Date: June 6, 2016
Source: XU Weiqun from the CAS Key Laboratory of Speech Acoustics and Content Understanding
Prof. C. Julian Chen from the Department of Applied Physics and Applied Mathematics, Columbia University visited our institute on May 4. Chen gave a talk on pitch-synchronous speech parameterization method and its applications.
In the talk, Chen began with the traditional pitch asynchronous method and analyzed the shortcomings of traditional parameterization methods. The traditional methods, such as MFCC and PLP, use a fixed-length window to process the audio wave with 10-millisecond overlapping. They do not separate the pitch and timbre and therefore cannot be used to construct a high quality speech synthesizing system.
Chen proposed a "Timbron Theory" based on the analysis of large amount of audio data. He showed the advantages of the new theory with some vivid synthesized audio clips.
The talk was very engaging and attracted broad audience. After the talk, Chen and Prof. YAN Yonghong, host and director of the CAS Key Laboratory of Speech Acoustics and Content Understanding, further discussed scientific cooperation, hoping to make great efforts to promote academic exchanges and scientific cooperation in the future.