A Study on Improvements of Noise Rebustness for HMM-based Speech Recognition

基于HMM的语音识别抗噪性改进研究

• 批准号: 05680294
• 负责人: MATSUMOTO Hiroshi
• 金额: $ 1.15万
• 依托单位: Shinshu University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for General Scientific Research (C)
• 财政年份: 1993
• 资助国家: 日本
• 起止时间: 1993 至 1994
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-05680294/
• 关键词: Speech Recognition; Hidden Markov Model; Noisy Environments; Frequency Weighting; Euclidean Distance; Noise Robustness

In order to realize robust continuous density Hidden Markov Models (HMM) for noisy speech recognition, this study develops a frequency-weighted HMM based on the human auditory characteristics which is seseitive to formant peaks in high SNR frequency region. In this HMM,the covariance matrices of Gaussian probability density functions are fixed to the inverse of frequency weighting matrices in order to utilize the robustness of group delay spectra and also to incorporate their relative perceptual importance in frequency domain into HMM.Several frequency weighting functions and the scaling methods of frequency weighting matrices are examined using the international data base of NOISEX-92. The results of word recognition tests are summarized as follows.(1) The smoothed power spectrum derived from each mean vector gives the most robust HMM.(2) The optimum scaling to convert the weighting matrices to the covariance matrices is such that the sum of weighting coefficients is equal to one or the determinants of the converted covariances are 50 to 150 times larger than those of initial HMMs.(3) A larger number of states is required to attain the robustness in the frequency-weighted HMM.(4) Adaptive preemphasis improves the robustness to noises which have less energy in the high frequency region.(5) The frequency-weighted HMM attains SNR gains of 6 to 12 dB over a standard diagonal HMM for white, pink, and car noises.(6) Even when preprocessing the noisy speech by the standard noise reduction method of spectral subtraction, the frequency weighted HMM attains about 10% higher recognition scores in very low SNR condition than the conventional HMM.

为了实现用于噪声语音识别的鲁棒连续密度隐马尔可夫模型（HMM），本研究开发了一种基于人类听觉特征的频率加权隐马尔可夫模型，该模型对高信噪比频率区域的共振峰峰值敏感。在该HMM中，将高斯概率密度函数的协方差矩阵固定为频率加权矩阵的逆，以利用群延迟谱的鲁棒性，并将它们在频域中的相对感知重要性纳入HMM中。使用NOISEX-92国际数据库研究了几种频率加权函数和频率加权矩阵的缩放方法。单词识别测试的结果总结如下：（1）从每个均值向量导出的平滑功率谱给出了最稳健的HMM。（2）将权重矩阵转换为协方差矩阵的最佳缩放比例是使得权重系数之和等于1或者转换后的协方差的行列式比初始HMM大50到150倍。（3）更多的状态 需要获得频率加权 HMM 的鲁棒性。(4) 自适应预加重提高了对高频区域能量较少的噪声的鲁棒性。(5) 对于白噪声、粉红噪声和汽车噪声，频率加权 HMM 比标准对角线 HMM 获得了 6 至 12 dB 的 SNR 增益。(6) 即使通过标准频谱降噪方法对带噪语音进行预处理 通过减法，频率加权 HMM 在非常低的 SNR 条件下获得比传统 HMM 高约 10% 的识别分数。

项目成果

永山 亮、松本 弘: "音声認識における雑音付加HMMの自動生成" 日本音響学会講演論文集. 59-60 (1995)

Ryo Nagayama、Hiroshi Matsumoto：“自动生成用于语音识别的噪声 HMM”，日本声学学会会议记录 59-60 (1995)。

H.Matsumoto: "Robust speech recognition in noisy environments" Proc. of Int. Workshop on Human Interfase Technology. 1-8 (1994)

H.Matsumoto：“嘈杂环境中的鲁棒语音识别”Proc。

H.Matsumoto & Y.Nimura: "A Frequency-weighted continuous density HMM for noisy speech recognition" Proc.of Int.Conf.on Spoken Language Processing. 1007-1010 (1994)

松本

H.Masumoto: "Robust speech recognition in noisy environments" Proc.of Int.Workshop on Human Interface Technology. 1-8 (1994)

H.Masumoto：“嘈杂环境中的鲁棒语音识别”Proc.of Int.Workshop on Human Interface Technology。

H.Matsumoto: "Robust speech recognition in noisy environments" Proc.of Int. Workshop on Human Interface Technology. 1-8 (1994)

H.Matsumoto：“嘈杂环境中的鲁棒语音识别”Proc.of Int。