intro algorithm sounds

Single-channel noise reduction - sound examples

The following sound examples illustrate the speech enhancement capabilities of single-channel noise reduction algorithms in different noise scenarios. For each scenario two sound files are presented :

  • the unprocessed sound file, which is the signal captured by the front microphone of the right hearing aid. The front microphone signal of the right hearing aid is the input to the noise reduction algorithm.
  • the processed sound file, which is the output of the MMSE optimal filter algorithm based on Laplacian speech and Gaussian noise priors.

Low-reverberant room with stationary directional background noise

  • low-reverberant room
  • male speaker in front of the listener (at 0°)
  • directional stationary noise source having the average spectrum of human speech at the right-hand side of the listener (at 60°)
  • signal-to-noise ratio :

We notice that the background noise has been effectively reduced in both cases (SNR=0 dB and SNR=10 dB). The desired speech signal shows very little distortion and the noise reduction has not affected the characteristic of the residual background noise, it still sounds natural.

Cafeteria with babble noise in the background

  • cafeteria room
  • female speaker in front of the listener (at 0°)
  • diffuse cocktail-party noise (crowded cafeteria)
  • signal-to-noise ratio :

The reduction of fluctuating noise such as cafeteria babble is a challenging task in single channel noise reduction. We observe little speech distortion, but still a noticeable reduction of the noise level. In particular for the case SNR=10 dB we not only notice noise reduction, but also a slight reduction of reverberation which is a property of most single channel noise reduction algorithms.