OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Audio & Voice AI

Deepfilternet

Deepfilternet is a deep-learning speech-enhancement framework that uses deep filtering to reduce noise in full-band 48 kHz audio. It includes pretrained models, a precompiled command-line tool for WAV files, Python components for enhancement and training, and a LADSPA plugin for real-time noise suppression with PipeWire.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Deepfilternet — OpenViST project profile in Audio & Voice AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use the precompiled tool for supported 48 kHz WAV files, the Python framework for Python-based enhancement or model work, and the LADSPA/PipeWire integration for a real-time microphone path.

Who it is for

  • People processing noisy 48 kHz WAV recordings.
  • Audio application developers exploring speech enhancement components.
  • Linux users configuring real-time microphone noise reduction through PipeWire and LADSPA.
  • Researchers and developers preparing datasets or training speech-enhancement models.

Documented uses

  • Suppress noise in supported WAV recordings using the precompiled command-line tool.
  • Use the Python package and PyTorch backend for audio enhancement workflows.
  • Integrate a noise-suppression microphone path using the LADSPA plugin and PipeWire filter-chain integration.
  • Prepare speech, noise and room-impulse-response datasets and train models.
Before you choose

Requirements & limits

  • The precompiled deep-filter binary is documented for 48 kHz WAV inputs.
  • The Python installation requires PyTorch; the training extra is described as Linux-only.
  • The framework is described as supporting Linux, macOS and Windows, while training is tested only under Linux.
  • The bundled demo’s run instructions are Linux-only and require Rust nightly.
  • The repository separates file processing, Python framework components, Rust audio/data components and real-time LADSPA integration.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Suppress noise in a 48 kHz WAV file with the precompiled binary

Process noisy WAV audio with the documented command-line tool.

Before you start

  • A precompiled deep-filter binary downloaded from the project’s release page.
  • A noisy WAV file sampled at 48 kHz.

Steps

  1. Download the precompiled binary from the DeepFilterNet releases page.

  2. Run deep-filter with the WAV file as its input; optionally set an output directory with -o.

Expected outcome

The tool processes the input WAV and writes output to the selected directory, or to its default out directory.

Guide limitations

  • The documented binary supports WAV files sampled at 48 kHz.
  • The command-line binary is distinct from the Python backend and framework installation.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
Rikorose/DeepFilterNet · Official homepage title · retained in its source language
How we review sourcesSuggest a correction