Maximum Likelihood Linear Regression Research Articles

Linear regression for Poisson count data: a new semi-analytical method with applications to COVID-19 events

This study presents the application of a new semi-analytical method of linear regression for Poisson count data to COVID-19 events. The regression is based on the maximum-likelihood solution for the best-fit parameters presented in an earlier publication, and this study introduces a simple analytical solution for the covariance matrix that completes the problem of linear regression with Poisson data for one independent variable. The analytical nature of both parameter estimates and their covariance matrix is made possible by a convenient factorization of the linear model proposed by J. Scargle. The method makes use of the asymptotic properties of the Fisher information matrix, whose inverse provides the covariance matrix. The combination of simple analytical methods to obtain both the maximum-likelihood estimates of the parameters and their covariance matrix constitutes a new and convenient method for the linear regression of Poisson-distributed count data, which are of common occurrence across a variety of fields. A comparison between this maximum-likelihood linear regression method for Poisson data and two alternative methods often used for the regression of count data—the ordinary least–square regression and the χ2 regression—is provided with the application of these methods to the analysis of recent COVID-19 count data. The study also discusses the relative advantages and disadvantages among these methods for the linear regression of Poisson count data.

Malaysian English Large Vocabulary Continuous Speech Recognizer: An Improvement Using Acoustic Model Adapation

This research project aims to develop Malaysian English Continuous Speech Recognition system by adapting US English acoustic model with Malaysian English speech corpus using Maximum a posteriori reasoning (MAP) and Maximum Likelihood Linear Regression (MLLR). During feature extraction stage, the Mel-Frequency Cepstral Coefficients (MFCC) technique was used. The Hidden Markov Model was used as the back end pattern comparison technique. For the purpose of implementation, the CMU Sphinx toolkit, which includes Pocketsphinx and Sphinxtrain as well as an acoustic model, was used to develop a speech recognition system for Malaysian English. Malaysian English speech sample will be recorded and transcribed to produce the training database required for acoustic model adaptation. The adaptation speech corpus were collected from a number of speakers. The outcome of this research could increase the application of Malaysian English speech recognition in Malaysia due to accent problem. As a result, speech recognition systems that have gone through the MAP adaptation had the best performance. Its average word error rate achieved was 32.84%. Average word recognition rate was 72.52% and average sentence error rate was 78.89%.

Maximum Likelihood Linear Regression (MLLR) for ASR Severity Based Adaptation to Help Dysarthric Speakers

Maximum Likelihood Linear Regression (MLLR) for ASR Severity Based Adaptation to Help Dysarthric Speakers

Discriminatively trained continuous Hindi speech recognition using integrated acoustic features and recurrent neural network language modeling

Abstract This paper implements the continuous Hindi Automatic Speech Recognition (ASR) system using the proposed integrated features vector with Recurrent Neural Network (RNN) based Language Modeling (LM). The proposed system also implements the speaker adaptation using Maximum-Likelihood Linear Regression (MLLR) and Constrained Maximum likelihood Linear Regression (C-MLLR). This system is discriminatively trained by Maximum Mutual Information (MMI) and Minimum Phone Error (MPE) techniques with 256 Gaussian mixture per Hidden Markov Model(HMM) state. The training of the baseline system has been done using a phonetically rich Hindi dataset. The results show that discriminative training enhances the baseline system performance by up to 3%. Further improvement of ~7% has been recorded by applying RNN LM. The proposed Hindi ASR system shows significant performance improvement over other current state-of-the-art techniques.

Speaker Feature Modeling Utilizing Constrained Maximum Likelihood Linear Regression and Gaussian Mixture Models

Speaker Feature Modeling Utilizing Constrained Maximum Likelihood Linear Regression and Gaussian Mixture Models

A comparative study of deep neural network based Punjabi-ASR system

HMM is regarded as the leader from last five decades for handling the temporal variability in an input speech signal for building automatic speech recognition system. GMM became an integral part of HMM so as to measure the efficiency of each state that stores the information of a short windowed frame. In order to systematically fit the frame, it reserves the frame coefficients and connects their posterior probability over HMM state that acts as an output. In this paper, deep neural network (DNN) is tested against the GMM through utilization of many hidden layers which helps the DNN to successfully evade the issue of overfitting on large training dataset before its performance becomes worse. The implementation DNN with robust feature extraction approach has brought a high performance margin in Punjabi speech recognition system. For feature extraction, the baseline MFCC and GFCC approaches are integrated with cepstral mean and variance normalization. The dimension reduction, decorrelation of vector information and speaker variability is later addressed with linear discriminant analysis, maximum likelihood linear transformation, SAT, maximum likelihood linear regression adaptation models. Two hybrid classifiers investigate the conceived acoustic feature vectors: GMM–HMM, and DNN–HMM to obtain improvement in performance on connected and continuous Punjabi speech corpus. Experimental setup shows a notable improvement of 4–5% and 1–3% (in connected and continuous datasets respectively).

Feature-space SVM adaptation for speaker adapted word prominence detection

Prosodic cues such as the word prominence play a fundamental role in human communication, e.g., to express important information. Since different speakers use a wide variety of features to express prominence, there is a large difference in performance between speaker dependently and speaker independently trained models. To cope with these variations without training a new speaker dependent model, in speech recognition speaker adaptation techniques such as feature-space Maximum Likelihood Linear Regression (fMLLR) turned out to be very useful. These methods are developed for GMM-HMM based classifiers under the assumption that the data can be well modeled via the mixture of a few Gaussian distributions. However, in many cases these assumptions are too restrictive. In particular a discriminative classifier such as an SVM often yields far superior results to a GMM. Therefore, we propose a new adaptation method, which adapts the data to the radial basis function kernel of the SVM. To avoid overfitting we apply two regularization terms. The first is based on fMLLR and the second is an L1 regularization to enforce a sparse transformation matrix. We analyze the method in the context of speaker adaptation for word prominence detection, with varying amounts of adaptation data and different weights of the regularization terms. We show that our novel method clearly outperforms fMLLR-GMM and fMLLR-SVM based adaptation.

Distant speech processing for smart home: comparison of ASR approaches in scattered microphone network for voice command

Voice command in multi-room smart homes for assisting people in loss of autonomy in their daily activities faces several challenges, one of them being the distant condition which impacts ASR performance. This paper presents an overview of multiple techniques for fusion of multi-source audio (pre, middle, post fusion) for automatic speech recognition for in-home voice command. The robustness of the models of speech is obtained by adaptation to the environment and to the task. Experiments are based on several publicly available realistic datasets with participants enacting activities of daily life. The corpora were recorded in natural condition, meaning background noise is sporadic, so there is no extensive background noise in the data. The smart home is equipped with one or two microphones in each room, the distance between them being larger than 1 m. An evaluation of the most suited techniques improves voice command recognition at the decoding level, by using multiple sources and model adaptation. Although Word Error Rate (WER) is between 26 and 40%, Domotic Error Rate (identical to the WER, but at the level of the voice command) is less than 5.8% for deep neural network models, the method using Feature space Maximum Likelihood Linear Regression (fMLLR) with speaker adaptation training and Subspace Gaussian Mixture Model (SGMM) exhibits comparable results.

Discriminatively trained continuous Hindi speech recognition system using interpolated recurrent neural network language modeling

This paper implements and evaluates the performance of a discriminatively trained continuous Hindi language speech recognition system. The system uses maximum mutual information and minimum phone error discriminative techniques with various numbers of Gaussian mixtures to train the automatic speech recognition (ASR) system. The training dataset consists of Hindi speech transcription. The experiments show a significant performance gain over maximum likelihood-based Hindi language speech recognition system. The system uses an efficient recurrent neural network (RNN)-based language modeling. The results indicate that the use of RNN-based language modeling enhances the performance of the ASR system. Further, the interpolation of n-gram language model (LM) with the RNNLM exhibits an additional increase in recognition performance of the implemented system. The proposed system introduces the concept of speaker adaption using maximum likelihood linear regression technique. The paper also gives an overview of the techniques used for discriminative training along with practical issues involved in their implementation.

FMLLR Speaker Normalization With i-Vector: In Pseudo-FMLLR and Distillation Framework

When an automatic speech recognition (ASR) system is deployed for real-world applications, it often receives only one utterance at a time for decoding. This single utterance could be of short duration depending on the ASR task. In these cases, robust estimation of speaker normalizing methods like feature-space maximum likelihood linear regression (FMLLR) and i -vectors may not be feasible. In this paper, we propose two unsupervised speaker normalization techniques—one at feature level and other at model level of acoustic modeling—to overcome the drawbacks of FMLLR and i -vectors in real-time scenarios. At feature level, we propose the use of deep neural networks (DNN) to generate pseudo-FMLLR features from time-synchronous pair of filterbank and FMLLR features. These pseudo-FMLLR features can then be used for DNN acoustic model training and decoding. At model level, we propose a generalized distillation framework, where a teacher DNN trained on FMLLR features guides the training and optimization of a student DNN trained on filterbank features. In both the proposed methods, the ambiguity in choosing the speaker-specific FMLLR transform can be reduced by augmenting i -vectors to the input filterbank features. Experiments conducted on 33-h and 110-h subsets of Switchboard corpus show that the proposed methods provide significant gains over DNNs trained on FMLLR, i -vector appended FMLLR, filterbank and i -vector appended filterbank features, in real-time scenario.

Speaker-Independent Silent Speech Recognition from Flesh-Point Articulatory Movements Using an LSTM Neural Network.

Silent speech recognition (SSR) converts non-audio information such as articulatory movements into text. SSR has the potential to enable persons with laryngectomy to communicate through natural spoken expression. Current SSR systems have largely relied on speaker-dependent recognition models. The high degree of variability in articulatory patterns across different speakers has been a barrier for developing effective speaker-independent SSR approaches. Speaker-independent SSR approaches, however, are critical for reducing the amount of training data required from each speaker. In this paper, we investigate speaker-independent SSR from the movements of flesh points on tongue and lip with articulatory normalization methods that reduce the inter-speaker variation. To minimize the across-speaker physiological differences of the articulators, we propose Procrustes matching-based articulatory normalization by removing locational, rotational, and scaling differences. To further normalize the articulatory data, we apply feature-space maximum likelihood linear regression and i-vector. In this paper, we adopt a bidirectional long short term memory recurrent neural network (BLSTM) as an articulatory model to effectively model the articulatory movements with long-range articulatory history. A silent speech data set with flesh points was collected using an electromagnetic articulograph (EMA) from twelve healthy and two laryngectomized English speakers. Experimental results showed the effectiveness of our speaker-independent SSR approaches on healthy as well as laryngectomy speakers. In addition, BLSTM outperformed standard deep neural network. The best performance was obtained by BLSTM with all the three normalization approaches combined.

Investigating Deep Neural Network Adaptation for Generating Exclamatory and Interrogative Speech in Mandarin

Currently, most speech synthesis systems only generate speech in a reading style, which greatly affects the expressiveness of the synthetized speech. To improve the expressiveness of the synthetized speech, this paper focuses on the generation of exclamatory and interrogative speech for Mandarin spoken language. A multi-style (exclamatory and interrogative) deep neural network-based acoustic model with a style-specific layer (which can have multiple layers) and several shared hidden layers is proposed. The style-specific layer is used to model the distinct style specific patterns. The shared layers allow maximum knowledge sharing between the declarative and multi-style speech. We investigate five major aspects of the multi-style adaptation: neural network type and topology, the number of layers in style-specific layer, initial model, adaptation parameters and adaptation corpus size. Both objective and subjective evaluations are carried out to evaluate the proposed method. Experiment results show the proposed multi-style BLSTM with top one layer adapted is superior to our prior work (which is trained by the combination of constrained Maximum likelihood linear regression and structural maximum a posterior), and achieves the best performance. We also find that adapting on both spectral and excitation parameters are more effective than only adapting on the excitation parameters.

Explicit Pitch Mapping for Improved Children’s Speech Recognition

Recognizing children’s speech on automatic speech recognition (ASR) systems developed using adults’ speech is a very challenging task. As reported by several earlier works, a severely degraded recognition performance is observed in such ASR tasks. This is mainly due to the gross mismatch in the acoustic and linguistic attributes between those two groups of speakers. One among the various identified sources of mismatch is that the vocal organs of the adult and child speakers are of significantly different dimensions. Feature-space normalization techniques are noted to effectively address the ill-effects arising from those differences. Two most commonly used approaches are the vocal-tract length normalization and the feature-space maximum-likelihood linear regression. Another important mismatch factor is the large variation in the average pitch values across the adult and child speakers. Addressing the ill-effects introduced by the pitch differences is the primary focus of the presented study. In this regard, we have explored the feasibility of explicitly changing the pitch of the children’s speech so that observed pitch differences between the two groups of speaker are reduced. In general, speech data from children is high-pitched in comparison with that from the adults’. Consequently, in this study, the pitch of the adults’ speech used for training the ASR system is kept unchanged while that for the children’s test speech data is reduced. Significant improvement in the recognition performance is noted by this explicit reduction of pitch. To conserve the critical spectral information and to avoid introducing perceptual artifacts, we have exploited timescale modification techniques for explicit pitch mapping. Furthermore, we also presented two schemes to automatically determine the factor by which the pitch of the given test data should be varied. Automatically determining the compensation factor is critical since an ASR system is expected to be accessed by both adult and child speakers. The effectiveness of proposed techniques is evaluated on adult data trained ASR systems employing different acoustic modeling approaches, viz. Gaussian mixture modeling (GMM), subspace GMM and deep neural networks (DNN). The proposed techniques are found to be highly effective in all the explored modeling paradigms. To further study the effectiveness of the proposed approaches, another DNN-based ASR system is developed on a mix of speech data from adult as well as child speakers. The use of pitch reduction is observed to be effective even in this case.

DNNs for unsupervised extraction of pseudo speaker-normalized features without explicit adaptation data

In this paper, we propose using deep neural networks (DNN) as a regression model to estimate speaker-normalized features from un-normalized features. We consider three types of speaker-specific feature normalization techniques, viz., feature-space maximum likelihood linear regression (FMLLR), vocal tract length normalization (VTLN) and a combination of both. The various un-normalized features considered were log filterbank features, Mel frequency cepstral coefficients (MFCC) and linear discriminant analysis (LDA) features. The DNN is trained using pairs of un-normalized features as input and corresponding speaker-normalized features as target. The network is optimized to reduce the mean square error between output and target speaker-normalized features. During test, un-normalized features are passed through this well trained DNN network to obtain pseudo speaker-normalized features without any supervision or adaptation data or first pass decode. As the pseudo speaker-normalized features are generated frame-by-frame, the proposed method requires no explicit adaptation data unlike in FMLLR or VTLN or i-vector. Our proposed approach is hence suitable for those scenarios where there is very little adaptation data. The proposed approach provides significant improvements over conventional speaker-normalization techniques when normalization is done at utterance level. The experiments done on TIMIT and 33-h subset and entire 300-h of Switchboard corpus supports our claim. With large amount of train data, the proposed pseudo speaker-normalized features outperforms conventional speaker-normalized features in the utterance-wise normalization scenario and gives consistent marginal improvements over un-normalized features.

Personalized Spontaneous Speech Synthesis Using a Small-Sized Unsegmented Semispontaneous Speech

A systematic approach is proposed to synthesizing personalized spontaneous speech using a small-sized unsegmented speech corpus of the target speaker. First, an automatic segmentation algorithm is employed to segment and label a collected semispontaneous speech corpus of the target speaker. Then, a pretrained average voice model is adapted to the voice model of the target speaker by using the segmented data. A postfilter based on modulation spectrum is adopted to further improve the speaker similarity of the synthesized speech as well as alleviate the over-smoothing problem of the synthesized speech. For generating spontaneous speech, a smoothing method applied at the prosodic word level is proposed to improve speech fluency. For objective evaluation on spontaneous speech segmentation, the segmentation accuracy of the proposed method is superior to that of Viterbi-based forced alignment. The results of subjective listening test also show that the proposed method can improve the spontaneity and speaker similarity of the synthesized speech compared to the maximum likelihood linear regression based speaker adaptation method.

A Fast Speaker Adaptation Approach for Large Vocabulary Continuous Speech Recognition

Acoustic model adaptation is an efficient way to compensate the acoustic mismatch due to speaker variability. We propose a linear projection (LP) based on transformation matrix speaker adaptation approach to reduce the amount of the required adaptation data. The method adopts LP function to transform multiple sets of speaker adapted (SA) acoustic models into a new single set of models. To save storage space, the transformation matrix obtained by maximum likelihood linear regression (MLLR) is used to represent the specific speaker information. Maximum likelihood (ML) criterion is used to select the SA models with the most important information from the candidate models in order to reduce the amount of transformation parameters. Supervised and unsupervised speaker adaptation experiments are carried out on a large vocabulary continuous speech recognition task to evaluate its effectiveness. Experimental results show that the proposed approach can realize fast adaptation.

Unobtrusive detection of body movements during sleep using Wi-Fi received signal strength with model adaptation technique

Although sleep is essential to healthy living, many people have issues related to insufficient or poor quality sleep. In this study, we propose a method for unobtrusively detecting body movements during sleep by measuring changes in Wi-Fi signal strength between two Wi-Fi-enabled devices because prior research has found a correlation between sleep state and body movements such as rolling over. In our method, users place two Wi-Fi-enabled devices, on the left and right sides of their bed when sleeping. Our method then detects body movement by measuring changes in Wi-Fi signal strength between the two devices. By doing so, our method can detect motion without any sensors connected to the user, using devices that are equipped with commercially available Wi-Fi modules. The main feature of our method is its ability to train a user’s body movement detection model on other users’ training data. We employ a model adaptation technique called the maximum likelihood linear regression (MLLR) to adapt a user-independent movement detection model to the user of interest. In this study, we evaluated our method using 60 sessions of data collected from six participants, and achieved approximately 82% accuracy on average with user-independent movement detection models.

A combined evaluation of established and new approaches for speech recognition in varied reverberation conditions

Robustness to reverberation is a key concern for distant-microphone ASR. Various approaches have been proposed, including single-channel or multichannel dereverberation, robust feature extraction, alternative acoustic models, and acoustic model adaptation. However, to the best of our knowledge, a detailed study of these techniques in varied reverberation conditions is still missing in the literature. In this paper, we conduct a series of experiments to assess the impact of various dereverberation and acoustic model adaptation approaches on the ASR performance in the range of reverberation conditions found in real domestic environments. We consider both established approaches such as WPE and newer approaches such as learning hidden unit contribution (LHUC) adaptations, whose performance has not been reported before in this context, and we employ them in combination. Our results indicate that performing weighted prediction error (WPE) dereverberation on a reverberated test speech utterance and decoding using a deep neural network (DNN) acoustic model trained with multi-condition reverberated speech with feature-space maximum likelihood linear regression (fMLLR) transformed features, outperforms more recent approaches and helps significantly reduce the word error rate (WER).

Bayesian Unsupervised Batch and Online Speaker Adaptation of Activation Function Parameters in Deep Models for Automatic Speech Recognition

We present a Bayesian framework to obtain maximum a posteriori MAP estimation of a small set of hidden activation function parameters in context-dependent-deep neural network-hidden markov model CD-DNN-HMM-based automatic speech recognition ASR systems. When applied to speaker adaptation, we aim at transfer learning from a well-trained deep model for a “general” usage to a “personalized” model geared toward a particular talker by using a collection of speaker-specific data. To make the framework applicable to practical situations, we perform adaptation in an unsupervised manner assuming that the transcriptions of the adaptation utterances are not readily available to the ASR system. We conduct a series of comprehensive batch adaptation experiments on the Switchboard ASR task and show that the proposed approach is effective even with CD-DNN-HMM built with discriminative sequential training. Indeed, MAP speaker adaptation reduces the word error rate WER to 20.1% from an initial 21.9% on the full NIST 2000 Hub5 benchmark test set. Moreover, MAP speaker adaptation compares favorably with other techniques evaluated on the same speech tasks. We also demonstrate its complementarity to other approaches by applying MAP adaptation to CD-DNN-HMM trained with speaker adaptive features generated through constrained maximum likelihood linear regression and further reduces the WER to 18.6%. Leveraging upon the intrinsic recursive nature in Bayesian adaptation and mitigating possible system constraints on batch learning, we also proposed an incremental approach to unsupervised online speaker adaptation by simultaneously updating the hyperparameters of the approximate posterior densities and the DNN parameters sequentially. The advantage of such a sequential learning algorithm over a batch method is not necessarily in the final performance, but in computational efficiency and reduced storage needs, without having to wait for all the data to be processed. So far, the experimental results are promising.

Arabic Speech Recognition for Connected Words Using HTK: Triphones Expanded to Gmm Based Quran Recognition

In this paper, the authors describe and propose an efficient and effective method of speaker-independent continuous Arabic speech recognition method, based on a phonetically rich speech corpus. This speech corpus contains two datasets: (1) the Spoken Arabic Digits (SAD), recorded by 66 speakers (33 male and 33 female), holds 6600 digit words in which each speaker pronounces each digit from zero to nine ten times, and (2) the second dataset is restrained to quranic sentences of three famous speakers with Tajweed rules (somehow of singing) of the last 30 chapters (Surat) of the Holy Quran. The problematic of co-articulation was treated in general with triphones as acoustic model and bigram as language model which was the most appropriate in this text; and the adjustment’s problem of sound duration in particular was also treated, with triphones expanded to Gaussians mixtures models (GMM). The proposed Arabic speech recognition system is based on the Cambridge Hidden Markov Model (HMM) Toolkit (HTK) tools. The Experimental tests show that the digit dataset using 3 emitting states per phone, has an excellent word recognition (WR) rate of 97.95% and sentence recognition (SR) rate of 93.14%. The best result of Quranic phrases obtained is 73.44% of WR rate and 14.38% of SR rate with 66.09% of accuracy within “Elmirigli” reader. Adapting the basic system to the speaker’s intonation voice, then applying an appropriate GMM to each tied-list triphone, it outperforms the Hidden Markov Models HMM-triphone acoustic model experiment by 16.93% of WR rate within “Alsudaissi” reader thanks to the solved duration vowels (mudud) problem. The contribution of this work is to apply triphones expanded to Gmms method which was compared to Maximum Likelihood Linear Regression (MLLR) and beat it by almost 5% in “Alsudaissi” reader.