A robust method for speech signal time-delay estimation in reverberant rooms

Brandstein, M.S.; Silverman, H.F.

doi:10.1109/icassp.1997.599651

Cited by 303 publications

(225 citation statements)

References 9 publications

Supporting

Mentioning

211

Contrasting

Unclassified

Order By: Relevance

“…The performance of the GCC approach is dependent on the design of the prefilters whose main function is to improve the accuracy of peak detection. We employed the phase transform (PHAT) prefilter which has shown to give satisfactory TDOA estimates in the presence of reverberation [13]. The estimated time delay between the ith and jth channel is given by τ = arg max…”

Section: The Gcc Algorithmmentioning

confidence: 99%

A Practical Adaptive Blind Multichannel Estimation Algorithm with Application to Acoustic Impulse Responses

Ahmad

Khong

Naylor

2007

2007 15th International Conference on Digital Signal Processing

View full text Add to dashboard Cite

We propose a noise robust adaptive blind multichannel identification algorithm for acoustic impulse responses. It has been known that the normalized multichannel frequency domain least-meansquare (NMCFLMS) algorithm misconverges under low signal-tonoise ratio. The coefficients of NMCFLMS converge initially toward the true impulse response after which they then misconverge. The extended NMCFLMS (ext-NMCFLMS) algorithm which has been proposed to mitigate this misconvergence problem assumes the knowledge of magnitude and time-differences-of-arrival (TDOA) of the direct paths for the acoustic impulse responses. In this work, we show how the TDOA can be obtained. More importantly, we present a novel approach to estimate the magnitude of the direct path component under practical conditions. We then show how these estimates can be incorporated to the proposed ext-NMCFLMS with direct path estimation algorithm. We analyze how errors in these estimates affect the performance of the proposed algorithm.

show abstract

Section: The Gcc Algorithmmentioning

confidence: 99%

A Practical Adaptive Blind Multichannel Estimation Algorithm with Application to Acoustic Impulse Responses

Ahmad

Khong

Naylor

2007

2007 15th International Conference on Digital Signal Processing

View full text Add to dashboard Cite

show abstract

“…2 because in speaker diarization we do not know the number of speakers or their locations. Therefore we use a modified version of the Generalized Cross Correlation (GCC) called "generalized cross correlation with phase transform" (GCC-PHAT) (see [7]). …”

Section: Tdoa Estimation Via Gcc-phatmentioning

confidence: 99%

“…To obtain the Time Delay of Arrival (TDOA) for each segment of a channel, the GCC-PHAT ( [7]) is computed between the segment and the corresponding segment in the reference channel. Such a measure is more robust and accurate than cross-correlation when the noise level is low and it outputs values normalized from 0 to 1.…”

Section: Robust Tdoa Estimationmentioning

confidence: 99%

“…This has the effect of improving the overall quality of the resulting signal with respect to the individual channels, regardless of the location of the speaker. To perform D&S, we use a scrolling window through the signal and compute the Time Delay of Arrival (TDOA) of each window using GCC-PHAT ( [7]). Two different filtering techniques are employed to smooth the computed TDOA to avoid instabilities due to overlapped speech, silence segments, or degraded channels.…”

Section: Fig 1 Delay-and-sum Systemmentioning

confidence: 99%

See 1 more Smart Citation

Speaker diarization for multi-party meetings using acoustic fusion

Anguera

Woofers

Hernando³

2005

IEEE Workshop on Automatic Speech Recognition and Understanding, 2005.

View full text Add to dashboard Cite

One of the sub-tasks of the Spring 2004 and Spring 2005NIST Meetings evaluations requires segmenting multi-party meetings into speaker-homogeneous regions using data from multiple distant microphones (the "MDM" sub-task). One approach to this task is to run a speaker segmentation system on each of the microphone channels separately, and then merge the results. This can be thought of as a many-to-one post-processing approach. In this paper we propose an alternative approach in which we use delay-andsum beamforming techniques to fuse the signals from each of the multiple distant microphones into a single enhanced signal. This approach can be thought of a many-to-one preprocessing approach. In the pre-processing approach we propose, the time delay of arrival (TDOA) between each of the multiple distant channels and a reference channel is computed incrementally using a window that steps through the signals from each of the multiple microphones. No information about the locations or setup of the microphones is required. Using the TDOA information, the channels are first aligned and then summed and the resulting "enhanced" signal is clustered using our standard speaker diarization system. We test our approach on the 2004 and 2005 NIST meetings evaluation databases and show that the technique performs very well.

show abstract

“…We used a modified version of the Generalized Cross Correlation with phase transform (GCC PHAT (f)) (see [11]) and estimate the delays between microphones with the following formula:…”

Section: Introductionmentioning

confidence: 99%

Speaker Diarization for Multi-microphone Meetings Using Only Between-Channel Differences

Pardo

Anguera

Wooters

2006

Machine Learning for Multimodal Interaction

View full text Add to dashboard Cite

Abstract. We present a method to extract speaker turn segmentation from multiple distant microphones (MDM) using only delay values found via a crosscorrelation between the available channels. The method is robust against the number of speakers (which is unknown to the system), the number of channels, and the acoustics of the room. The delays between channels are processed and clustered to obtain a segmentation hypothesis. We have obtained a 31.2% diarization error rate (DER) for the NIST´s RT05s MDM conference room evaluation set. For a MDM subset of NIST´s RT04s development set, we have obtained 36.93% DER and 35.73% DER * . Comparing those results with the ones presented by Ellis and Liu [8], who also used between-channels differences for the same data, we have obtained 43% relative improvement in the error rate.

show abstract

A robust method for speech signal time-delay estimation in reverberant rooms

Cited by 303 publications

References 9 publications

A Practical Adaptive Blind Multichannel Estimation Algorithm with Application to Acoustic Impulse Responses

A Practical Adaptive Blind Multichannel Estimation Algorithm with Application to Acoustic Impulse Responses

Speaker diarization for multi-party meetings using acoustic fusion

Speaker Diarization for Multi-microphone Meetings Using Only Between-Channel Differences

Contact Info

Product

Resources

About