International Audio Laboratories Erlangen

Evaluation and Comparison of Late Reverberation Power Spectral Density Estimators

Sebastian Braun, Adam Kuklasinski, Ofer Schwartz, Oliver Thiergart, Emanuel A. P. Habets, Sharon Gannot, Simon Doclo and Jesper Jensen

Published in the IEEE/ACM Transactions on Audio, Speech and Language Processing, Vol. 26, Issue 6, pp. 1052-1067, Jun. 2018

Abstract

Reduction of late reverberation can be achieved using spatio-spectral filters such as the multichannel Wiener filter (MWF). To compute this filter, an estimate of the late reverberation power spectral density (PSD) is required. In recent years, a multitude of late reverberation PSD estimators have been proposed. In this contribution, these estimators are categorized into several classes, their relations and differences are discussed, and a comprehensive experimental comparison is provided. To compare their performance, simulations in controlled as well as practical scenarios are conducted. It is shown that a common weakness of spatial coherence-based estimators is their performance in high direct-to-diffuse ratio (DDR) conditions. To mitigate this problem, a correction method is proposed and evaluated. It is shown that the proposed correction method can decrease the speech distortion without significantly affecting the reverberation reduction.

Audio Examples

Acoustic setup:

  • Uniform circular array of 6 omnidirectional microphones with radius 4.5 cm
  • Measured room impulse responses in large conference room with T60 = 800 ms
  • SNR to additive pink noise of 15 dB

Description:

  • A multichannel Wiener filter (MWF) is used to extract the direct sound while suppressing late reverberation and noise.
  • The source position and noise PSD matrix are known.
  • The PSD of the late reverberation is estimated using various estimators.
  • Switch between the various processed and unprocessed audio files to compare how the MWF sounds using the different PSD estimators.

Note: Please use Google Chrome if you experience playback problems.

Comparison between all diffuse PSD estimators (source distance 3.5 m):

Activate

  • Play
  • Stop
  • Repeat
  • --:--:--:--- / --:--:--:---
  • Noisy reverberant input signal at reference mic
    • Solo
  • Desired target signal
    • Solo
  • MWF output using oracle PSD with diffuse coherence matrix
    • Solo
  • MWF output using oracle PSD with reverberation coherence matrix
    • Solo
  • MWF output using blocking PSD least-squares
    • Solo
  • MWF output using blocking ML root
    • Solo
  • MWF output using blocking ML Newton
    • Solo
  • MWF output using LCMV diffuse beamformer
    • Solo
  • MWF output using joint ML Newton
    • Solo
  • MWF output using ML EM with theoretical diffuse coherence matrix
    • Solo
  • MWF output using ML EM with estimated coherence matrix
    • Solo
  • MWF output using PSD least-squares
    • Solo
  • MWF output using CDR estimator
    • Solo
  • MWF output using LRSV estimator
    • Solo
  • MWF output using CTF based estimator
    • Solo

Comparison between selected estimators without and with bias compensation (source distance 2.5 m):

Activate

  • Play
  • Stop
  • Repeat
  • --:--:--:--- / --:--:--:---
  • Noisy reverberant input signal at reference mic
    • Solo
  • Desired target signal
    • Solo
  • MWF output using oracle PSD with diffuse coherence matrix
    • Solo
  • MWF output using blocking PSD least-squares, no bias compensation
    • Solo
  • MWF output using blocking PSD least-squares, with bias compensation
    • Solo
  • MWF output using blocking ML root, with bias compensation
    • Solo
  • MWF output using blocking ML root, with bias compensation
    • Solo
  • MWF output using PSD least-squares, no bias compensation
    • Solo
  • MWF output using PSD least-squares, with bias compensation
    • Solo