Novel Mongolian speech synthesis method

• Main Authors: LANG JIAJUN, ZHANGBILIGETU, ZHANG WENJING, RENQINGDAOERJI, ZHANG QIAN, JI YATU, SA HEYA, LIU XINYUAN
• Format: Patent
• Language: Chinese; English
• Published: 24.12.2021
• Subjects: ACOUSTICS; MUSICAL INSTRUMENTS; PHYSICS; SPEECH ANALYSIS OR SYNTHESIS; SPEECH OR AUDIO CODING OR DECODING; SPEECH OR VOICE PROCESSING; SPEECH RECOGNITION

The invention discloses a novel Mongolian speech synthesis method, wherein the method specifically comprises the following steps: S1, processing a Mongolian word sequence based on BiLSTM: providing a Mongolian rhythm modeling method fusing a form vector and a phonetic system vector based on a BiLSTM neural network, wherein the Mongolian rhythm modeling method comprises an input layer, an attention layer, a BiLSTM layer and an output layer; and processing an input Mongolian word sequence, specifically, giving a word vector WE, a form vector ME and a phonetic system vector PE of a Mongolian word. The method relates to the technical field of speech synthesis. According to the novel Mongolian speech synthesis method, Mongolian rhythm modeling fusing the form vector and the phonetic system vector is provided based on the BiLSTM neural network, an input Mongolian word sequence is processed, characters are input by using a synthesizer to generate acoustic features, and then a vocoder is used for generating waveforms