Confusion analysis in phoneme based speech recognition in Hindi

The primary goal of nonlinear frequency compression (NFC) and other frequency-lowering strategies is to increase the audibility of high-frequency sounds that are not otherwise audible with conventional hearing aid (HA) processing due to the degree of hearing loss, limited HA bandwidth, or a combination of both factors. The aim of the present study was to compare estimates of speech audibility processed by NFC with improvements in speech recognition for a group of children and adults with high-frequency hearing loss. Monosyllabic word recognition was measured in noise for 24 adults and 12 children with mild to severe sensorineural hearing loss. Stimuli were amplified based on each listener's audiogram with conventional processing (CP) with amplitude compression or with NFC and presented under headphones using a software-based HA simulator. A modification of the speech intelligibility index (SII) was used to estimate audibility of information in frequency-lowered bands. The mean improvement in SII was compared with the mean improvement in speech recognition. All but 2 listeners experienced improvements in speech recognition with NFC compared with CP, consistent with the small increase in audibility that was estimated using the modification of the SII. Children and adults had similar improvements in speech recognition with NFC. Word recognition with NFC was higher than CP for children and adults with mild to severe hearing loss. The average improvement in speech recognition with NFC (7%) was consistent with the modified SII, which indicated that listeners experienced an increase in audibility with NFC compared with CP. Further studies are necessary to determine whether changes in audibility with NFC are related to speech recognition with NFC for listeners with greater degrees of hearing loss, with a greater variety of compression settings, and using auditory training.

Previous research in cochlear implant (CI) recipients with bilateral severe-to-profound sensorineural hearing loss showed improvements in speech recognition in noise using remote wireless microphone systems. However, to our knowledge, no previous studies have addressed the benefit of these systems in CI recipients with single-sided deafness. The objective of this study was to evaluate the potential improvement in speech recognition in noise for distant speakers in single-sided deaf (SSD) CI recipients obtained using the digital remote wireless microphone system, Roger. In addition, we evaluated the potential benefit in normal hearing (NH) participants gained by applying this system. Speech recognition in noise for a distant speaker in different conditions with and without Roger was evaluated with a two-way repeated-measures design in each group, SSD CI recipients, and NH participants. Post hoc analyses were conducted using pairwise comparison t-tests with Bonferroni correction. Eleven adult SSD participants aided with CIs and eleven adult NH participants were included in this study. All participants were assessed in 15 test conditions (5 listening conditions × 3 noise levels) each. The listening conditions for SSD CI recipients included the following: (I) only NH ear and CI turned off, (II) NH ear and CI (turned on), (III) NH ear and CI with Roger 14, (IV) NH ear with Roger Focus and CI, and (V) NH ear with Roger Focus and CI with Roger 14. For the NH participants, five corresponding listening conditions were chosen: (I) only better ear and weaker ear masked, (II) both ears, (III) better ear and weaker ear with Roger Focus, (IV) better ear with Roger Focus and weaker ear, and (V) both ears with Roger Focus. The speech level was fixed at 65 dB(A) at 1 meter from the speech-presenting loudspeaker, yielding a speech level of 56.5 dB(A) at the recipient's head. Noise levels were 55, 65, and 75 dB(A). Digitally altered noise recorded in school classrooms was used as competing noise. Speech recognition was measured in percent correct using the Oldenburg sentence test. In SSD CI recipients, a significant improvement in speech recognition was found for all listening conditions with Roger (III, IV, and V) versus all no-Roger conditions (I and II) at the higher noise levels (65 and 75 dB[A]). NH participants significantly benefited from the application of Roger in noise for higher levels, too. In both groups, no significant difference was detected between any of the different listening conditions at 55 dB(A) competing noise. There was also no significant difference between any of the Roger conditions III, IV, and V across all noise levels. The application of the advanced remote wireless microphone system, Roger, in SSD CI recipients provided significant benefits in speech recognition for distant speakers at higher noise levels. In NH participants, the application of Roger also produced a significant benefit in speech recognition in noise.

Hearing loss can lead to isolation and social withdrawal. The telephone oftentimes connects persons with hearing loss to society; however, telephone use is impeded by narrow bandwidth, loss of visual cues, electromagnetic interference, and inherent phone-line noise. In the past, research assessing telephone communication has consistently reported that switching from the microphone to a telecoil will typically result in the acoustic signal being discernibly softer. Properly used telecoils improve the signal-to-noise ratio (SNR), decrease the chance for acoustic feedback, and overcome the impact of distance and reverberation creating an opportunity for clearer telephone communication. Little research, however, has examined matching the telecoil frequency response to the prescribed target of the microphone frequency response (National Acoustics Laboratories, Non-Linear, version 1 [NAL-NL1]). The primary goal of this study was to determine if differences exist in speech recognition for sentences (AZ-BIO) and consonant-vowel nucleus-consonant monosyllabic words (CNC) between two telecoil conditions (default and programmed). A secondary goal was to determine if differences exist in speech recognition for sentences between male and female talkers. A single-blinded randomized controlled trial. Twenty experienced adult hearing aid users with bilateral symmetric slight to severe sensorineural hearing loss were recruited from Washington University in St. Louis School of Medicine. In addition, ten normal-hearing participants were recruited to determine the presentation level of the speech stimuli for the hearing aid participants. Participants underwent real-ear measures to program the microphone frequency response of a receiver-in-the-canal hearing aid to NAL-NL1. Using the manufacturer software, one telecoil program remained as the manufacturer default and a second telecoil program was programmed so the sound pressure level for an inductive telephone simulator frequency response matching the microphone's frequency response to obtain as close to a 0 dB relative simulated equivalent telephone sensitivity value as possible. Participants then completed speech recognition measures including AZ-BIO sentences (male and female talkers) and CNC monosyllabic words and phonemes, using both telecoil programs. A mixed model analysis was performed to examine if significant differences in speech recognition exist between the two conditions and speech stimuli. Results revealed significant improvement in overall speech recognition for the programmed telecoil performance compared with default telecoil performance (p < 0.001). Also, improved performance in the programmed telecoil was reported with a male talker (p < 0.001) and performance for sentences compared with monosyllabic words (p < 0.001) or phonemes (p < 0.001). The programmed telecoil condition revealed significant improvement in speech recognition for all speech stimuli conditions compared with the default telecoil (sentences, monosyllables, and phonemes). Additional improvement was observed in both telecoil conditions when the talker was male.

The aim of the study was to investigate whether children with bilateral conductive hearing loss benefit from their second device (i.e., the bilateral bone conduction device [BCD]). Speech recognition in noise was assessed in 10 children fitted with bilateral BCDs during childhood. Speech recognition was measured in 2 conditions with both BCDs active. Spatial resolution was tested with the Minimum Audible Angle test in the bilateral and monaural listening conditions. Children demonstrated an improvement in speech recognition when speech was presented from the front and noise was presented from the right-hand side as compared with both speech and noise being presented from the front. The minimum audible angle decreased from 57° in the best monaural condition to 13° in the bilateral condition. The audiological outcomes demonstrate the advantage of bilateral BCD fitting in children with bilateral conductive hearing loss.

This current research deals with the occlusives of the contemporary Baltic languages, Lithuanian and Latvian. In the framework of this study two acoustic cues, a duration of the release phase (for the plosives and affricates) and energy distribution in the FFT spectra obtained at the burst onset (for the plosives), have been analysed. The aim of this paper is to test the role of the former acoustic cues in the classification of the occlusives according to their voicing, manner, and place of articulation. There are 20 occlusive phonemes in Standard Lithuanian: plosives/p, pj, b, bj, t, tj, d, dj, k, kj, g, gj/ and affricates /ʦ, ʦj, ʣ, ʣj, ʧ, ʧj, ʤ, ʤj/. Standard Latvian has 12 occlusive phonemes: plosives /p, b, t, d, c, ɟ, k, g/ and affricates /ʦ, ʣ, ʧ, ʤ/.The occlusives (C) have been studied in the phonetic context of all monophthongs (V) of Standard Lithuanian or Latvian. The recorded material consisted of isolated CVC syllables pronounced by 12 Lithuanian and 12 Latvian native speakers (6 males and 6 females within each language group) of the ages from 20 to 48. Their pronunciation was flawless and consistent with the norms of Standard Lithuanian or Standard Latvian. Each syllable was recorded 3–5 times. In the framework of the current research, the material of 6192 CVC syllables (3600 pronounced by the Lithuanian speakers and 2592 pronounced by the Latvian speakers) was analysed. Data processing was performed using the software PRAAT, MS EXCEL, and SPSS. The results indicate that the set of the two acoustic cues analysed in this paper provides a distinction between the following groups of the occlusives: 1) voiced vs. voiceless – displayed by the duration of the release phase (a classification of the occlusives according to their voicing); 2) plosives vs. affricates – displayed by the duration of the release phase (a classification of the occlusives according to their manner of articulation); 3) bilabials and dentals vs. palatals (la.) and velars – both displayed by the duration of the release phase and energy distribution in the FFT spectra (a classification of the plosives according to their place of articulation). The duration of the release phase provides a distinction between plosives and affricates. Within the class of plosives it is possible to separate bilabials and dentals (articulated in the frontal part of the oral cavity) vs. palatals (la.) and velars (articulated further back in the oral cavity). The results also reveal a correlation between the duration of the release phase and voicing: voiced occlusives usually have a shorter duration of the release phase in comparison to their voiceless counterparts with the same manner and place of the articulation both in Standard Lithuanian and Standard Latvian. The duration of the release phase for the occlusive (judging by mean values) is resistant to the effect of palatalisation (in Standard Lithuanian) and to gender-related differences (both in Standard Lithuanian and Standard Latvian). According to the FFT spectra obtained at the burst onset for the plosives of Standard Lithuanian and Standard Latvian, it is possible to distinguish the same groups of these consonants as by applying the duration of the release phase: bilabials and dentals vs. palatals (la.) and velars. Energy distribution in the FFT spectra is considered to be unaffected by voicing. Since it is imposible to distinguish bilabials vs. dentals or palatals (la.) vs. velars within the class of plosives, as well as to separate dental vs. alveolar affricates, it can be concluded that the set of two acoustic cues analysed in the paper is not sufficient for a complete classification of the occlusives in Standard Lithuanian or Standard Latvian. In order to achieve more accurate results, a complex analysis of the occlusives is required using a wider set of acoustic cues. Keywords: Standard Lithuanian, Standard Latvian, occlusives, duration of the release phase, FFT spectra. DOI: http://dx.doi.org/10.15823/zz.2016.3

The aim of this study is to quantify the gap between the recognition performance of human listeners and an automatic speech recognition (ASR) system with special focus on intrinsic variations of speech, such as speaking rate and effort, altered pitch, and the presence of dialect and accent. Second, it is investigated if the most common ASR features contain all information required to recognize speech in noisy environments by using resynthesized ASR features in listening experiments. For the phoneme recognition task, the ASR system achieved the human performance level only when the signal-to-noise ratio (SNR) was increased by 15 dB, which is an estimate for the human-machine gap in terms of the SNR. The major part of this gap is attributed to the feature extraction stage, since human listeners achieve comparable recognition scores when the SNR difference between unaltered and resynthesized utterances is 10 dB. Intrinsic variabilities result in strong increases of error rates, both in human speech recognition (HSR) and ASR (with a relative increase of up to 120%). An analysis of phoneme duration and recognition rates indicates that human listeners are better able to identify temporal cues than the machine at low SNRs, which suggests incorporating information about the temporal dynamics of speech into ASR systems.

In this work, we present an alternative model to Arabic speech recognitions to boost classification accuracy of basic speech units in Industrial Internet of Things. This approach integrates both top-down and bottom-up knowledge into the automatic speech attribute transcription (ASAT) framework. ASAT is a new family of lattice-based speech recognition systems grounded on accurate detection of speech attributes. Two state-of-the-art deep neural networks were used for attribute speech feature detection, convolutional neural networks (CNNs) and feed-forward neural network, with pretraining using a stack of restricted Boltzmann machines (DBN-DNN). These attribute detectors are trained using a data-driven approach for Arabic speech data using the Buckwalter code for Arabic transcription. The detectors work by detecting phonologically distinctive articulatory features of Arabic based on manner and place of articulation, which extract characteristics of the human speech production process from the speech dataset. The results of the work are shown in terms of average equal error rate inferring that the attribute model based on the CNN consistently outperforms the DBN-DNN model. Attribute detectors with CNN, versus DBN-DNN model, reduce the AEER producing 10.42% relative improvements for manner of articulation and producing 13.11% relative improvements for place of articulation.

A recurrent error propagation network speech recognition system

In this paper, we propose a novel framework to integrate articulatory features (AFs) into HMM- based ASR system. This is achieved by using posterior probabilities of different AFs (estimated by multilayer perceptrons) directly as observation features in Kullback-Leibler divergence based HMM (KL-HMM) system. On the TIMIT phoneme recognition task, the proposed framework yields a phoneme recognition accuracy of 72.4% which is comparable to KL-HMM system using posterior probabilities of phonemes as features (72.7%). Furthermore, a best performance of 73.5% phoneme recognition accuracy is achieved by jointly modeling AF probabilities and phoneme probabilities as features. This shows the efficacy and flexibility of the proposed approach.

This work presents a method for phoneme recognition based on a model of the human binaural auditory system. Performance of the binaural system is compared to that of an equivalent monaural system across a range of signal-to-noise ratios (SNRs) and simulated azimuth separations of speech and noise sources. Comparison of binaural and monaural phoneme recognition accuracy for both speech and noise at 0/spl deg/ suggests an effective increase in SNR of 5 dB for the binaural system. Experiments are also conducted to investigate binaural system performance for speech and noise at differing azimuth angles. Two binaural speech and noise conditions are considered: i) speech from 0/spl deg/ plus noise from 0/spl deg/; and ii) speech from 0/spl deg/ plus noise from 90/spl deg/. A comparison of relative performance differences for each binaural system with the monaural system shows the greatest increase in phoneme recognition accuracy when the speech and noise are simulated using different azimuth angles (0/spl deg/ speech plus noise from 90/spl deg/).

In this paper, we compare several approaches for the extraction of modulation frequency features from speech signal using a phoneme recognition system. The general framework in these approaches is to decompose the speech signal into a set of sub-bands. Amplitude modulations (AM) in the sub-band signal are used to derive features for automatic speech recognition (ASR). Then, we propose a feature extraction technique which uses autoregressive models (AR) of sub-band Hilbert envelopes in relatively long segments of speech signal. AR models of Hilbert envelopes are derived using frequency domain linear prediction (FDLP). Features are formed by converting the FDLP envelopes into static and dynamic modulation frequency components. In the phoneme recognition experiments using the TIMIT database, the FDLP based modulation frequency features provide significant improvements compared to other techniques (average relative improvement of 7.5 % over the base-line features). Furthermore, a detailed analysis is performed to determine the relative contribution of various processing stages in the proposed technique.

Forward masking was investigated as a measure of spectral and temporal interactions. Such interactions may adversely affect speech recognition in cochlear-implant subjects. Seven subjects, implanted with the Nucleus 22 device, performed a forward-masking task. They also performed an electrode-discrimination task in order to measure spectral interactions without temporal interactions. Correlation analysis indicated a significant relationship between data obtained in the two tasks (p < 0.1). The two tasks were also correlated with the subjects' scores from five measures of speech recognition. Forward masking and electrode discrimination were strongly correlated with measures requiring consonant and phoneme recognition, respectively. These results indicate that the relationship between forward masking and speech recognition may be due, in part, to a lack of spectral resolution. The data also indicate that consonants may be more readily masked than vowels. Forward-masking data measured for all clinically programmed electrodes in three of the seven subjects were used with a model of the spectral maxima sound processor (SMSP) to estimate the number of electrodes stimulated during a consonant that might be masked by prior presentation of a vowel. These results suggest that temporal interactions across electrodes may be a factor in speech-recognition abilities of some cochlear-implant subjects.

This chapter presents Bangla (widely known as Bengali) Automatic Speech Recognition (ASR) techniques by evaluating the different speech features, such as Mel Frequency Cepstral Coefficients (MFCCs), Local Features (LFs), phoneme probabilities extracted by time delay artificial neural networks of different architectures. Moreover, canonicalization of speech features is also performed for Gender-Independent (GI) ASR. In the canonicalization process, the authors have designed three classifiers by male, female, and GI speakers, and extracted the output probabilities from these classifiers for measuring the maximum. The maximization of output probabilities for each speech file provides higher correctness and accuracies for GI speech recognition. Besides, dynamic parameters (velocity and acceleration coefficients) are also used in the experiments for obtaining higher accuracy in phoneme recognition. From the experiments, it is also shown that dynamic parameters with hybrid features also increase the phoneme recognition performance in a certain extent. These parameters not only increase the accuracy of the ASR system, but also reduce the computation complexity of Hidden Markov Model (HMM)-based classifiers with fewer mixture components.

Despite recent advances in the use of Artificial Neural Network (ANN) architectures for automatic speech recognition (ASR), relatively little attention has been given to using feature inputs beyond MFCCs in such systems. In this paper, we propose an alternative to conventional MFCC or filterbank features, using an approach based on the Generalised Hough Transform (GHT). The GHT is a common approach used in the field of image processing for the task of object detection, where the idea is to learn the spatial distribution of a codebook of feature information relative to the location of the target class. During recognition, a simple weighted summation of the codebook activations is commonly used to detect the presence of the target classes. Here we propose to learn the weighting discriminatively in an ANN, where the aim is to optimise the static phone classification error at the output of the network. As such an ANN is common to hybrid ASR architectures, the output activations from the GHT can be considered as a novel feature for ASR. Experimental results on the TIMIT phoneme recognition task demonstrate the state-of-the-art performance of the approach.

A number of techniques have been proposed in the literature for phoneme based speech recognition system. In this paper, a technique for automatic phoneme recognition using zero-crossings (ZC) and magnitude sum function (MSF) is proposed. The number of zero-crossings and magnitude sum function per frame are extracted and a minimum distance classifier is proposed to recognize the phonemes in each frame with these features. In order to increase the recognition accuracy of phonemes, a finite state machine is also proposed. The performance of the proposed phoneme recognition system is evaluated using TTS database and compared with the system using Linear Predictive Coefficients(LPC) feature inputs. Phoneme recognition accuracies of 70.93% and 55.25% are obtained for the system using LPC and the one using ZC along with MSF respectively. However, using the finite state machine proposed in this paper, 100% recognition accuracy is obtained for both the techniques. The computational costs required for recognizing various sentences using both of the feature extraction techniques are evaluated. It is observed that the proposed technique requires about 9.3 times lower computational cost than the one using LPC. The proposed phoneme recognition system is also implemented on an Altera Cyclone II FPGA using Nios II soft-core processor and custom instructions. The custom instructions for floating point arithmetic and Minimum distance classifier provide an acceleration factor of 41 and 1.87 respectively. The technique proposed here is also applicable for speech inputs from other database.