loudness.ai
EN
English (EN)Español (ES)

Guide

Install the AI Noise Remover SDK

This guide installs the AI Noise Remover SDK, checks that it works and removes the noise from your first recording. It takes about ten minutes, and everything runs on your own computer.

Updated 16 October 2026

  1. 1Check what you need

    You need Python 3.11, 3.12, 3.13 or 3.14 and FFmpeg, which includes FFprobe. The JavaScript edition needs Node.js 22 or 24 instead of Python. No graphics card is needed: the speech model runs on the CPU.

    python3 --version
ffmpeg -version

    On Windows, use py --version. If Python is missing, install it from python.org and, on Windows, tick “Add python.exe to PATH”.

  2. 2Install FFmpeg

    On macOS with Homebrew:

    brew install ffmpeg

    On Ubuntu or Debian:

    sudo apt-get install ffmpeg

    On Windows:

    winget install --id Gyan.FFmpeg

    Close the terminal and open a new one, so it finds the new program, then run ffmpeg -version again.

  3. 3Install the SDK in a virtual environment

    A virtual environment keeps the SDK and its dependencies apart from your other Python projects. On macOS and Linux:

    python3 -m venv .venv
source .venv/bin/activate
python -m pip install loudness-audio-kit

    On Windows, in PowerShell:

    py -m venv .venv
.venv\Scripts\activate
python -m pip install loudness-audio-kit

  4. 4Check the installation

    loudness-audio processing-doctor

    The command prints a JSON report. "ready": true and "speech_engine": "ready" mean that FFmpeg was found and the speech model loads.

  5. 5Remove noise from your first recording

    loudness-audio denoise recording.wav --output clean.wav --progress

    Use enhance instead of denoise to also reduce low rumble, compress gently and normalize loudness, for example for a podcast:

    loudness-audio enhance podcast.mp3 --output podcast-clean.wav --target-lufs -16

    The same from Python:

    from loudness_audio_kit import denoise, enhance

denoise("recording.wav", "clean.wav")
enhance("podcast.mp3", "podcast-clean.wav", target_lufs=-16)

  6. 6Optional: build the JavaScript and TypeScript edition

    The Node.js edition is built from the GitHub repository:

    git clone https://github.com/Loudness-ai/ai-noise-remover-sdk.git
cd ai-noise-remover-sdk/javascript
npm ci --ignore-scripts
npm run build
node dist/cjs/cli.js processing-doctor

  7. 7Optional: run the local web demo

    The repository includes a small web page for uploading, processing and downloading audio on your own machine. From the javascript folder, after the build:

    npm ci --prefix demo --ignore-scripts
npm run demo

    Then open http://127.0.0.1:7860/. A Docker setup is described in the demo guide on GitHub.

Common problems

FFmpeg is not found

Install FFmpeg as in step 2, then open a new terminal. On Windows, check that ffmpeg -version works in the new window before you run the SDK.

“loudness-audio” is not recognized

Activate the virtual environment first (step 3). Each new terminal needs source .venv/bin/activate, or .venv\Scripts\activate on Windows.

pip cannot find a matching version

The SDK needs Python 3.11 to 3.14. Check python --version and create the environment with a supported version, for example py -3.12 -m venv .venv on Windows.

PowerShell refuses to activate the environment

Run Set-ExecutionPolicy -Scope CurrentUser RemoteSigned once, then activate it again.

Music sounds worse after processing

The speech model is made for voice recordings and can suppress music. For music, use normalize, which changes loudness only.

Next steps

More guides

All developer guides

More from Loudness.ai

Sign in to Loudness.ai