Title :
The effects of whispered speech on state-of-the-art voice based biometrics systems
Author :
Sarria-Paja, Milton ; Senoussaoui, Mohammed ; Falk, Tiago H.
Author_Institution :
Inst. Nat. de la Rech. Sci., Univ. of Quebec, Montréal, QC, Canada
Abstract :
In this paper, automatic speaker verification using whispered speech is explored. In the past, whispered speech has been shown to convey relevant speaker identity and gender information, nevertheless it is not clear how to efficiently use this information in speech-based biometric systems. This study compares the performance of three different speaker verification systems trained and tested under different scenarios and with two different feature representations. First, we show the benefits of using AM-FM based features as well as their effectiveness for i-vectors extraction. Second, for the classical mel-frequency cepstral coefficient (MFCC) features we show that gains of up to 40% could be achieved with the fusion of traditional Gaussian mixture model (GMM) based systems and more recent i-vector based ones, relative to using either system alone for normal speech. Additionally, for MFCC, fusion schemes were shown to be more robust to addition of whispered speech data during training or enrollment. Overall, AM-FM based features were shown to be more robust to varying training/testing conditions and to improve speaker verification performance for both normal and whispered speech by using the GMM based system alone.
Keywords :
Gaussian processes; biometrics (access control); cepstral analysis; feature extraction; gender issues; mixture models; sensor fusion; speaker recognition; AM-FM based features; GMM based systems; Gaussian mixture model based system; MFCC features; automatic speaker verification; classical Mel-frequency cepstral coefficient features; feature representation; fusion schemes; gender information; i-vector extraction; speaker identity; speech-based biometric systems; voice based biometrics system; whispered speech; Databases; Error analysis; Feature extraction; Mel frequency cepstral coefficient; Speech; Training; GMM; Whispered speech; i-vectors; modulation features; speaker verification;
Conference_Titel :
Electrical and Computer Engineering (CCECE), 2015 IEEE 28th Canadian Conference on
Conference_Location :
Halifax, NS
Print_ISBN :
978-1-4799-5827-6
DOI :
10.1109/CCECE.2015.7129458