speaker-encoder-pipeline 0.1.1

Multilingual Speaker Encoder with Multi-branch Architecture for Few-shot Voice Synthesis

Speaker Encoder Pipeline

A multilingual speaker encoder with multi-branch architecture for few-shot voice synthesis. This package provides a robust speaker encoding system that captures timbre, cadence, and self-supervised learning features.

Features

• Multi-branch Architecture: Combines timbre, cadence, and SSL features
• Multilingual Support: English and Tamil language support
• Few-shot Learning: Efficient speaker encoding with minimal audio samples
• Modular Design: Easy to extend and customize
• CI/CD Ready: Fully configured with GitHub Actions

Installation

From PyPI (when published)

pip install speaker-encoder-pipeline

From Source

git clone https://github.com/yourusername/Zero_shotVoiceClone.git
cd speaker_encoder_pipeline

# Install in development mode
pip install -e ".[dev]"

Quick Start

Basic Usage

from model.encoder import MultilingualSpeakerEncoder
from config import SAMPLE_RATE
import torch

# Initialize model
encoder = MultilingualSpeakerEncoder(config)

# Prepare audio (mono, 16kHz)
audio = torch.randn(batch_size, SAMPLE_RATE * duration)

# Get speaker embedding
embedding = encoder(audio)
print(embedding.shape)  # (batch_size, embedding_dim)

Command-line Interface

# Prepare data
speaker-encoder-prepare --data-dir ./data

# Train model
speaker-encoder-train --epochs 100 --batch-size 32

# Validate speakers
speaker-encoder-validate --model-path ./models/best_model.pt

# Evaluate performance
speaker-encoder-evaluate --test-dir ./test_data

Development Setup

Installation with Dev Dependencies

pip install -e ".[dev]"

Running Tests

# Run all tests
pytest

# Run with verbose output
pytest -v

# Run specific test file
pytest tests/test_model.py

# Run specific test
pytest tests/test_model.py::TestMultilingualSpeakerEncoder::test_forward_pass

# Run with coverage report
pytest --cov=model --cov=data --cov=train --cov=utils --cov-report=html

# Run only unit tests
pytest -m unit

# Run without slow tests
pytest -m "not slow"

# Run in parallel (faster)
pytest -n auto

Code Quality

# Format code with black
black model/ data/ train/ utils/

# Check imports with isort
isort model/ data/ train/ utils/

# Lint with flake8
flake8 model/ data/ train/ utils/

# Type checking with mypy
mypy model/ data/ train/ utils/

Package Structure

speaker_encoder_pipeline/
├── model/
│   ├── encoder.py          # Main encoder architecture
│   ├── timbre_branch.py    # Timbre extraction
│   ├── cadence_branch.py   # Cadence extraction
│   ├── ssl_branch.py       # SSL features
│   ├── fusion.py           # Feature fusion
│   └── loss.py             # Loss functions
├── data/
│   ├── dataset.py          # Dataset classes
│   ├── preprocessor.py     # Audio preprocessing
│   ├── augmentations.py    # Data augmentation
│   ├── sampler.py          # Sampling strategies
│   └── __init__.py
├── train/
│   ├── dataloader.py       # Data loading
│   └── __init__.py
├── utils/
│   ├── audio_utils.py      # Audio utilities
│   └── __init__.py
├── tests/
│   ├── test_model.py       # Model tests
│   ├── test_data.py        # Data tests
│   ├── test_train.py       # Training tests
│   ├── test_package.py     # Package structure tests
│   └── conftest.py         # Pytest configuration
├── config.py               # Configuration
├── setup.py                # Package setup
├── pyproject.toml          # Modern Python packaging
├── pytest.ini              # Pytest configuration
├── requirements.txt        # Core dependencies
├── requirements-dev.txt    # Development dependencies
└── README.md               # This file

Configuration

Edit config.py to customize:

• Audio parameters (sample rate, channels, format)
• Dataset paths and sources
• Model architecture parameters
• Training hyperparameters

Publishing to PyPI

1. Update Version

# Update version in version.txt
echo "0.2.0" > version.txt

# Commit changes
git add version.txt
git commit -m "Bump version to 0.2.0"

2. Create a Git Tag

git tag -a v0.2.0 -m "Release version 0.2.0"
git push origin v0.2.0

3. GitHub Actions will automatically:

• Run all tests on multiple Python versions and OSes
• Build distribution packages (wheel and sdist)
• Publish to PyPI (requires PYPI_API_TOKEN secret in GitHub)
• Create a GitHub Release

4. Configure PyPI Token (one-time setup)

In your GitHub repository:

1. Go to Settings → Secrets and variables → Actions
2. Add a new secret: PYPI_API_TOKEN
3. Generate token at https://pypi.org/manage/account/token/

Testing

The project uses pytest with the following markers:

• @pytest.mark.unit - Unit tests
• @pytest.mark.integration - Integration tests
• @pytest.mark.slow - Slow tests
• @pytest.mark.gpu - GPU-required tests
• @pytest.mark.model - Model-specific tests
• @pytest.mark.data - Data processing tests
• @pytest.mark.train - Training tests

CI/CD Workflow

The GitHub Actions workflow (python-package.yml):

1. Test Job: Runs on every push and pull request

   • Tests on Ubuntu, Windows, and macOS
   • Python 3.9, 3.10, 3.11
   • Linting with flake8
   • Code formatting with black
   • Import sorting with isort
   • Coverage reports to Codecov

2. Build and Publish Job: Runs on version tags

   • Builds distribution packages
   • Validates with twine
   • Publishes to PyPI
   • Creates GitHub Release

Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/your-feature
3. Make changes and write tests
4. Ensure all tests pass: pytest
5. Format code: black and isort
6. Commit and push to your fork
7. Create a Pull Request

License

MIT License - see LICENSE file for details

Citation

If you use this package in your research, please cite:

@software{speaker_encoder_2025,
  title={Speaker Encoder Pipeline},
  author={Your Name},
  year={2025},
  url={https://github.com/yourusername/Zero_shotVoiceClone}
}

Support

For issues and questions:

• Open an issue on GitHub
• Check existing documentation
• Review test cases for usage examples

Changelog

Version 0.1.0 (Initial Release)

• Multi-branch speaker encoder
• Multilingual support (English, Tamil)
• Complete training pipeline
• CI/CD with GitHub Actions
• PyPI packaging support