https://github.com/michael-borck/deep-talk
Transcribes and analyzes audio/video conversations locally with AI-powered insights.
https://github.com/michael-borck/deep-talk
ai ai-analysis ai-powered-transcription audio-processing audio-video-processing conversation-analysis cross-platform desktop-app edtech electron ffmpeg javascript local-processing natural-language-processing privacy privacy-first speech-to-text transcription typescript video-processing
Last synced: 28 days ago
JSON representation
Transcribes and analyzes audio/video conversations locally with AI-powered insights.
- Host: GitHub
- URL: https://github.com/michael-borck/deep-talk
- Owner: michael-borck
- License: mit
- Created: 2025-05-30T03:37:40.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2026-04-12T22:01:51.000Z (about 1 month ago)
- Last Synced: 2026-04-12T22:26:59.003Z (about 1 month ago)
- Topics: ai, ai-analysis, ai-powered-transcription, audio-processing, audio-video-processing, conversation-analysis, cross-platform, desktop-app, edtech, electron, ffmpeg, javascript, local-processing, natural-language-processing, privacy, privacy-first, speech-to-text, transcription, typescript, video-processing
- Language: TypeScript
- Size: 85.7 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# DeepTalk
[](https://github.com/topics/ai-powered-transcription) [](https://github.com/topics/cross-platform) [](https://github.com/topics/desktop-app) [](https://github.com/topics/ffmpeg) [](https://github.com/topics/local-processing) [](https://github.com/topics/natural-language-processing) [](https://github.com/topics/privacy-first) [](https://github.com/topics/typescript) [](https://github.com/topics/audio-video-processing) [](https://github.com/topics/edtech)
[](https://github.com/michael-borck/deep-talk)
[](https://github.com/michael-borck/deep-talk/tree/main/docs)
[](https://gitingest.com/michael-borck/deep-talk)
[](https://deepwiki.com/michael-borck/deep-talk)
AI-powered conversation analysis and insight discovery platform with local processing and privacy-first design.
## Features
- **Audio/Video Support**: MP3, WAV, MP4, AVI, MOV, M4A, WebM, OGG, and more
- **Privacy-First**: Transcription and speaker diarisation run entirely on your machine. AI analysis uses your choice of local Ollama or any of OpenAI / Anthropic / Groq / Gemini / OpenRouter / custom — your call.
- **Local Whisper**: Built-in English transcription via `@huggingface/transformers`. No external server required.
- **Local Speaker Diarisation**: pyannote-segmentation-3.0 + wespeaker for real audio-level "who said what". No LLM guessing from text.
- **In-app Documentation**: Rendered with the app's theme, ships with the binary, no internet required.
- **Cross-Platform**: macOS, Windows, Linux. FFmpeg bundled.
## Installation
Download the latest release for your platform from the [Releases](https://github.com/michael-borck/deep-talk/releases) page.
## Development
### Prerequisites
- Node.js 20+
- npm or yarn
- (Optional) Speaches service running on http://localhost:8000
- (Optional) Ollama service running on http://localhost:11434
### Setup
```bash
# Clone the repository
git clone https://github.com/michael-borck/deep-talk.git
cd deep-talk
# Install dependencies
npm install
# Start development server
npm start
# Build for production
npm run dist
```
## Release Process
### GitHub Secrets Required
To enable automatic builds when you create a release tag, set up these GitHub secrets:
1. **For macOS Code Signing (Optional)**:
- `MAC_CERTS`: Base64 encoded .p12 certificate
- `MAC_CERTS_PASSWORD`: Certificate password
- `APPLE_ID`: Your Apple ID
- `APPLE_ID_PASS`: App-specific password
- `APPLE_TEAM_ID`: Your Apple Developer Team ID
2. **Automatic (Already exists)**:
- `GITHUB_TOKEN`: Automatically provided by GitHub Actions
### Creating a Release
1. Update version in `package.json`
2. Commit changes: `git commit -am "Bump version to v1.0.0"`
3. Create tag: `git tag v1.0.0`
4. Push tag: `git push origin v1.0.0`
5. GitHub Actions will automatically build for all platforms
6. Edit the draft release on GitHub and publish
### Build Outputs
- **Windows**: `.exe` installer
- **macOS**: `.dmg` installer and `.pkg` for Mac App Store
- **Linux**: `.AppImage` and `.deb` packages
## Architecture
```
DeepTalk/
├── src/ # React TypeScript source
├── public/ # Electron main process
├── database/ # SQLite schema
└── ffmpeg-binaries/ # Platform-specific FFmpeg
```
## Technologies
- **Frontend**: React + TypeScript
- **Desktop**: Electron
- **Database**: SQLite (better-sqlite3)
- **Styling**: Tailwind CSS
- **Transcription**: Speaches API
- **AI Analysis**: Ollama API
- **Media Processing**: FFmpeg (bundled)
## License
MIT