Optimising selection of units from speech databases for concatenative synthesis

By Alan W. Black, Nick Campbell. 1995

Concatenating units of natural speech is one method of speech synthesis 1 . Most such systems use an inventory of (cid:12)xed length units, typically diphones or triphones with one instance of each type. An alternative is to use more varied, non-uniform units extracted from large speech databases containing multiple instances of each. The greater variability in such natural speech segments allows closer modeling of naturalness and di(cid:11)erences in speaking styles, and eliminates the need for specially-recorded, single-use databases. However, with the greater variability comes the problem of how to select between the many instances of units in the database. This paper addresses that issue and presents a general method for unit selection.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

WORLD: A Vocoder-Based High-Quality Speech Synthesis System for Real-Time Applications (2016, background): “However, the speech synthesized by most of the conventional vocoder systems is inferior to that of waveform-based systems [4].”
Analysis of Speaker Adaptation Algorithms for HMM-Based Speech Synthesis and a Constrained SMAPLR Adaptation Algorithm (2009, background): “R ECENT unit-selection and concatenative approaches [1]–[4] produce high-quality synthetic speech, but require large-scale speech corpora if the speech is to sound”
An Overview of Voice Conversion and Its Challenges: From Statistical Modeling to Deep Learning (2020, background): “It is known to produce high speaker similarity and voice quality [77], [153],”
An HMM-based speech synthesis system applied to English (2002, methodology): “Figure 3 shows the widely-used unit selection scheme [16].”
In Other News: a Bi-style Text-to-speech Model for Synthesizing Newscaster Voice with Limited Data (2019, background): “For many years, concatenative-based speech synthesis (Black and Campbell, 1995; Taylor, 2006; Qian et al., 2013; Merritt et al., 2016; Wan et al., 2017) has been the industry standard.”
A survey on speech synthesis techniques in Indian languages (2020, background): “ui − 1, ui ) algorithms have been known to select segments from a place that results in less than ideal synthesis [104].”