# NVIDIA/waveglow

A Flow-based Generative Network for Speech Synthesis

Repository: https://github.com/NVIDIA/waveglow
Canonical: https://ross.abutalabs.com/products/waveglow
Language: Python
License: BSD-3-Clause
License Family: permissive
Last push: 2023-10-19T15:19:59+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2856, "days_push": 1049, "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 2339, forks 535 (observed 2026-08-28T04:06:38.557060+00:00)

## What it is
WaveGlow is a PyTorch implementation of a flow-based generative network that synthesizes high-quality speech audio from mel-spectrograms, combining ideas from Glow and WaveNet without auto-regression. It includes pretrained models, inference scripts, and training code for single- or multi-GPU setups.

## Use cases
- generate speech audio from mel-spectrograms
- vocoder for text-to-speech pipeline
- train a flow-based audio synthesis model on LJSpeech
- fast non-autoregressive waveform generation on GPU
- convert mel-spectrogram files to wav audio
- research on generative models for speech

## When to choose
- you need a GPU-accelerated vocoder to turn mel-spectrograms into audio
- you want non-autoregressive synthesis with quality comparable to WaveNet
- you want to train or fine-tune a flow-based vocoder on your own dataset

## When to avoid
- you need actively maintained code with recent framework compatibility
- you lack an NVIDIA GPU, since inference and training are GPU-oriented
- you need the latest state-of-the-art vocoder quality or CPU-friendly inference

## Facets
- artifact type: library
- maturity: maintenance
- function: tts, audio-processing, deep-learning, machine-learning
- domain: speech-processing, deep-learning, machine-learning
- platform: python, cross-platform
- tags: vocoder, flow-based-generative-model, mel-spectrogram, pytorch, speech-synthesis, waveglow, audio, gpu, linux

## Member repositories
- NVIDIA/waveglow (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:38.557060+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 2026-08-30T02:37:26.534372+00:00, confidence not recorded.
  - readme: https://github.com/NVIDIA/waveglow (fetched 2026-08-28T04:06:38.557060+00:00, sha d70f8c50ad15)
- Data as of 2026-08-30T08:39:29.467469+00:00.