lifeiteng/vall-e

Health v2 · maintenance only

Adoption not part of the score

2215 stars · 329 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

An unofficial PyTorch implementation of VALL-E, a zero-shot text-to-speech model that treats TTS as a conditional language modeling task over neural codec codes. It supports training on a single GPU and reproduces the paper's results with demo audio for English and Chinese.

Use cases

• synthesize speech in a cloned voice from a 3-second audio prompt
• train a zero-shot text-to-speech model on my own dataset
• reproduce the VALL-E paper results
• convert text to speech with emotion and acoustic environment preserved
• experiment with neural codec language models for audio generation

When to choose

• you want an open-source, trainable VALL-E implementation in PyTorch
• you need zero-shot TTS with voice cloning from short prompts
• you want to train TTS on a single consumer GPU

When to avoid

• you need a production-ready pretrained model or hosted service (none is provided to prevent misuse)
• you need a simple pip-install TTS API without training setup
• you require official support or guaranteed parity with the original model

Facets

library · maturity active

tts machine-learning deep-learning llm-training speech-processing deep-learning machine-learning python pytorch zero-shot-tts neural-codec voice-cloning speech-synthesis vall-e in-context-learning natural-language-processing linux macos gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="lifeiteng/vall-e")