Voicemo

An accessibility-focused desktop application that analyzes speech emotion during online meetings and represents estimated emotional tone using emojis.

Planned Features

• Turkish and English speech recognition
• Speech emotion recognition
• Whisper integration
• emotion2vec integration
• PyTorch-based inference
• Real-time emotion visualization
• Emoji-based emotional cues
• Local-first processing
• Accessibility-focused meeting experience

Tech Stack

• Python 3.12
• PySide6
• PyTorch
• Whisper
• emotion2vec
• ONNX Runtime
• pytest

Development

1. Clone the repository

git clone <repository-url>
cd Voicemo

2. Create a virtual environment

python3 -m venv .venv

3. Activate the virtual environment

macOS / Linux:

source .venv/bin/activate

Windows:

.venv\Scripts\activate

4. Install dependencies

Install the project and development dependencies:

python -m pip install -e ".[dev]"

This installs the required application dependencies, including PySide6, as well as development tools such as pytest and pytest-qt.

5. Start the UI

Run the application with:

python -m app.main

6. Run tests

python -m pytest

Project Structure

Voicemo/
├── app/
│   ├── audio/
│   ├── config/
│   ├── core/
│   ├── inference/
│   ├── models/
│   ├── ui/
│   ├── utils/
│   └── main.py
├── models/
│   ├── whisper/
│   └── emotion2vec/
├── tests/
├── docs/
├── .github/
│   └── workflows/
├── .gitignore
├── README.md
└── pyproject.toml

Model Pipeline

Audio Input
     │
     ├───────────────┐
     ↓               ↓
  Whisper        emotion2vec
     │               │
     ↓               ↓
Transcript     Emotional Tone
     │               │
     └───────┬───────┘
             ↓
      Emotion Engine
             ↓
       Emoji / UI

License

This project is currently under development.