# NATSpeech/NATSpeech

A Non-Autoregressive Text-to-Speech (NAR-TTS) framework, including official PyTorch implementation of PortaSpeech (NeurIPS 2021) and DiffSpeech (AAAI 2022)

Repository: https://github.com/NATSpeech/NATSpeech
Canonical: https://ross.abutalabs.com/products/natspeech
Language: Python
License: MIT
License Family: permissive
Topics: speech-synthesis, pytorch, tts, speech, huggingface, portaspeech, diffsinger, diffspeech
Last push: 2023-04-02T00:55:24+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1662, "days_push": 1250, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1004, forks 99 (observed 2026-09-03T02:15:06.754031+00:00)

## What it is
A PyTorch framework for non-autoregressive text-to-speech (NAR-TTS), containing official implementations of PortaSpeech (NeurIPS 2021) and DiffSpeech (AAAI 2022). It provides data processing with Montreal Forced Aligner, training, and inference pipelines.

## Use cases
- synthesize speech from text with a non-autoregressive TTS model
- train a PortaSpeech text-to-speech model
- run DiffSinger singing voice synthesis
- reproduce NeurIPS/AAAI TTS paper results
- build a custom TTS training pipeline in PyTorch
- use pretrained TTS models from Hugging Face

## When to choose
- you need official implementations of PortaSpeech or DiffSpeech for research
- you want a trainable non-autoregressive TTS framework in PyTorch
- you need forced-alignment-based data prep for TTS

## When to avoid
- you need a production-ready, actively maintained TTS system
- you want autoregressive or latest large-scale TTS models
- you need Windows or macOS support without extra setup

## Facets
- artifact type: framework
- maturity: maintenance
- function: tts, machine-learning, deep-learning, audio-processing
- domain: speech-processing, machine-learning, deep-learning
- platform: python
- tags: text-to-speech, non-autoregressive, pytorch, diffusion, portaspeech, diffspeech, speech-synthesis, research-code, audio, linux, gpu

## Member repositories
- NATSpeech/NATSpeech (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:06.754031+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 2026-08-30T07:13:42.226783+00:00, confidence not recorded.
  - readme: https://github.com/NATSpeech/NATSpeech (fetched 2026-09-03T02:15:06.754031+00:00, sha c5bf43744ad1)
- Data as of 2026-08-30T08:39:29.467469+00:00.