Eigenvoice speaker adaptation with minimal data for statistical speech synthesis systems using a MAP approach and nearest-neighbors

Name: Eigenvoice speaker adaptation with minimal data for statistical speech synthesis systems using a MAP approach and nearest-neighbors
Author: Mohammadi, Amir, Sarfjoo, Seyyed Saeed, Demiroğlu, Cenk

İsim	Eigenvoice speaker adaptation with minimal data for statistical speech synthesis systems using a MAP approach and nearest-neighbors
Yazar	Mohammadi, Amir, Sarfjoo, Seyyed Saeed, Demiroğlu, Cenk
Basım Tarihi:	2014-12
Basım Yeri	- IEEE
Konu	Cluster adaptive training, Eigenvoice adaptation, Nearest neighbor, Speaker adaptation, Statistical speech synthesis
Tür	Süreli Yayın
Dil	İngilizce
Dijital	Evet
Yazma	Hayır
Kütüphane:	Özyeğin Üniversitesi
Demirbaş Numarası	2329-9290
Kayıt Numarası	381836f7-5637-4ed5-aa56-4a55290fcaf5
Lokasyon	Electrical & Electronics Engineering
Tarih	2014-12
Notlar	Due to copyright restrictions, the access to the full text of this article is only available via subscription.
Örnek Metin	Statistical speech synthesis (SSS) systems have the ability to adapt to a target speaker with a couple of minutes of adaptation data. Developing adaptation algorithms to further reduce the number of adaptation utterances to a few seconds of data can have substantial effect on the deployment of the technology in real-life applications such as consumer electronics devices. The traditional way to achieve such rapid adaptation is the eigenvoice technique which works well in speech recognition but known to generate perceptual artifacts in statistical speech synthesis. Here, we propose three methods to alleviate the quality problems of the baseline eigenvoice adaptation algorithm while allowing speaker adaptation with minimal data. Our first method is based on using a Bayesian eigenvoice approach for constraining the adaptation algorithm to move in realistic directions in the speaker space to reduce artifacts. Our second method is based on finding pre-trained reference speakers that are close to the target speaker and utilizing only those reference speaker models in a second eigenvoice adaptation iteration. Both techniques performed significantly better than the baseline eigenvoice method in the objective tests. Similarly, they both improved the speech quality in subjective tests compared to the baseline eigenvoice method. In the third method, tandem use of the proposed eigenvoice method with a state-of-the-art linear regression based adaptation technique is found to improve adaptation of excitation features.
DOI	10.1109/TASLP.2014.2362009
Cilt	22

Kaynağa git Özyeğin Üniversitesi

Aramaya Dön

Özyeğin Üniversitesi

Kaynağa git

Eigenvoice speaker adaptation with minimal data for statistical speech synthesis systems using a MAP approach and nearest-neighbors

Yazar Mohammadi, Amir, Sarfjoo, Seyyed Saeed, Demiroğlu, Cenk

Basım Tarihi 2014-12

Basım Yeri - IEEE

Konu Cluster adaptive training, Eigenvoice adaptation, Nearest neighbor, Speaker adaptation, Statistical speech synthesis

Tür Süreli Yayın

Dil İngilizce

Dijital Evet

Yazma Hayır

Kütüphane Özyeğin Üniversitesi

Demirbaş Numarası 2329-9290

Kayıt Numarası 381836f7-5637-4ed5-aa56-4a55290fcaf5

Lokasyon Electrical & Electronics Engineering

Tarih 2014-12

Notlar Due to copyright restrictions, the access to the full text of this article is only available via subscription.

Örnek Metin Statistical speech synthesis (SSS) systems have the ability to adapt to a target speaker with a couple of minutes of adaptation data. Developing adaptation algorithms to further reduce the number of adaptation utterances to a few seconds of data can have substantial effect on the deployment of the technology in real-life applications such as consumer electronics devices. The traditional way to achieve such rapid adaptation is the eigenvoice technique which works well in speech recognition but known to generate perceptual artifacts in statistical speech synthesis. Here, we propose three methods to alleviate the quality problems of the baseline eigenvoice adaptation algorithm while allowing speaker adaptation with minimal data. Our first method is based on using a Bayesian eigenvoice approach for constraining the adaptation algorithm to move in realistic directions in the speaker space to reduce artifacts. Our second method is based on finding pre-trained reference speakers that are close to the target speaker and utilizing only those reference speaker models in a second eigenvoice adaptation iteration. Both techniques performed significantly better than the baseline eigenvoice method in the objective tests. Similarly, they both improved the speech quality in subjective tests compared to the baseline eigenvoice method. In the third method, tandem use of the proposed eigenvoice method with a state-of-the-art linear regression based adaptation technique is found to improve adaptation of excitation features.

DOI 10.1109/TASLP.2014.2362009

Cilt 22