Accuracy Of Speech Recognition Systems Research Articles

Speaker Anonymization for Personal Information Protection Using Voice Conversion Techniques

As speech-based user interfaces integrated in the devices such as AI speakers become ubiquitous, a large amount of user voice data is being collected to enhance the accuracy of speech recognition systems. Since such voice data contain personal information that can endanger the privacy of users, the issue of privacy protection in the speech data has garnered increasing attention after the introduction of the General Data Protection Regulation in the EU, which implies that restrictions and safety measures for the use of speech data become essential. This study aims to filter the speaker-related voice biometrics present in speech data such as voice fingerprint without altering the linguistic content to preserve the usefulness of the data while protecting the privacy of users. To achieve this, we propose an algorithm that produces anonymized speeches by adopting many-to-many voice conversion techniques based on variational autoencoders (VAEs) and modifying the speaker identity vectors of the VAE input to anonymize the speech data. We validated the effectiveness of the proposed method by measuring the speaker-related information and the original linguistic information retained in the resultant speech, using an open source speaker recognizer and a deep neural network-based automatic speech recognizer, respectively. Using the proposed method, the speaker identification accuracy of the speech data was reduced to 0.1-9.2%, indicating successful anonymization, while the speech recognition accuracy was maintained as 78.2-81.3%.

Accuracy of Speech Recognition System’s Medical Report and Physicians' Experience in Hospitals

Introduction: Speech recognition(SR) technology has been existing for more than two decades. But, it has been rarely used in health care institutions and not applied uniformly in all the clinical domains. The aim of this study was to investigate the accuracy of speech recognition system in four different situations in the real environment of health services. We also report physicians' experience of using speech recognition technology.Method:. To do this study, NEVISA SR software professional v.3 was installed on the computers of expert physicians. The pre-designated medical report was tested by the physicians in four different modes including slow expression in a silent environment, slow expression in crowded environments, rapid expression in a silent environment and rapid expression in a busy environment. After using the speech recognition software by 15 physicians in hospitals, a designed questionnaire was distributed among them. .Results: The results showed that the highest average accuracy of speech recognition software was in the silent environment by slow expression and the minimum average accuracy was in the busy environment by rapid expression. Of all the participants in the study, 53.3% of the physicians believed that the use of speech recognition system promoted the workflow.Conclusion: We found that software accuracy was generally higher than the expectation and its use required to upgrade the system and its operation. In order to achieve the highest level of recognition rate and error reduction by speech recognition, influential factors such as environmental noise, type of software or hardware, training and experience of participants can be also considered.

Enhancement Quality and Accuracy of Speech Recognition System Using Multimodal Audio-Visual Speech signal

Enhancement Quality and Accuracy of Speech Recognition System Using Multimodal Audio-Visual Speech signal

Noise Robust Speech Recognition Using Deep Belief Networks

Most current speech recognition systems use Hidden Markov Models (HMMs) to deal with the temporal variability of speech and Gaussian mixture models (GMMs) to determine how well each state of each HMM fits a frame or a short window of frames of coefficients that represents the acoustic input. In these systems acoustic inputs are represented by Mel Frequency Cepstral Coefficients temporal spectrogram known as frames. But MFCC is not robust to noise. Consequently, with different train and test conditions the accuracy of speech recognition systems decreases. On the other hand, using MFCCs of larger window of frames in GMMs needs more computational power. In this paper, Deep Belief Networks (DBNs) are used to extract discriminative information from larger window of frames. Nonlinear transformations lead to high-order and low-dimensional features which are robust to variation of input speech. Multiple speaker isolated word recognition tasks with 100 and 200 words in clean and noisy environments has been used to test this method. The experimental results indicate that this new method of feature encoding result in much better word recognition accuracy.

Speech Recognition System: A Review

able to control devices by voice has always intrigued mankind. Today after intense research, Speech Recognition System, have made a niche for themselves and can be seen in many walks of life. The accuracy of Speech Recognition Systems remains one of the most important research challenges e.g. noise, speaker variability, language variability, vocabulary size and domain. The design of speech recognition system requires careful attentions to the challenges such as various types of Speech Classes and Speech Representation, Speech Preprocessing stages, Feature Extraction techniques, Database and Performance evaluation. This paper presents the advances made as well as highlights the pressing problems for a speech recognition system. The paper also classifies the system into Front End and Back End for better understanding and representation of speech recognition system in each part.

Robust automatic speech recognition in reverberation: onset enhancement versus binaural source separation

The precedence effect describes the auditory system’s ability to suppress later-arriving components of sound in a reverberant environment, maintaining the perceived arrival azimuth of a sound in the direction of the actual source, even though the later reverberant components may arrive from other directions. It is also widely believed that precedence-like processing can also improve speech intelligibility for humans and the accuracy of speech recognition systems in reverberant environments. While the mechanisms underlying the precedence effect have traditionally been assumed to be binaural in nature, it is also possible that the suppression of later-arriving components may take place monaurally, and that the suppression of the corresponding components of the spatial image may be a consequence of this more peripheral processing. This paper compares potential contributions of onset enhancement (and consequent steady-state suppression) of the envelopes of subband components of speech at the monaural and binaural levels. Experimental results indicate that substantial improvement in recognition accuracy can be obtained in reverberant environments if feature extraction includes both onset enhancement and binaural interaction. Recognition accuracy appears to be relatively unaffected by which stage in the binaural processing is the site of the suppression mechanism. [Work supported by the LG Yonam Foundation and Cisco.]

Accent Recognition for Indian English using Acoustic Feature Approach

Accent is the basic pattern of acoustic feature and pronunciation. It can identify the person’s social and linguistic background. It is an important source of inter as well as intra speaker variability. The accent dependent dictionary or model can be used to improve accuracy of speech recognition system. In this study we present an experimental approach of acoustic speech feature for Marathi & Arabic accents for English speaking. The detail study of acoustics correlates the accent using formant frequency, energy and pitch characteristics. The database consists of speech from speaker with Marathi as their mother tongue and speakers from Iraq with Arabic language as mother tongue. Both the speakers were asked to speak English number from zero to nine. Through experimental results the fifth formant frequency found to be very effective for accent recognition.

Spoken English Alphabet Recognition with Mel Frequency Cepstral Coefficients and Back Propagation Neural Networks

alphabet recognition as one of the subsets of speechrecognition and pattern recognition has many applications. Unfortunately, spoken alphabet recognition might not be a simple task due to highly confusable set of letters as presented in the English alphabets. The highly acoustic similarities that contribute to the confusability may hinder the accuracy of speech recognition systems. One of the confusable set is called the E-set letters which consist of the letters B, C, D, E, G, P, T, V and Z. In this study, we present aninvestigation of isolated alphabet speech recognition system using the Mel Frequency Cepstral Coefficients (MFCC) and Back-propagation Neural Network (BPNN) for the E-set and for all the 26 English alphabets. Learning rates and momentum rates of the BPNN are adjusted and varied in order to achieve the best recognition rate for the E-set and all the 26 alphabets. By adjusting these parameters,we managed to achieve 62.28% and 70.49% recognition rate for E-set recognition under speaker-independent and speaker-dependent conditions respectively.

Phone-based filter parameter optimization of filter and sum robust speech recognition using likelihood maximization

Because of noise and reverberation, accuracy of speech recognition systems decreases when the distance between talker and microphone increases. By the using of microphone arrays and appropriate filtering of received signals, the accuracy of recognizer can be increased. Many different methods for using microphone arrays have been proposed that can be classified into two main approaches: systems that perform in two independent stages of array processing and then recognition and systems that use array processing to generate a sequence of features which maximize the likelihood of generating the correct hypothesis in recognition phase. Following second approach, in this paper a new method for microphone array processing is proposed in which the parameters of array processing are adjusted in calibration phase based on phones used in language and maximum likelihood method. Optimized filter parameters are stored and used during recognition phase. A new modified Viterbi algorithm using optimal phone-based filter parameters is used for recognition phase. The proposed algorithm is analytically formulated and Persian language is used to find any improvement in speech recognition accuracy compared with results of delay and sum and utterance-based filter and sum algorithms. The results show 12.2% improvement in accuracy compared to utterance-based algorithm.

Improvement in speech recognition using phone-based filter and sum parameter optimization

Because of noise and reverberation, accuracy of speech recognition systems decreases when the distance between talker and microphone increases. In this paper a new method for microphone array processing is proposed in which parameters of array processing are adjusted in calibration phase based on phones used in language and maximum likelihood method. Optimum filter length first studied and then optimized filter parameters are stored and used during recognition phase. A new modified Viterbi algorithm using optimal phone-based filter parameters is introduced for recognition phase. The proposed algorithm is analytically formulated and improvement in recognition accuracy in various noise level and reverberation time is studied.

Improving language models for radiology speech recognition

Speech recognition systems have become increasingly popular as a means to produce radiology reports, for reasons both of efficiency and of cost. However, the suboptimal recognition accuracy of these systems can affect the productivity of the radiologists creating the text reports. We analyzed a database of over two million de-identified radiology reports to determine the strongest determinants of word frequency. Our results showed that body site and imaging modality had a similar influence on the frequency of words and of three-word phrases as did the identity of the speaker. These findings suggest that the accuracy of speech recognition systems could be significantly enhanced by further tailoring their language models to body site and imaging modality, which are readily available at the time of report creation.

Transputer-based human hearing simulation

Simulations of the human hearing system can help in a number of research fields; including work with the speech and hearing impaired as well as improving the accuracy of speech recognition systems and the naturalness of speech synthesis technology. The results from psychoacoustic experiments carried out over the last few decades have enabled models of the human peripheral hearing system to be developed. Conventionally, analyses such as the Fast Fourier Transform are used to analyze speech and other sounds to establish the acoustic cues which are important for human perception. Such analyses can be shown to be inappropriate in a number of ways. Additional insights could be gained into the importance of various acoustic cues if analyses based on hearing models were used. This paper describes an implementation of a real-time spectrograph based on a contemporary model of the peripheral human hearing system, executing on a network of T9000 transputers. The differences between it and conventional spectrographs are illustrated by means of test signals and speech sounds.

Pre-recognition measures of speaking rate

The accuracy of speech recognition systems is known to be affected by fast speech. If fast speech can be detected by means of a measure of speaking rate, the acoustic as well as language models of a speech recognition system can be adapted to compensate for fast speech effects. We have studied several measures of speaking rate which have the advantage that they can be computed prior to speech recognition. The proposed measures have been compared with conventional measures, viz., word and phone rate on the TIMIT database. Some of the proposed measures have significant correlations with phone rate and vowel duration. We have shown that the mismatch between actual and expected durations of test vowels reduces if the vowel duration models are adapted to speaking rate, as estimated by the proposed measures. These measures can be computed from features commonly employed in speech recognition, do not entail significant additional computational load and do not need labeling or segmentation of unknown utterance in terms of linguistic units.

Compensation for speech recognition in degraded acoustical environments

The accuracy of speech recognition systems degrades when operated in adverse acoustical environments. This paper discusses two ways in which more detailed mathematical descriptions of the effects of environmental degradation can improve speech recognition accuracy using both ‘‘data-driven’’ and ‘‘model-based’’ compensation strategies. Data-driven methods learn environmental characteristics through direct comparisons of speech recorded in the noisy environment with the same speech recorded under optimal conditions. Model-based methods use a mathematical model of the environment and attempt to use samples of the degraded speech to estimate model parameters. Two approaches to data-driven compensation, RATZ and STAR, are described, as well as a new approach to model-based compensation, referred to as the vector Taylor series (VTS) algorithm. Compensation algorithms are evaluated in a series of experiments measuring recognition accuracy for speech from the ARPA Wall Street Journal database that is corrupted by artificially added noise at various signal-to-noise ratios (SNRs). For any particular SNR, the greatest recognition accuracy obtained using a practical compensation algorithm is observed when that system is trained using noisy data at that SNR. The RATZ, VTS, and STAR algorithms achieve this bound at global SNRs as low as 15, 10, and 5 dB, respectively. [Work supported by ARPA.]

Whither speech recognition: the next 25 years

The fundamentals of speech recognition are reviewed. The dimensions of the speech recognition task, speech feature analysis, pattern classification using hidden Markov models, language processing, and the current accuracy of speech recognition systems are discussed. The applications of speech recognition in telecommunications, voice dictation, speech understanding for data retrieval, and consumer products are examined. >