Monophone Models Research Articles

Recognition and Spectral Analysis of Adi Digits

This article presents a method for automatically recognizing spoken digits in the Adi language using the highly effective open-source voice recognition toolkit, “Kaldi”. The study also includes an analysis of the formant frequencies and spectral characteristics of Adi digits. Adi is a zero-resource indigenous tribal language of Arunachal Pradesh with its roots in Tibeto-Burman. UNESCO's 2009 Atlas of the World's Languages in Danger classifies Adi as an endangered language in India. The research utilizes a modest digit corpus from 42 native Adi speakers, employing Mel Frequency Cepstral Coefficients (MFCC) and Perceptual Linear Prediction (PLP) features from continuous Adi speech utterances. The study explores both Monophone and Triphone models for digit recognition, with notable improvements in accuracy from 79.14% in the Monophone model to 84.29%, 88.43%, and 90.86% in the Triphone models (Tri1, Tri2, and Tri3). This proposed model holds promise for the development of Adi language speech recognition applications.

Speech recognition using an english multimodal corpus with integrated image and depth information

Traditional English corpora mainly collect information from a single modality, but lack information from multimodal information, resulting in low quality of corpus information and certain problems with recognition accuracy. To solve the above problems, this paper proposes to introduce depth information into multimodal corpora, and studies the construction method of English multimodal corpora that integrates electronic images and depth information, as well as the speech recognition method of the corpus. The multimodal fusion strategy adopted integrates speech signals and image information, including key visual information such as the speaker’s lip movements and facial expressions, and uses deep learning technology to mine acoustic and visual features. The acoustic model in the Kaldi toolkit is used for experimental research.Through experimental research, the following conclusions were drawn: Under 15-dimensional lip features, the accuracy of corpus A under monophone model was 2.4% higher than that of corpus B under monophone model when the SNR (signal-to-noise ratio) was 10dB, and the accuracy of corpus A under the triphone model at the signal-to-noise ratio of 10dB was 1.7% higher than that of corpus B under the triphone model at the signal-to-noise ratio of 10dB. Under the 32-dimensional lip features, the speech recognition effect of corpus A under the monophone model at the SNR of 10dB was 1.4% higher than that of corpus B under the monophone model at the SNR of 10dB, and the accuracy of corpus A under the triphone model at the SNR of 10dB was 2.6% higher than that of corpus B under the triphone model at the SNR of 10dB. The English multimodal corpus with image and depth information has a high accuracy, and the depth information helps to improve the accuracy of the corpus.

DNN based continuous speech recognition system of Punjabi language on Kaldi toolkit

This paper demonstrates the effect of incorporating Deep Neural Network techniques in speech recognition systems. Speech recognition through hybrid Deep Neural Networks on the Kaldi toolkit for the Punjabi language is implemented. Performance of the automatic speech recognition system drastically improves using DNN, and further Karel's DNN model gives better recognition performance as compared to Dan's DNN model. Out of MFCC and PLP features, the MFCC feature gives better results. The triphone model gives a lower word error rate than the monophone model, and 3-g gives a lower word error rate as compared to a 2-g model on the Kaldi toolkit for the continuous Punjabi speech recognition system.

Performance of Isolated and Continuous Digit Recognition System using Kaldi Toolkit

A digit recognition system is built for recognizing the sequence of digits through 0-9. The system is experimented with speech corpus created in the room environment. The acoustic information to feature representation is achieved using PLP and MFCC features. The system initially utilized the conventional GMM-HMM framework, state of the art hybrid classifier with varied number of states to complete the speech recognition task, i.e., the system is first trained and tested using Monophone models, and system’s recognition accuracy is then evaluated using Triphone Models: Triphone1 models, which was later followed by Triphones2 models and Triphones3 Models. The Ngram Language model is used for both Monophone and Triphone training. The system performance is evaluated with the use of MFCC and PLP parameterisation techniques on Kaldi toolkit. The system performance is evaluated using metrics word error rate (WER) and Word Recognition Accuracy (WRA). The proposed system can be utilized for building speech applications

Agricultural price information acquisition using noise-robust Mandarin auto speech recognition

Speech-recognition technology was applied to collect agricultural-price information. In this paper, we propose a robust continuous Mandarin speech-recognition method suitable for environments where agricultural product prices are acquired. To mitigate the decrease in recognition rate caused by the mismatch between training and real tests, we developed acoustic models based on the Hidden Markov Model (HMM) and trained the models by collecting data in different environments. The results showed that the recognition performance of triphone models was superior to that of monophone models. Both male and female HMMs performed better than the male and female mixed acoustic models. Although the decision-tree clustering method could not significantly improve the recognition rate, it evidently would reduce the quantity of triphone models. Gaussian mixture components improved the recognition rate on one hand, but they increased the calculation tasks on the other hand. The cepstral mean normalization and cepstral variance normalization methods significantly improved the identification-system performance. Under different locations and different speaker tests, the methods we used demonstrated varying degrees of improvement in recognition performance. The ultimate recognition rates were 95.04% for the males and 97.62% for the females. Speech-recognition technology can possibly be applied to collection of agricultural-price information. The experimental results showed that the models trained by these methods exhibited good recognition performance. Furthermore, the approach adopted by our research lays the foundation for the development of an application system in the future.

Efficient data selection for speech recognition based on prior confidence estimation using speech and monophone models

This paper proposes an efficient speech data selection technique that can identify those data that will be well recognized. Conventional confidence measure techniques can also identify well-recognized speech data. However, those techniques require a lot of computation time for speech recognition processing to estimate confidence scores. Speech data with low confidence should not go through the time-consuming recognition process since they will yield erroneous spoken documents that will eventually be rejected. The proposed technique can select the speech data that will be acceptable for speech recognition applications. It rapidly selects speech data with high prior confidence based on acoustic likelihood values and using only speech and monophone models. Experiments show that the proposed confidence estimation technique is over 50 times faster than the conventional posterior confidence measure while providing equivalent data selection performance for speech recognition and spoken document retrieval.

Unsupervised acoustic model adaptation algorithm using MLLR in a noisy environment

This paper proposes an unsupervised noisy environment adaptation algorithm based on the HMM acoustic model, using MLLR and a multispeaker database. An arbitrary single sentence uttered by the target speaker, together with living room noise, is used as the input, and the data with superposed noise for environment adaptation are generated from the speech database. By the above procedure, a large amount of adaptation data can be acquired without burdening the speaker. Specifically, the adaptation procedure is composed of the following three stages. (1) Speaker identification by GMM is used to select speakers with short acoustic distances from the input speaker from the database. (2) Utterances read out by the selected speakers are extracted from the database, and living room noise is superposed. (3) Using the constructed speech with superimposed noise as adaptation samples, adaptation by MLLR is performed. Then, by combining the method with unsupervised speaker adaptation based on sufficient statistics and the speaker distance, and HMM synthesis, a highly precise unsupervised integrated adaptation system is constructed. The system is evaluated through large-vocabulary continuous speech recognition, and it is shown that the adaptation model by the proposed method can realize the same or better recognition accuracy than the environment matched model, and realize performance close to supervised MLLR using several tens of samples. In a noisy environment with an SNR of 20 dB, the proposed adaptation system improved the recognition rate from 48.3% to 70.5% in the monophone model, and from 60.1% to 79.9% in the PTM model. © 2005 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 89(3): 48–58, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20227

A CALL system for teaching the duration and phone quality of Japanese tokushuhaku

This paper describes a CALL (computer-aided language learning) system that teaches the pronunciation of Japanese tokushuhaku (long vowels, the mora nasal, and mora obstruents) to beginning learners of Japanese as a second language. The system detects mispronounced tokushuhaku durations and quality, including, for instance, shortened or lengthened tokushuhaku, dentalized mora nasals, diphthongized long vowels, and aspirated plosives. Learners are prompted by the system to read tokushuhaku minimal pairs aloud. Tokushuhaku durations are measured using speech recognition and compared with distributions of native speakers. Mistakes in phone quality are detected using a speech recognizer incorporating bilingual monophone models of both the learner’s native and target languages. HMMs for the two languages are trained separately on language-dependent speech data, but are bundled together during recognition so that the closest phone recognized indicates the non-nativeness of the utterance should the recognized phone not be Japanese. Knowing the phonetics and phonology of the learner’s native language can identify non-native articulatory gestures that result in Japanese pronunciation errors, thus allowing precise corrective feedback to the learner. The paper includes an overview of the system, its reliability, validity, and effectiveness in the foreigen language classroom.

Stochastic trajectory modeling and sentence searching for continuous speech recognition

The paper first points out a defect in hidden Markov modeling (HMM) of continuous speech, referred as trajectory folding phenomenon. A new approach to modeling phoneme-based speech units is then proposed, which represents the acoustic observations of a phoneme as clusters of trajectories in a parameter space. The trajectories are modeled by a mixture of probability density functions of a random sequence of states. Each state is associated with a multivariate Gaussian density function, optimized at the state sequence level. Conditional trajectory duration probability is integrated in the modeling. An efficient sentence search procedure based on trajectory modeling is also formulated. Experiments with a speaker-dependent, 2010-word continuous speech recognition application with a word-pair perplexity of 50, using vocabulary-independent acoustic training, monophone models trained with 80 sentences per speaker, reported about a 1% word error rate. The new models were experimentally compared to continuous density mixture HMM (CDHMM) on the same recognition task, and gave significantly smaller word error rates. These results suggest that the stochastic trajectory model provides a more in-depth modeling of continuous speech signals.