CAJ | 학술논문

设计一款嵌入式语音识别系统，该系统硬件平台以凌阳公司的SPCE061A为核心。采用离散隐马尔科夫模型识别算法对非特定人的孤立词语音识别。实验结果表明，该系统对非特定人孤立词识别率达90%以上，性价比高，可应用于许多场合，有一定的市场前景。文章阐述了该系统各个电路模块及软件流程图的设计。
설계일관감입식어음식별계통，해계통경건평태이릉양공사적SPCE061A위핵심。채용리산은마이과부모형식별산법대비특정인적고립사어음식별。실험결과표명，해계통대비특정인고립사식별솔체90%이상，성개비고，가응용우허다장합，유일정적시장전경。문장천술료해계통각개전로모괴급연건류정도적설계。
We design an embedded speech recognition system. The system was centered SPCE061A of Sunpluse company. It was realization of speaker-independent’s isolated word recognition through DHMM. The experiment confirms that its speech recognition accuracy reaches 90 percent for speaker-independent and short vocabulary. The system has advantages of high rate of quantity and price. It can be used in some special situations and has much marker potential. We also elaborate the circuit modules and software flow diagram of the system.