Noise Suppression

What is Noise Suppression?

Noise Suppression (also known as Noise Cancellation or Noise Reduction) is a signal processing technology used to filter out unwanted background sounds from a microphone's audio input. It ensures that only the speaker's voice is transmitted, removing distractions like keyboard typing, computer fans, traffic, or office chatter.

How It Works

Traditional noise suppression algorithms use spectral subtraction. They analyze the audio frequencies to identify consistent, stationary noise patterns (like the hum of an air conditioner) and subtract them from the signal.

Modern, advanced noise suppression relies on Deep Learning (AI). Neural networks (such as RNNoise) are trained on thousands of hours of speech and noise data. These models can distinguish human speech from non-speech audio in real-time, effectively suppressing complex, non-stationary noises like a dog barking or a baby crying, which traditional methods often miss.

Noise Suppression in WebRTC

WebRTC implementations in browsers typically include a built-in noise suppression module as part of the audio processing pipeline. This standard suppression is effective for steady-state noise. However, many modern video calling applications implement additional, AI-based noise suppression layers (using WebAssembly or cloud processing) to provide professional-grade clarity.

Trade-offs

While highly effective, aggressive noise suppression can sometimes degrade speech quality, making voices sound robotic or cutting off the ends of words. It also adds a small amount of processing latency to the call.