欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为了减少在噪声环境下测试条件与训练条件不匹配导致的语音识别性能下降，提出了一种结合多频带谱减法的抗噪声语音识别系统。首先提取带噪语音的前几帧作为估计的噪声信号，将带噪语音、估计的噪声信号按频率划分肘个互不相交的频带，然后根据每个频带内带噪语音与估计噪声信号的信噪比，来确定该频带噪声的谱减参数。语音增强作为前端处理，与语音识别器级连构成抗噪声语音识别系统。通过实验仿真表明，基于多频带谱减法的抗噪声语音识别系统在不同信噪比、不同类型的噪声下，识别性能明显优于基本谱减法。
英文摘要：
In order to reduce the degradation of the speech recognition accuracy while the testing condition are mismatched with the training condition around noisy environment, one kind of multi-band spectral subtraction is put forward. The estimated noise signals are extracted from the first few frames of the noisy speech, and the noisy speech and estimation of noisy signals by the frequency were divided into non-interacted M frequency bands, and then according to the SNR （signal-to-noise ratio） of noisy speech in each frequency band to determine the band noise spectral subtraction parameters. Speech enhancement as the front-end processing module and concatenate with speech recognizer constitute robust speech recognition system. The results of simulated experiments indicate that the recognition accuracy of multi-band spectral subtraction robust speech recognition system is obviously superior to the basic spectral subtraction in different signal-to-noise ratios and different noise＇ s types.
同期刊论文项目
同项目期刊论文
期刊信息
位置：