ARM Cortex-M3-Based Speech Synthesis Software System Design And Implementation

With the rise of the mobile Internet era, the application of embedded speech synthesis technology is more and more attention. HMM-based speech synthesis technology is being widely used, however, due to the large corpus system needs to take up a lot of system resources, large amount of computation and other reasons, so when used in embedded systems because of insufficient system resources and cause the response is not timely and not smooth voice and other issues.In response to these problems, this paper first describes in detail the history and current status of speech synthesis technology, and then analyzes the HMM algorithm principle, HMM corpus design methods, HMM unit selection method, LZSS compression algorithm and waveform stitching algorithm TD-PSOLA, then the speech synthesis system for embedded systems do a needs analysis, and based on actual market demand and functional requirements, design and implement the Arm Cortex-M3core-based embedded systems speech synthesis system, the system makes full use of the Arm Cortex-M3 core features to solve the speech synthesis technology requires high computational efficiency and high data transmission requirements, In order to support parallel processing text analysis module, HMM unit selection module and waveform splice module, the system uses a multi-task scheduling RTOS, and rely on real-time multi-tasking operating system, time-sharing scheduling features, the system uses a pipelined parallel processing method significantly improves the speed of speech synthesis and quality, both to meet the embedded speech synthesis system for converting text to speech rhythm correctness and reasonableness requirements, while also taking into account the real-time playback, smooth user experience needs. Through the design and implementation of this system, speech synthesis technology applications in the field of embedded products has been expanded, the system is ideally suited for the current low-end mobile devices for speech synthesis technology to market demand.