Adding DeepFilterNet Noise Reduction to Easy Effects on Arch Linux

Introduction

Easy Effects (formerly PulseEffects) is a powerful audio equalizer and effects system for Linux that provides studio-quality audio processing right from your desktop. While the basic installation through most Linux distributions includes a comprehensive suite of audio effects, one crucial component is often missing: the neural network-based noise reduction plugin from DeepFilterNet.

DeepFilterNet is a state-of-the-art speech enhancement framework that uses deep learning to remove background noise from audio in real-time. Unlike traditional noise gates or spectral subtraction methods, DeepFilterNet leverages advanced neural networks trained on thousands of hours of audio to distinguish between speech and background noise with remarkable accuracy. This makes it particularly effective for removing complex, dynamic background sounds like keyboard typing, fan noise, or ambient chatter while preserving the clarity and natural quality of human speech.

Unfortunately, many Linux distributions, including Arch Linux, don't package the DeepFilterNet LADSPA plugin alongside Easy Effects due to licensing restrictions, file size considerations, or simply packaging oversights. This leaves users with a powerful audio processing suite but missing one of its most impressive capabilities.

In this guide, we'll walk through the process of manually installing the DeepFilterNet plugin to complete your Easy Effects setup, enabling professional-grade noise reduction for podcasts, voice calls, recordings, and any other audio application where clean speech matters.

Understanding the Components

Before diving into the installation process, it's helpful to understand what we're working with:

Easy Effects provides the framework and user interface for audio processing. It acts as a host for various LADSPA (Linux Audio Developer's Simple Plugin API) plugins, managing the audio routing and providing an intuitive GUI for controlling the effects.

DeepFilterNet is the neural network engine that performs the actual noise reduction. It processes audio in real-time, analyzing the frequency spectrum and using machine learning models to identify and suppress background noise while preserving speech components.

LADSPA Plugin serves as the bridge between Easy Effects and DeepFilterNet. This plugin follows the LADSPA standard, allowing Easy Effects to load and control the DeepFilterNet processing engine as if it were any other audio effect.

Prerequisites

Before proceeding with the installation, ensure you have:

On Arch Linux, you can install Easy Effects using:

sudo pacman -S easyeffects

The Installation Process

Step 1: Identify Your System Architecture

First, determine whether you're running a 64-bit system (most common) or need a different architecture. You can check this using:

uname -m

For most modern systems, this will return x86_64, indicating a 64-bit architecture.

Step 2: Download the DeepFilterNet Plugin

The DeepFilterNet project provides pre-compiled LADSPA plugins through their GitHub releases. Visit the official releases page to find the appropriate version for your system:

GitHub Releases URL: https://github.com/Rikorose/DeepFilterNet/releases

Look for the latest stable release and download the appropriate LADSPA plugin file. For 64-bit systems, you'll typically want a file named something like: - libdeep_filter_ladspa-0.5.6-x86_64-unknown-linux-gnu.so

You can download this file directly using wget or curl:

# Example downloading version 0.5.6 for x86_64
curl -LO https://github.com/Rikorose/DeepFilterNet/releases/download/v0.5.6/libdeep_filter_ladspa-0.5.6-x86_64-unknown-linux-gnu.so

Alternatively, you can use your web browser to download the file from the releases page.

Step 3: Install the Plugin to the System LADSPA Directory

Linux audio applications typically look for LADSPA plugins in standard system directories. The most common location is /usr/lib64/ladspa/ on 64-bit systems. We'll move the downloaded plugin to this location and give it a simpler name that Easy Effects can easily recognize.

# Move the plugin to the system LADSPA directory and rename it
sudo mv -v libdeep_filter_ladspa-0.5.6-x86_64-unknown-linux-gnu.so /usr/lib64/ladspa/libdeep_filter_ladspa.so

The -v flag provides verbose output, confirming the move operation. The command should output something like:

renamed '/path/to/downloaded/libdeep_filter_ladspa-0.5.6-x86_64-unknown-linux-gnu.so' -> '/usr/lib64/ladspa/libdeep_filter_ladspa.so'

Why rename the file? The version-specific naming in the downloaded file helps with package management, but Easy Effects typically looks for plugins with simpler, standardized names. Renaming to libdeep_filter_ladspa.so makes it easier for the application to discover and load the plugin.

Step 4: Verify the Installation

After installing the plugin, you can verify that it's properly placed and accessible:

# Check that the file exists and has appropriate permissions
ls -la /usr/lib64/ladspa/libdeep_filter_ladspa.so

This should show the plugin file with read permissions for all users and appropriate ownership (typically root:root).

Step 5: Configure Easy Effects

Now that the plugin is installed, it's time to configure it in Easy Effects:

  1. Launch Easy Effects: Start the application from your application menu or using the command line: easyeffects

  2. Select Input Device: In the Easy Effects window, make sure you're working with the correct input device (your microphone).

  3. Add DeepFilterNet Plugin:

  4. Click on the "Add effect" or "+" button in the input effects section
  5. Look for "DeepFilterNet" in the list of available plugins

  6. Select it to add to your effects chain

  7. Configure the Plugin:

  8. Noise Reduction Level: Start with a moderate setting (around 50-70%)
  9. Aggressiveness: Higher settings remove more noise but may affect speech quality

  10. Voice Preservation: Balance between noise removal and natural speech quality

  11. Test the Configuration:

  12. Speak into your microphone while playing some background noise
  13. Adjust the settings in real-time to find the optimal balance
  14. Use the bypass toggle to compare the processed and unprocessed audio

Important Configuration Steps

Configure Easy Effects to Launch on Startup

To ensure your noise reduction settings are persistent and always active when you need them, you should configure Easy Effects to launch automatically when you log in:

Option 1: Use Easy Effects Built-in Startup Setting (Recommended)

  1. Open Easy Effects
  2. Click the menu button (☰) in the top-right corner
  3. Select Preferences
  4. In the General tab, look for "Launch Easy Effects at startup" or "Start on login"
  5. Enable this toggle option

Option 2: System-level Startup Configuration

If you prefer system-level configuration or the built-in option isn't available:

For systems with GNOME Desktop:

  1. Open Settings → Applications → Startup Applications
  2. Click Add and browse to Easy Effects
  3. Alternatively, run this command to add it to startup:
cp /usr/share/applications/com.github.wwmm.easyeffects.desktop ~/.config/autostart/

For systems with KDE Plasma:

  1. Open System Settings → Startup and Shutdown → Autostart
  2. Click Add Application and select Easy Effects

Why is this important? Easy Effects needs to be running to process audio through the DeepFilterNet plugin. Without it configured to launch on startup, you'll need to manually start the application each time you log in, and your noise reduction settings won't be automatically applied.

Configure Applications to Use the Correct Audio Input

This is a crucial step that many users miss: applications must use the "Easy Effects Source" virtual input device, not your standard microphone input.

When Easy Effects is running, it creates a virtual input source called "Easy Effects Source" (or similar depending on your system configuration). This virtual device captures your raw microphone input, processes it through DeepFilterNet and any other effects you've configured, and then outputs the filtered audio.

How to configure your applications:

For audio recording software (Audacity, etc.):

For video conferencing applications:

For system-wide configuration:

⚠️ Important Warning: If applications use your standard microphone input instead of "Easy Effects Source," they will bypass all audio processing entirely. This means DeepFilterNet noise reduction (and any other effects) will not be applied, and you'll wonder why the noise reduction isn't working despite everything being correctly configured in Easy Effects.

To verify it's working:

  1. Start Easy Effects with DeepFilterNet enabled
  2. Open your system sound settings or pavucontrol
  3. Look at the input devices - you should see both your physical microphone and "Easy Effects Source"
  4. Configure your recording/conferencing app to use "Easy Effects Source"
  5. Test by recording audio or joining a call - the noise reduction should now be active

Troubleshooting Common Issues

Plugin Not Appearing in Easy Effects

If DeepFilterNet doesn't appear in your list of available plugins:

  1. Restart Easy Effects: Completely close and restart the application
  2. Check file permissions: Ensure the plugin is readable by all users: sudo chmod 644 /usr/lib64/ladspa/libdeep_filter_ladspa.so
  3. Verify LADSPA path: Some systems may use different paths. Check if plugins are loaded from:
  4. /usr/lib/ladspa/
  5. /usr/local/lib/ladspa/
  6. ~/.ladspa/

Audio Quality Issues

If you experience poor audio quality or distortion:

  1. Reduce noise reduction intensity: Lower the settings to preserve more natural speech
  2. Check input levels: Ensure your microphone isn't clipping or too quiet
  3. Update audio drivers: Outdated audio drivers can cause compatibility issues

Performance Issues

If you experience high CPU usage or audio stuttering:

  1. Close other applications: Neural network processing can be CPU-intensive
  2. Adjust buffer sizes: In Easy Effects settings, increase the buffer size if available
  3. Check system resources: Monitor CPU usage during processing

Advanced Configuration

For users wanting more control over the noise reduction process:

Multiple Microphone Setup

If you use multiple microphones, you can:

Integration with Other Effects

DeepFilterNet works well in combination with other audio effects:

Automation and Scripts

For advanced users, you can automate the noise reduction setup:

Performance Considerations

DeepFilterNet represents a significant advancement in noise reduction technology, but it's important to understand its resource requirements:

Alternative Solutions

While DeepFilterNet is excellent, it's worth knowing about other noise reduction options:

Conclusion

Installing DeepFilterNet for Easy Effects transforms your Linux audio setup from good to professional-grade. The neural network-based noise reduction can dramatically improve audio quality for voice recordings, podcasts, video calls, and any application where clear speech communication is essential.

While the manual installation process might seem intimidating at first, it's actually quite straightforward and provides access to one of the most advanced noise reduction technologies available in open-source audio processing. The ability to remove complex background noise while preserving natural speech quality makes this setup particularly valuable for content creators, remote workers, and anyone who needs clear audio in less-than-ideal environments.

Take the time to experiment with different settings and find the configuration that works best for your specific use case and environment. The results can be truly impressive, turning noisy, unusable audio into clear, professional-quality sound.

Further Reading