Publication Details
A Unified Approach for Audio Characterization and its Application to Speaker Recognition
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Plchot Oldřich, Ing., Ph.D. (DCGM)
Scheffer Nicolas
audio characterization, speaker recognition, i-vector, calibration metadata
The technique proposed in this work allows for extracting a very low-dimensional vector encoding information about chosen characteristics of audio signal such as: type and level of background noise, reverberation and transmission channel. The reported experimental shows that such information can be very useful for calibration and fusion of speaker verification systems.
Systems designed to solve speech processing tasks like speech or speaker recognition, language identification, or emotion detection are known to be affected by the recording conditions of the acoustic signal, like the channel, background noise, reverberation, and so on. Knowledge of the nuisance characteristics present in the signal can be used to improve performance of the system. In some cases, the nature of these nuisance characteristics is known a priori, but in most practical cases it is not. Most approaches used to automatically detect the characteristics of a signal are designed for a specific type of effect: noise, reverberation, language, type of channel, and so on. We propose a method for detecting the audio characteristics of a signal in a unified way, based on iVectors. We show results for the detector itself and for its use as metadata during calibration of a state-ofthe- art speaker recognition system based on iVectors extracted from Mel frequency cepstral coefficients. Results show relative gains in equal error rate of up to 15% in a variety of recording conditions.
@inproceedings{BUT96992,
author="Luciana {Ferrer} and Lukáš {Burget} and Oldřich {Plchot} and Nicolas {Scheffer}",
title="A Unified Approach for Audio Characterization and its Application to Speaker Recognition",
booktitle="Proceedings of Odyssey 2012, The Speaker and Language Recognition Workshop",
year="2012",
pages="317--323",
publisher="International Speech Communication Association",
address="Singapur",
isbn="978-981-07-3093-2",
url="http://www.fit.vutbr.cz/research/groups/speech/publi/2012/ferrer_odyssey2012_317-323-59.pdf"
}