Fast and local neural text-to-speech engine

A PyTorch implementation of Location-Relative Attention Mechanisms For Robust Long-Form Speech Synthesis

Encode and decode audio samples to/from compressed latent representations!

Efficient optimizers

Lightweight package that tracks and summarizes code changes using LLMs (Large Language Models)

Minimal implementation of scalable rectified flow transformers, based on SD3's approach

The python library for real-time communication

NanoGPT (124M) in 3 minutes

Powerful & Easy-to-Use Video Face Swapping and Editing Software

Implementation of Band Split Roformer, SOTA Attention network for music source separation out of ByteDance AI Labs

Meta Lingua: a lean, efficient, and easy-to-hack codebase to research LLMs.

Minimalistic, hackable PyTorch implementation of SimSiam in ~400 lines. Achieves good performance on ImageNet with ResNet50. Features distributed training, real-time KNN eval, and AMP. Perfect for …

Official repository of Wavehax vocoder

Code for vec2wav 2.0, a speech token vocoder for VC. Paper: https://arxiv.org/abs/2409.01995

Amphion (/æmˈfaɪən/) is a toolkit for Audio, Music, and Speech Generation. Its purpose is to support reproducible research and help junior researchers and engineers get started in the field of audi…

Minimal, clean code for the Byte Pair Encoding (BPE) algorithm commonly used in LLM tokenization.

MP-SENet: A Speech Enhancement Model with Parallel Denoising of Magnitude and Phase Spectra

Tools for handling speech data in machine learning projects.

A simple but complete full-attention transformer with a set of promising experimental features from various papers

Implementation of E2-TTS, "Embarrassingly Easy Fully Non-Autoregressive Zero-Shot TTS", in Pytorch

State-of-the-art audio codec with 90x compression factor. Supports 44.1kHz, 24kHz, and 16kHz mono/stereo audio.

A fast implementation of T5/UL2 in PyTorch using Flash Attention

VAE modified from Descript Audio Codec, which replaces the RVQ with VAE

Tacotron 2 - PyTorch implementation with faster-than-realtime inference