Allophone-based speech synthesis system for Russian
System of speech synthesis
| First version: | 1.0 1997 | Last version: | 2.0 1999 |
| Application: | ||||
| The 2nd version is a prototype and is used only by its developers, for R&D and experiments | ||||
| Description: | ||||
| The system includes a dictionary of allophones and a linguistic processor. Modification of physical parameters (frequency, duration) of sound signals is made by means of a new technology, similar to PSOLA. The dictionary of allophones contains about 3200 units. These are all the allophones possible in Russian language. The allophones have been selected from words and logotoms produced by a professional speaker who had so-called "normative saint-peterburgian pronunciation". The lexicographic basis of the linguistic processor consists of about 110,000 words (it corresponds to about 4 million word forms). The algorithm of linguistic analysis is specially designed for allophonic synthesis. The linguistic processor provides conversion of any written text into its sound realisation. It consists of two units, each including independent parts. The first unit is a software tool converting a text into a chain of allophones selected from the dictionary. The second block is a software tool doing prosodic analysis of this chain. | ||||
| Languages processed: | ||||
| Russian | ||||
| Type of processing: | ||||
| Text transcription by means of a
lingustic processor. Use of the dictionary of allophones. Concatenation of allophones. |
||||
| Hardware/software requirements: | ||||
| IBM PC compatible, Windows 95/98 The system is implemented in C |
||||
| Distribution: | The 1st version is disributed by "Kometa" R&D centre for innovation introduction (Moscow) | |||
| Developer team: | Contact person: | |||
| Department for phonetics and methods of teaching foreign languages, SPbGU | Pavel A. Skrelin Phone: (+7 812) 328 95 65 E-mail: paul@PS1089.spb.edu, paul@phonet.lang.pu.ru |
|||