# enhuiz/vall-e

An unofficial PyTorch implementation of the audio LM VALL-E

Repository: https://github.com/enhuiz/vall-e
Canonical: https://ross.abutalabs.com/products/vall-e
Language: Python
License: MIT
License Family: permissive
Topics: vall-e, valle, text-to-speech, pytorch, tts, audio-lm
Last push: 2023-05-10T05:55:34+00:00

## Health v2 (maintenance only)
Score: 31/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 95
- inputs: {"age_days": 1330, "days_push": 1211, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2976, forks 399 (observed 2026-08-28T04:07:32.977193+00:00)

## What it is
An unofficial PyTorch implementation of the VALL-E text-to-speech audio language model, built on the EnCodec tokenizer. It provides training scripts for AR and NAR models plus a CLI for synthesizing speech from text with a reference audio sample.

## Use cases
- train a VALL-E text-to-speech model on custom audio data
- clone a voice from a short reference wav clip
- synthesize speech from text using an audio language model
- experiment with AR and NAR transformer models for TTS
- quantize audio with EnCodec for neural TTS training

## When to choose
- you want to train or fine-tune VALL-E from scratch on your own data
- you need an open-source PyTorch codebase for audio-LM TTS research
- you have a DeepSpeed-compatible GPU setup and want to experiment

## When to avoid
- you need a ready-to-use pretrained model with polished demos
- you want production TTS without training infrastructure
- you cannot meet DeepSpeed GPU and CUDA/ROCm compiler requirements
- you need permissively licensed audio decoding (EnCodec is CC-BY-NC 4.0)

## Facets
- artifact type: library
- maturity: experimental
- function: tts, machine-learning, deep-learning, llm-training, audio-processing, cli
- domain: speech-processing, machine-learning, deep-learning
- platform: python, cli
- tags: vall-e, text-to-speech, audio-language-model, pytorch, encodec, voice-cloning, unofficial-implementation, audio, linux, gpu

## Member repositories
- enhuiz/vall-e (main) score 31

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:32.977193+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:47:28.332461+00:00, confidence not recorded.
  - readme: https://github.com/enhuiz/vall-e (fetched 2026-08-28T04:07:32.977193+00:00, sha 196790881f0b)
- Data as of 2026-08-30T08:39:29.467469+00:00.