基于DDBHMM的数字语音识别

数字语音识别具有很高的识别率，具有较高的实用价值。为实现在真实噪声环境下能达到高识别率的数字语音识别系统，采用基于段长分布的隐马尔可夫模型（DDBHMM）进行了安静环境和带噪环境下，特定人和非特定人的数字语音识别试验。试验结果表明，基于DDBHMM模型的数字语音识别技术对真实非平稳噪声环境下录制的特定人和非特定人语音都具有较高识别率。

Digital Speech Recognition Based on DDBHMM

Digital speech recognition technology has a very high recognition rate and is valuable in application. In order to implement a digit speech recognition system with a high recognition rate in real environment, experiments of speaker dependent and independent digit speech recognition based on DDBHMM （Duration Distribution Based on Hidden Markov Model） are carried out to test the performance of digit speech under the clear and non- stationary noisy environment respectively. Experiment results show that the algorithm based on the DDBHMM has a high recognition rate for speaker dependent and independent digit recognition on the speech database recorded under the real non-stationary noisy environments.