Unit Selection in Concatenative TTS Synthesis Systems Based on Mel Filter Bank Amplitudes and Phonetic Context

T. Lambert; Andrew P. Breen; Barry Eggleton; Stephen J. Cox; Ben P. Milner

Unit Selection in Concatenative TTS Synthesis Systems Based on Mel Filter Bank Amplitudes and Phonetic Context

T. Lambert, Andrew P. Breen, Barry Eggleton, Stephen J. Cox, Ben P. Milner

Research output: Contribution to conference › Paper

3 Citations (Scopus)

Abstract

In concatenative text-to-speech (TTS) synthesis systems unit selection aims to reduce the number of concatenation points in the synthesized speech and make concatenation joins as smooth as possible. This research considers synthesis of completely new utterances from non-uniform units, whereby the most appropriate units, according to acoustic and phonetic criteria, are selected from a myriad of similar speech database candidates. A Viterbi-style algorithm dynamically selects the most suitable database units from a large speech database by considering concatenation and target costs. Concatenation costs are derived from mel filter bank amplitudes, whereas target costs are considered in terms of the phonemic and phonetic properties of required units. Within subjects and between subjects ANOVA [9] evaluation of listeners' scores showed that the TTS system with this method of unit selection was preferred in 52% of test sentences.

Original language	English
Pages	273-276
Number of pages	4
Publication status	Published - Sept 2003
Event	8th European Conference on Speech Communication and Technology - Geneva, Switzerland Duration: 1 Sept 2003 → 4 Sept 2003

Conference

Conference	8th European Conference on Speech Communication and Technology
Abbreviated title	EUROSPEECH 2003
Country/Territory	Switzerland
City	Geneva
Period	1/09/03 → 4/09/03

Cite this

@conference{ff82222a2d3645318d67e589389e6208,

title = "Unit Selection in Concatenative TTS Synthesis Systems Based on Mel Filter Bank Amplitudes and Phonetic Context",

abstract = "In concatenative text-to-speech (TTS) synthesis systems unit selection aims to reduce the number of concatenation points in the synthesized speech and make concatenation joins as smooth as possible. This research considers synthesis of completely new utterances from non-uniform units, whereby the most appropriate units, according to acoustic and phonetic criteria, are selected from a myriad of similar speech database candidates. A Viterbi-style algorithm dynamically selects the most suitable database units from a large speech database by considering concatenation and target costs. Concatenation costs are derived from mel filter bank amplitudes, whereas target costs are considered in terms of the phonemic and phonetic properties of required units. Within subjects and between subjects ANOVA [9] evaluation of listeners' scores showed that the TTS system with this method of unit selection was preferred in 52% of test sentences.",

author = "T. Lambert and Breen, {Andrew P.} and Barry Eggleton and Cox, {Stephen J.} and Milner, {Ben P.}",

year = "2003",

month = sep,

language = "English",

pages = "273--276",

note = "8th European Conference on Speech Communication and Technology, EUROSPEECH 2003 ; Conference date: 01-09-2003 Through 04-09-2003",

}

TY - CONF

T1 - Unit Selection in Concatenative TTS Synthesis Systems Based on Mel Filter Bank Amplitudes and Phonetic Context

AU - Lambert, T.

AU - Breen, Andrew P.

AU - Eggleton, Barry

AU - Cox, Stephen J.

AU - Milner, Ben P.

PY - 2003/9

Y1 - 2003/9

N2 - In concatenative text-to-speech (TTS) synthesis systems unit selection aims to reduce the number of concatenation points in the synthesized speech and make concatenation joins as smooth as possible. This research considers synthesis of completely new utterances from non-uniform units, whereby the most appropriate units, according to acoustic and phonetic criteria, are selected from a myriad of similar speech database candidates. A Viterbi-style algorithm dynamically selects the most suitable database units from a large speech database by considering concatenation and target costs. Concatenation costs are derived from mel filter bank amplitudes, whereas target costs are considered in terms of the phonemic and phonetic properties of required units. Within subjects and between subjects ANOVA [9] evaluation of listeners' scores showed that the TTS system with this method of unit selection was preferred in 52% of test sentences.

AB - In concatenative text-to-speech (TTS) synthesis systems unit selection aims to reduce the number of concatenation points in the synthesized speech and make concatenation joins as smooth as possible. This research considers synthesis of completely new utterances from non-uniform units, whereby the most appropriate units, according to acoustic and phonetic criteria, are selected from a myriad of similar speech database candidates. A Viterbi-style algorithm dynamically selects the most suitable database units from a large speech database by considering concatenation and target costs. Concatenation costs are derived from mel filter bank amplitudes, whereas target costs are considered in terms of the phonemic and phonetic properties of required units. Within subjects and between subjects ANOVA [9] evaluation of listeners' scores showed that the TTS system with this method of unit selection was preferred in 52% of test sentences.

M3 - Paper

SP - 273

EP - 276

T2 - 8th European Conference on Speech Communication and Technology

Y2 - 1 September 2003 through 4 September 2003

ER -