The Infona portal uses cookies, i.e. strings of text saved by a browser on the user's device. The portal can access those files and use them to remember the user's data, such as their chosen settings (screen view, interface language, etc.), or their login data. By using the Infona portal the user accepts automatic saving and using this information for portal operation purposes. More information on the subject can be found in the Privacy Policy and Terms of Service. By closing this window the user confirms that they have read the information on cookie usage, and they accept the privacy policy and the way cookies are used by the portal. You can change the cookie settings in your browser.
Recently, the minimum mean squared error (MMSE) has been a benchmark of optimization criterion for deep neural network (DNN) based speech enhancement. In this study, a probabilistic learning framework to estimate the DNN parameters for single-channel speech enhancement is proposed. First, the statistical analysis shows that the prediction error vector at the DNN output well follows a unimodal density...
We first examine the generalization issue with the noise samples used in training nonlinear mapping functions between noisy and clean speech features for deep neural network (DNN) based speech enhancement. Then an empirical proof is established to explain why the DNN-based approach has a good noise generalization capability provided that a large collection of noise types are included in generating...
We present a joint noise and mask aware training strategy for deep neural network (DNN) based speech enhancement with sub-band features. First, based on the analysis of the previously proposed dynamic noise aware training approach tested on the wide-band (16 KHz) speech data, the full-band dynamic noise features cannot always improve the enhancement performance due to inaccurate noise estimation....
In this study, we explore long short-term memory recurrent neural networks (LSTM-RNNs) for speech enhancement. First, a regression LSTM-RNN approach for a direct mapping from the noisy to clean speech features is presented and verified to be more effective than deep neural network (DNN) based regression techniques in modeling long-term acoustic context. Then, a comprehensive comparison between the...
In this study, we investigate on the learning behaviors of DNN by explicit feature transformations. As a demonstration, linear and logarithm transformations, corresponding to the amplitude spectra and log-power spectra, are compared with the same minimum mean squared error (MMSE) objective function for optimizing DNN parameters. Based on the experimental analysis of the DNN learning behaviors, we...
This paper proposes a novel regression approach to binaural speech segregation based on deep neural network (DNN). In contrast to the conventional ideal binary mask (IBM) method using DNN with the interaural time difference (ITD) and in-teraural level difference (ILD) as the auditory features, the log-power spectra (LPS) features of target speech are directly predicted via a regression DNN model by...
Speech enhancement and speech separation are important frontends of many speech processing systems. In real tasks, the background noises are often mixed with some human voice interferences. In this paper, we explore a framework to unify speech enhancement and speech separation for a speaker-dependent scenario based on deep neural networks (DNNs). Using a supervised method, DNN is adopted to directly...
Based on the recently proposed speech pre-processing front-end with deep neural networks (DNNs), we first investigate different feature mapping directly from noisy speech via DNN for robust speech recognition. Next, we propose to jointly train a single DNN for both feature mapping and acoustic modeling. In the end, we show that the word error rate (WER) of the jointly trained system could be significantly...
In this paper, we present a synthesized stereo-based stochastic mapping approach for robust speech recognition. We extend the traditional stereo-based stochastic mapping (SSM) in two main aspects. First, the constraint of stereo-data, which is not practical in real applications, is relaxed by using HMM-based speech synthesis. Then we make feature mapping more focused on those incorrectly recognized...
When active learning is applied to real-world applications, human experts usually act as oracles to provide labels. However, human make mistakes, thus noise might be introduced during the learning process. Most previous studies simplify the problem by assuming uniformly-distributed noise over the sample space. Such assumption, however, might fail to precisely reflect the human experts' behaviour in...
In this paper, we present a novel approach to relax the constraint of stereo-data which is needed in a series of algorithms for noise-robust speech recognition. As a demonstration in SPLICE algorithm, we generate the pseudo-clean features to replace the ideal clean features from one of the stereo channels, by using HMM-based speech synthesis. Experimental results on aurora2 database show that the...
Set the date range to filter the displayed results. You can set a starting date, ending date or both. You can enter the dates manually or choose them from the calendar.