Two methods for the analysis of the acoustic transmission of the respiratory system are presented. Continuous speech utterance is used as acoustic stimulation. The transmitted acoustic signal is recorded from various sites over the chest wall. The AR method analyzes the power spectral density function of the transmitted sound, which heavily depends on the microphone assembly and the utterance. The method was applied to a screening problem and was tested on a small database that consisted of 19 normal and five abnormal patients. Using the first five AR coefficients and the prediction error of an AR(10) model, as discriminating features, the system screened all abnormals. An ARMA method is suggested, which eliminates the dependence on microphone and utterance. In this method, the generalized least squares identification algorithm is used to estimate the ARMA transfer function of the respiratory system. The normal transfer function demonstrates a peak at the range of 130-250 Hz and sharp decrease in gain for higher frequencies. A pulmonary fibrotic patient demonstrated a peak at the same frequency range, a much higher gain in the high frequency range with an additional peak at about 700 Hz.

JiBSTRACTThe problem of Speech Recognition in a noisy environment is addressed. Particularly the mismatch problem originated when training a system in a l1cleantt environment and operating it in a noisy one. When measuring the similarity between a noisy test utterance and a list of clean templates a correction process, based on a series of Wiener filters built using the hypothesized clean template, is applied to the feature vectors of the noisy word .The filtering process is optimized as a by product of the Dynamic Programing algorithm of the scoring step. Tests were conducted on two data bases, one in Hebrew and the second in Japanese, using additive white and car noise at different SNRs. The method shows a very good performance and compares well with other methods proposed in the literature. . INTRODUCTIONThe performance of speech recognition systems designed to work in noise free conditions is strongly affected by the presence of noise. If the system has to be operated in different noise environments, training the system in one environment and operating it in a different one leads to a mismatch problem responsible for a poor performance.In contrast with other methods which use a speech enhancement step in order to input to the recognition system with noise reduced utterances, the key feature of the proposed method is doing the filtering at the scoring step using the information present in the clean templatesThe proposed method performs a feature correction on the noisy tests utterances in order to eliminate noise effects. The correcting mechanism is based on optimal filtering and on Dynamic Programming. The optimal filter at each state of the Dynamic programming is based on information present in each state. This correction is performed when computing the local distance between two feature vectors, one pertaining to a clean template and the other to a noisy test word.An estimate of the background noise and different template hypothesis are used to built the correcting filters. In this way the decisions are made using all the useful information present in the recognizer. The method was implemented using the system depicted in Figure 1. The system is trained only in a quiet environment and the background noise present in the operating environment (assumed to be stationary or slowly time variant) is learned in the neighborhoods of the words to be recognized. The Hypothesized Wiener Filtering (HWF) mechanism is performed during the DTW scoring step.

No abstract