Non-invasive estimate of blood glucose and blood pressure from a photoplethysmograph by means of machine learning techniques

In the process of speech recognition, it is especially crucial to precisely locate endpoints of the input utterance to be free of non-speech regions. This paper proposes a novel approach that finds robust features for endpoint detection in a noisy in-car environment. In the proposed method, we integrate both the widely used energy and entropy to form a new feature that possesses advantages of each individual while compensating for the drawback of each other. By monitoring the transition of the extracted new features, more precise endpoints could be found. Experiments in a real noisy environment, inside a Honda Civic car with background radio music and free chat, reveal an accuracy improvement which reached over 10% higher compared with a pure energy-based algorithm.

A novel approach to robust speech endpoint detection in car environments

Vibration analysis is widely used to reveal the fundamental cutting mechanics in machining condition monitoring. In this work, vibration signals generated in different chatter conditions as well as stable cutting are studied to understand chatter characteristics. Considering the nonlinear and non-stationary properties of chatter vibration in milling process, a self-adaptive analysis method named ensemble empirical mode decomposition (EEMD) is adopted to analyze vibration signals and two nonlinear indices are extracted as chatter indicators. Firstly, the vibration signal is preprocessed with a comb filter to eliminate the interference of rotation frequency, tooth passing frequency and their harmonics. Secondly, EEMD is applied to decompose the filtered signal into a set of intrinsic mode functions (IMFs). Sensitive IMFs containing rich chatter information are selected. With the development of chatter, an accumulation phenomenon appears in the spectrum of sensitive IMFs and chatter frequencies are modulated by the rotation frequency and tooth passing frequency. Finally, two nonlinear dimensionless indices within the range of [0, 1], i.e., C0 complexity and power spectral entropy, are extracted from the sensitive IMFs in both time domain and frequency domain. The proposed method is verified with well-designed cutting tests. It is found that, the stochastic noise dominates in the sensitive IMFs of stable cutting and both the C0 complexity value and power spectral entropy are the largest; with the increase of chatter severity level, the periodic chatter components dominate gradually and the proportion of stochastic noise decreases, and thus these two indicators decrease.

Chatter identification in end milling process based on EEMD and nonlinear dimensionless indicators

In speech processing, endpoint detection in noisy environments is difficult, especially in the presence of nonstationary noise. Robust endpoint detection is one of the most important areas of speech processing. Generally, the feature parameters used for endpoint detection are highly sensitive to the environment. Endpoint detection is severely degraded at low signal-to-noise ratios (SNRs) since those feature parameters cannot adequately describe the characteristics of a speech signal. As a result, this study seeks the banded structure on speech spectrogram to distinguish a speech from a nonspeech, especially in adverse environments. First, this study proposes a feature parameter, called band-partitioning spectral entropy (BSE), which exploits the use of the banded structure on speech spectrogram. A refined adaptive band selection (RABS) method is extended from the adaptive band selection method proposed by Wu et al., which adaptively selects useful bands not corrupted by noise. The successful RABS method is strongly depended on an on-line detection with minimal processing delay. In this paper, the RABS method is combined with the BSE parameter. Finally, a novel robust feature parameter, adaptive band-partitioning spectral entropy (ABSE), is presented to successfully detect endpoints in adverse environments. Experimental results indicate that the ABSE parameter is very effective under various noise conditions with several SNRs. Furthermore, the proposed algorithm outperforms other approaches and is reliable in a real car.

https://ir.nctu.edu.tw:443/bitstream/11536/13353/1/000231517400004.pdf

Robust endpoint detection algorithm based on the adaptive band-partitioning spectral entropy in adverse environments

A review on speaker diarization systems and approaches

This paper presents an entropy-based algorithm for accurate and robust endpoint detection for speech recognition under noisy environments. Instead of using the conventional energy-based features, the spectral entropy is developed to identify the speech segments accurately. Experimental results show that this algorithm outperforms the energy-based algorithms in both detection accuracy and recognition performance under noisy environments, with an average error rate reduction of more than 16%.

/pdf/robust-entropy-based-endpoint-detection-for-speech-19dgg90lt0.pdf

Robust entropy-based endpoint detection for speech recognition in noisy environments.

Although Mel-frequency cepstral coefficients (MFCC) have been proven to perform very well under most conditions, some limited efforts have been made in optimizing the shape of the filters in the filter-bank in the conventional MFCC approach. This paper presents a new feature extraction approach that designs the shapes of the filters in the filter-bank. In this new approach, the filter-bank coefficients are data-driven and obtained by applying principal component analysis (PCA) to the FFT spectrum of the training data. The experimental results show that this method is robust under noisy environment and is well additive with other noise-handling techniques.

Improved MFCC feature extraction by PCA-optimized filter-bank for speech recognition

Previous studies indicate that noise and speaker variations can degrade the performance of deep-learning–based speech– enhancement systems. To increase the system performance over environmental variations, we propose a novel speaker-aware system that integrates a deep denoising autoencoder (DDAE) with an embedded speaker identity. The overall system first extracts embedded speaker identity features using a neural network model; then the DDAE takes the augmented features as input to generate enhanced spectra. With the additional embedded features, the speech-enhancement system can be guided to generate the optimal output corresponding to the speaker identity. We tested the proposed speech-enhancement system on the TIMIT dataset. Experimental results showed that the proposed speech-enhancement system could improve the sound quality and intelligibility of speech signals from additive noisecorrupted utterances. In addition, the results suggested system robustness for unseen speakers when combined with speaker features.

Speaker-aware Deep Denoising Autoencoder with Embedded Speaker Identity for Speech Enhancement

In this paper, we proposed an algorithm used to improve the performance of the metric-based segmentation techniques, by which the segmentation points are found at maxima of a distance measured between two contiguous windows shifted along the stream of speech features. In our proposed method, the PCA processes are first performed on the speech features to obtain more robust features, and then the above metric-based segmentation was applied on the PCA-derived features to decide the segmentation points. Experiment results show that our proposed method can efficiently improve the detection rates of the segmentation points up to 7% while the false alarm rates remain unchanged.

/pdf/automatic-metric-based-speech-segmentation-for-broadcast-4kagthx3fr.pdf

Automatic metric-based speech segmentation for broadcast news via principal component analysis.

Parallel model combination (PMC) techniques have been very successful and popularly used in many applications to improve the performance of speech recognition systems under noisy environments. However, it is believed that some assumptions and approximations made in this approach, primarily in the domain transformation and parameter combination processes, are not necessarily accurate enough in certain practical situations, which may degrade the achievable performance of PMC. In this paper, the possible sources that cause the performance degradation in these processes are carefully analyzed and discussed. Three new approaches, including the truncated Gaussian approach and the split mixture approach for the domain transformation process and the estimated cross-term approach for parameter combination process, are proposed in this paper in order to handle these problems, minimize such degradation, and improve the accuracy of the PMC techniques. These proposed approaches were analyzed and discussed with two recognition tasks, one relatively simple, and the other more complicated and realistic. Both sets of experiments showed that these proposed approaches are able to provide significant improvements over the original PMC method, especially when the SNR condition is worse.

Jeih-weih Hung

Papers

Robust entropy-based endpoint detection for speech recognition in noisy environments.

Improved MFCC feature extraction by PCA-optimized filter-bank for speech recognition

Speaker-aware Deep Denoising Autoencoder with Embedded Speaker Identity for Speech Enhancement

Automatic metric-based speech segmentation for broadcast news via principal component analysis.

New approaches for domain transformation and parameter combination for improved accuracy in parallel model combination (PMC) techniques