SVoice is a PyTorch-based implementation of Facebook Research’s study on speaker voice separation as described in the paper “Voice Separation with an Unknown Number of Multiple Speakers.” This project presents a deep learning framework capable of separating mixed audio sequences where several people speak simultaneously, without prior knowledge of how many speakers are present. The model employs gated neural networks with recurrent processing blocks that disentangle voices over multiple computational steps, while maintaining speaker consistency across output channels. Separate models are trained for different speaker counts, and the largest-capacity model dynamically determines the actual number of speakers in a mixture. The repository includes all necessary scripts for training, dataset preparation, distributed training, evaluation, and audio separation.
Features
- End-to-end PyTorch implementation for speech separation with unknown speaker counts
- Uses gated RNN blocks and convolutional encoders for robust multi-speaker modeling
- Configurable via Hydra with automatic checkpointing and experiment management
- Supports distributed multi-GPU training and easy dataset configuration
- Includes dataset generation tools for noisy and reverberant synthetic mixtures
- Built-in evaluation and inference tools for separating and scoring speech samples
Project Samples
