| Speech synthesis technology can convert the information such as words, signals, or numbers produced by itself or inputted from outside to continuous speech signalsin terms of output. This technology is also called Text To Speech (TTS), whichinvolves many areas, such as computer science, linguistics, phonetics, sound signalprocessing and psychology.Mongolian language is the official language of Inner Mongolia AutonomousRegion, and above six million Mongolian uses it in daily life. So the research is worth, and the meaning is great to Mongolian's information industry. Therefore, this paperhas a major discussion on how to create a comparatively natural and widely-applicable speech synthesis system based on the features of the Mongolian languageand phonetics. We introduce in the following aspects:1. Based on the analysis of the Mongolian phonetic structure as well as thephonetics, we chose the etyma and the affix of Mongolian as the basic unit for thisspeech synthesis. According to Mongolian word formation rules, we summarizedMongolian configuration affix from the massive corpus. And we solved the problem of the separation of the etyma and the affix in the text analysis process, thus carriedon the retrieval in the speech database, found corresponding the speech data.2. By having a general analysis and statistics of the phonetic rhythm parametersin the natural speech flow, we summarized the rhythm transformation rules of theMongolian language, such as time-based transformation rules, the stresstransformation rules, pause rules and tone changes in the four types of sentencestructures in the Mongolian language, and the influence of these sentence structureson the rhythm of the speech flow.3. Using TD-PSOLA algorithm and FD-PSOLA algorithm, we realized theadjustment for the rhythm of the synthesized speech, which greatly improved thenaturalness of the synthesized speech of the Mongolian language. At the same timewe used the algorithms of the direct concatenation and the soft concatenation toconcatenate the speech units, in order to achieve the better synthesis effect. |
