Speech recognition in reverberant and noisy environments employing multiple feature extractors and i-vector speaker adaptation

Abstract

The REVERB challenge provides a common framework for the evaluation of feature extraction techniques in the presence of both reverberation and additive background noise. State-of-the-art speech recognition systems perform well in controlled environments, but their performance degrades in realistic acoustical conditions, especially in real as well as simulated reverberant environments. In this contribution, we utilize multiple feature extractors including the conventional mel-filterbank, multi-taper spectrum estimation-based mel-filterbank, robust mel and compressive gammachirp filterbank, iterative deconvolution-based dereverberated mel-filterbank, and maximum likelihood inverse filtering-based dereverberated mel-frequency cepstral coefficient features for speech recognition with multi-condition training data. In order to improve speech recognition performance, we combine their results using ROVER (Recognizer Output Voting Error Reduction). For two-and eight-channel tasks, to get benefited from the multi-channel data, we also use ROVER, instead of the multi-microphone signal processing method, to reduce word error rate by selecting the best scoring word at each channel. As in a previous work, we also apply i-vector-based speaker adaptation which was found effective. In speech recognition task, speaker adaptation tries to reduce mismatch between the training and test speakers. Speech recognition experiments are conducted on the REVERB challenge 2014 corpora using the Kaldi recognizer. In our experiments, we use both utterance-based batch processing and full batch processing. In the single-channel task, full batch processing reduced word error rate (WER) from 10.0 to 9.3 % on SimData as compared to utterance-based batch processing. Using full batch processing, we obtained an average WER of 9.0 and 23.4 % on the SimData and RealData, respectively, for the two-channel task, whereas for the eight-channel task on the SimData and RealData, the average WERs found were 8.9 and 21.7 %, respectively.

Research on the Efficiency of Intelligent Algorithm for English Speech Recognition and Sentence Translatione

“…Therefore, this study used LSTM instead of RNN to encode the English text. LSTM [13] is also a kind of recurrent neural network algorithm. Compared with the traditional RNN, LSTM introduces the structural units of input gate, forgetting gate, and output gate to simulate the phenomena of deep impression and forgetting in the process of human brain memory, thus reducing the unimportant parts in the English text, highlighting the key points, reducing the amount of computation while enhancing the accuracy.…”

Speaker Adaptation Using i-Vector Based Clustering

“…The combined, 60-dimensional features are referred to as LDA+STC [15] and used as an input to the i-vector extractor. The GMM-UBM using full-covariance GMMs with 512 components is trained using Baum-Welch statistics extraction [18]. All the parameters of the trained GMM-UBM are converted into a single supervector, and reduced to 100 dimensional i-vectors using the i-vector extractor (the total variability matrix T).…”

Data Restoration by Linear Estimation of the Principal Components From Lossy Data

“…A variety of methods for the analysis of sensor data [1]- [4] and the extraction of meaningful patterns from these data have been proposed in recent decades [5]. Data collected by various sensors such as image, voice, electromyography (EMG) and chemical sensors are used for different applications such as image recognition [6]- [8], speech recognition [9], [10], gesture recognition [11]- [14] and gas classification [15]- [20]. The performance of classification techniques using sensor data varies greatly depending not only on the amount of data collected but also on the quality of the data.…”