noise-suppression-for-voice is a real-time noise suppression plugin leveraging Xiph's RNNoise library. It aims to reduce a wide spectrum of background noise, including computer fans, office sounds, and traffic, to improve voice clarity. The plugin is designed to work with single or multiple audio channels at 48000 Hz sample rates. It provides a minimalistic GUI for parameter adjustments and diagnostic stats.
Notable for its cross-platform compatibility (VST, VST3, LV2, LADSPA, AU, AUv3, PipeWire), ease of configuration via GUI and command-line tools, and efficient noise reduction algorithm based on RNNoise. The plugin offers flexible implementation across different operating systems and audio frameworks, addressing the need for noise cancellation in various applications.
- Core Functionality: Real-time noise reduction using the RNNoise algorithm. Supports single and multi-channel audio input.
- Supported Platforms: Windows (using Equalizer APO and native plugins), Linux (using PipeWire, LADSPA, and PulseAudio), macOS (to be implemented).
- Configuration: GUI with VAD Threshold, VAD Grace Period, and Retroactive VAD Grace Period. Command-line options available for PulseAudio.
- Performance: Designed for low latency, suitable for real-time applications. The architecture supports efficient resource utilization.
- Developer Experience: Vendored dependencies ensure easy builds and no external submodule management. Clear build instructions are provided.
The project is actively maintained with recent commits and issue responses. Documentation is available for various platforms. It has been tested on multiple operating systems, including Windows, Linux, and macOS, indicating reasonable stability. However, there are unresolved issues and open problems, particularly regarding cross-platform support and configuration complexities. The project adopts a contribution-focused approach.
This project benefits users seeking to improve speech intelligibility in noisy environments. It's valuable for applications such as voice communication, audio recording, and voice-controlled systems. It provides a straightforward way to integrate a powerful noise suppression algorithm into existing workflows, offering a practical alternative to manual noise reduction or more complex audio processing pipelines.