Paper details
- TITLE: Comparative evaluation of howling detection criteria in notch-filter-based howling suppression
- AUTHORS: Toon van Waterschoot and Marc Moonen
- ABSTRACT: Notch-filter-based howling suppression (NHS) is a popular approach to acoustic feedback control in public address and hands-free communication systems. The NHS method consists of two stages: howling detection and notch filter design. While the design of notch filters is well established, there is little agreement in the NHS literature on how the detection problem should be tackled. Moreover since the NHS literature consists mainly of patents, only few experimental results have been reported. A unifying framework for howling detection is provided and a comparative evaluation of existing and novel howling detection criteria is performed.
- STATUS: Published in J. Audio Eng. Soc., vol. 58, no. 11, Nov. 2010, pp. 923-940 (invited paper).
- REPORT NUMBER: 09-207
Downloads
- paper: PDF (1.3 MB)
- acoustic feedback path impulse responses:
- first impulse response (before acoustic feedback path change): WAV (44.1 kHz, 16 bit, 8.65 KB)
- second impulse response (after acoustic feedback path change): WAV (44.1 kHz, 16 bit, 8.65 KB)
- source signal: WAV (44.1 kHz, 16 bit, 5.04 MB)
- howling-compensated signals (see paper for definition of acronyms):
- FEP criterion:
- FEP threshold = 0.9: WAV (44.1 kHz, 16 bit, 5.04 MB)
- FEP threshold = 0.95: WAV (44.1 kHz, 16 bit, 5.04 MB)
- FEP threshold = 0.99: WAV (44.1 kHz, 16 bit, 5.04 MB)
- PHPR & IMSD criterion:
- PHPR threshold = 40 dB, IMSD threshold = 1 dB WAV (44.1 kHz, 16 bit, 5.04 MB)
- PHPR threshold = 42 dB, IMSD threshold = 1 dB WAV (44.1 kHz, 16 bit, 5.04 MB)
- PHPR threshold = 42 dB, IMSD threshold = 0.25 dB WAV (44.1 kHz, 16 bit, 5.04 MB) [Mind your ears! This sound file contains a severe howling component!]
- PHPR & PNPR & IMSD criterion:
- PHPR threshold = 36 dB, PNPR threshold = 12 dB, IMSD threshold = 0.5 dB WAV (44.1 kHz, 16 bit, 5.04 MB)
- PHPR threshold = 36 dB, PNPR threshold = 12 dB, IMSD threshold = 0.1 dB WAV (44.1 kHz, 16 bit, 5.04 MB)
- PHPR threshold = 44 dB, PNPR threshold = 12 dB, IMSD threshold = 0.1 dB WAV (44.1 kHz, 16 bit, 5.04 MB) [Mind your ears! This sound file contains a severe howling component!]
- FEP criterion:
BibTeX entry
@Article{vanwaterschoot09-207,
author = {T. van Waterschoot and M. Moonen},
title = {Comparative evaluation of howling detection criteria in notch-filter-based howling suppression},
journal = {J. Audio Eng. Soc.},
year = {2010},
volume = {58},
number = {11},
pages = {923--940},
month = Nov
}
Related papers [Acoustic feedback control]
- 15-48: A PEM-based frequency-domain Kalman filter for adaptive feedback cancellation
- 14-197: Feedback path characterization for a direct acoustic cochlear implant
- 12-214: Wiener variable step size and gradient spectral variance smoothing for double-talk-robust acoustic echo cancellation and acoustic feedback cancellation
- 12-163: Improved prediction error filters for adaptive feedback cancellation in hearing aids
- 12-112: Transform domain prediction error method for improved acoustic echo and feedback cancellation
- 10-57: Prediction-error-method-based adaptive feedback cancellation in hearing aids using pitch estimation
- 09-185: Adaptive feedback cancellation in hearing aids using a sinusoidal near-end signal model
- 09-58: Design and evaluation of digital signal processing algorithms for acoustic feedback and echo cancellation
- 09-43: Regularized adaptive notch filters for acoustic howling suppression
- 09-01: Assessing the acoustic feedback control performance of adaptive feedback cancellation in sound reinforcement systems
- 08-216: Comparative evaluation of howling detection criteria in notch-filter-based howling suppression
- 08-13: Fifty years of acoustic feedback control: state of the art and future challenges
- 08-IAP: Adaptive feedback cancellation for audio signals: an application of prediction error identification with cascaded noise models
- 07-177: Adaptive feedback cancellation for audio signals using a warped all-pole near-end signal model
- 07-49: Dually regularized recursive prediction error identification for acoustic feedback and echo cancellation
- 07-30: Adaptive feedback cancellation for audio applications
- 06-177: A pole-zero placement technique for designing second-order IIR parametric equalizer filters
- 06-117: MSE optimal regularization of APA and NLMS algorithms in room acoustic applications
- 06-81: Proactive notch filtering for acoustic feedback cancellation
- 05-258: Robust and efficient implementation of the PEM-AFROW algorithm for acoustic feedback cancellation
- 05-184: Optimally regularized adaptive filtering algorithms for room acoustic signal enhancement
- 05-156: Identification of undermodelled room impulse responses
- 05-71: Acoustic feedback cancellation for long acoustic paths using a nonstationary source model
- 05-14: Instrumental variable methods for acoustic feedback cancellation
- 04-151: Acoustic feedback cancellation for long acoustic paths using a nonstationary source model
- 04-24: On the performance of decorrelation by prefiltering for adaptive feedback cancellation in Public Address systems
- 03-ERNSI: Acoustic feedback cancellation in a closed-loop system identification framework
- 02-IAP: Design and evaluation of DSP algorithms for feedback cancellation in Public Address systems
- EP1675374: Circuit and method for estimating a room impulse response