app/ # Electron desktop app
src/
main/ # Main process
index.js # App window, permissions, tray
ipc-handlers.js # Recording logic, 3-file output
file-manager.js # FFmpeg conversion, file naming
settings.js # Settings + MCP config sync
tray.js # System tray
renderer/ # Frontend UI
index.html # Dark theme, waveform, player
styles.css # Professional dark UI
app.js # Recording, playback, waveform
preload.js # Secure IPC bridge
assets/icon.svg # App icon
ffmpeg/ # FFmpeg binaries (not in repo)
package.json
mcp-server/ # Python MCP server
main.py # 15 tools + Whisper transcription
config.json # Storage path config
pyproject.toml # Dependencies
recordings/ # Auto-created recording storage
LICENSE # MIT License
README.md
Tech Stack
| Component | Technology | Purpose |
|---|---|---|
| Desktop App | Electron 33 | Cross-platform recording UI |
| Audio/Video | FFmpeg | MP3 conversion, A+V merging |
| Transcription | faster-whisper | Local Whisper AI (no cloud) |
| AI Integration | MCP (Model Context Protocol) | Claude tools |
| MCP Server | FastMCP (Python) | 15-tool server |
| UI | HTML/CSS/JS | Dark theme with live waveform |
Contributing
Contributions are welcome! Feel free to:
- Open issues for bugs or feature requests
- Submit pull requests
- Add support for more platforms (macOS, Linux)
- Improve Whisper model selection
- Add speaker diarization
Roadmap
- macOS and Linux support
- Speaker diarization (who said what)
- Auto-record when Zoom/Meet/Teams call starts
- Calendar integration for auto-naming
- Export reports to PDF/Word
- Electron auto-updater
License
MIT License - see LICENSE for details.
Author
Abhishek Tansukh Joshi
GitHub: @abhishekjoshi-eng