Analysis of Data Collected in Listening Tests for the Purpose of Evaluation of Concatenation Cost Functions
In this paper we present an analysis of data, which were collected in listening tests, and are planned to be used for the development and evaluation of concatenation cost functions for unit selection based TTS systems. The aim of the analysis was to evaluate a “richness” of the collected data with respect to the intended utilization. No effort was made to propose a new method for measuring concatenation artifacts. The study was limited to two speakers (male and female), and five short Czech vowels as these sounds are characterized by being highly energetic and having rich spectral content, which induces complexity and wide range of possible discontinuities at concatenation points.
Keywordsspeech synthesis speech analysis unit selection concatenation cost data collection
Unable to display preview. Download preview PDF.
- 1.Hunt, A., Black, A.: Unit selection in a concatenative speech synthesis system using a large speech database. In: ICASSP 1996, Atlanta, Georgia, vol. 1, pp. 373–376 (1996)Google Scholar
- 2.Vepa, J.: Join cost for unit selection speech synthesis. Ph.D. thesis, University of Edinburgh (2004)Google Scholar
- 3.Stylianou, Y., Syrdal, A.K.: Perceptual and objective detection of discontinuities in concatenative speech synthesis. In: ICASSP 2001, Salt Lake City, Utah, vol. 2, pp. 837–840 (2001)Google Scholar
- 5.Bellegarda, J.R.: A novel discontinuity metric for unit selection text-to-speech synthesis. In: SSW5 2004, Pittsburgh, PA, USA, pp. 133–138 (2004)Google Scholar
- 9.Legát, M., Matoušek, J., Tihelka, D.: A robust multi-phase pitch-mark detection algorithm. In: INTERSPEECH 2007, Antwerp, Belgium, vol. 1, pp. 1641–1644 (2007)Google Scholar