# facebookresearch/denoiser

Real Time Speech Enhancement in the Waveform Domain (Interspeech 2020)We provide a PyTorch implementation of the paper Real Time Speech Enhancement in the Waveform Domain. In which, we present a causal speech enhancement model working on the raw waveform that runs in real-time on a laptop CPU. The proposed model is based on an encoder-decoder architecture with skip-connections. It is optimized on both time and frequency domains, using multiple loss functions. Empirical evidence shows that it is capable of removing various kinds of background noise including stationary and non-stationary noises, as well as room reverb. Additionally, we suggest a set of data augmentation techniques applied directly on the raw waveform which further improve model performance and its generalization abilities.

Repository: https://github.com/facebookresearch/denoiser
Canonical: https://ross.abutalabs.com/products/denoiser
Language: Python
License: NOASSERTION
License Family: other
Archived: true
Last push: 2023-03-14T18:56:57+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2193, "days_push": 1268, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1899, forks 320 (observed 2026-08-28T04:05:51.411255+00:00)

## What it is
A PyTorch library implementing a causal, real-time speech enhancement model that operates on raw waveforms to remove background noise and reverb. It is based on the Demucs encoder-decoder architecture and ships with pretrained models usable offline or in streaming mode.

## Use cases
- remove background noise from speech recordings
- real-time noise suppression for voice calls
- denoise audio in python
- speech enhancement pretrained model
- remove room reverb from voice
- streaming audio denoising

## When to choose
- you need real-time or streaming speech denoising on CPU
- you want a pretrained model out of the box via pip
- you need to remove stationary, non-stationary noise, or reverb from speech

## When to avoid
- you need commercial use - the model weights are CC-BY-NC licensed
- you need music separation or general audio enhancement beyond speech
- you need actively maintained software - releases have been infrequent since 2023

## Facets
- artifact type: library
- maturity: maintenance
- function: audio-processing, machine-learning, deep-learning
- domain: speech-processing, machine-learning
- platform: python, cross-platform, windows
- tags: speech-enhancement, noise-reduction, pytorch, real-time, demucs, pretrained-model, research, audio, macos, linux

## Member repositories
- facebookresearch/denoiser (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:51.411255+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 2026-08-30T03:12:10.952502+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/denoiser (fetched 2026-08-28T04:05:51.411255+00:00, sha ed83aeb09f51)
  - registry_pypi: https://pypi.org/pypi/denoiser/json (fetched 2026-08-29T10:51:17.892628+00:00, sha 2b2ae6a749f6)
- Data as of 2026-08-30T08:39:29.467469+00:00.