Vocal Tract Length Normalization Research Articles

Improving End-to-End Models for Children’s Speech Recognition

Children’s Speech Recognition (CSR) is a challenging task due to the high variability in children’s speech patterns and limited amount of available annotated children’s speech data. We aim to improve CSR in the often-occurring scenario that no children’s speech data is available for training the Automatic Speech Recognition (ASR) systems. Traditionally, Vocal Tract Length Normalization (VTLN) has been widely used in hybrid ASR systems to address acoustic mismatch and variability in children’s speech when training models on adults’ speech. Meanwhile, End-to-End (E2E) systems often use data augmentation methods to create child-like speech from adults’ speech. For adult speech-trained ASRs, we investigate the effectiveness of augmentation methods; speed perturbations and spectral augmentation, along with VTLN, in an E2E framework for the CSR task, comparing these across Dutch, German, and Mandarin. We applied VTLN at different stages (training/test) of the ASR and conducted age and gender analyses. Our experiments showed highly similar patterns across the languages: Speed Perturbations and Spectral Augmentation yield significant performance improvements, while VTLN provided further improvements while maintaining recognition performance on adults’ speech (depending on when it is applied). Additionally, VTLN showed performance improvement for both male and female speakers and was particularly effective for younger children.

Experimental studies for improving the performance of children's speaker verification system using short utterances

The task of developing an automatic speaker verification (ASV) system for children's speech is a formidable one due to a number of reasons. The dearth of domain-specific data is one among them. The challenge further intensifies with the introduction of short utterances of speech, a relatively unexplored domain in the case of children's ASV. Voice-based biometric systems suffers miserably when speech data, inadequate both in volume as well as in duration, is used either for enrollment or verification. To circumvent the issue arising due to data scarcity, the work in this paper extensively explores in-domain as well as various out-of-domain data augmentation techniques. A data augmentation approach is proposed that encompasses both in-domain and out-of-domain data augmentation techniques. The in-domain data augmentation approach, incorporates speed perturbation of children's speech. The out-of-domain data used are from adult speakers which are known to have acoustic attributes in stark contrast to child speakers. The acoustic characteristics of the adult speech data in this study are altered on two fronts namely speech waveform modification and feature-level modification, in order to modify the adult acoustic features and render it acoustically similar to children's speech prior to augmentation. While the speech waveform modification involves various signal processing techniques like prosody modification, formant modification and voice-conversion. The feature-level modification on the other hand involves Vocal-tract length normalization technique (VTLN) which explicitly models and compensates for the ill-effects of variations in vocal tract length by linearly warping the frequency axis of speech signals. The proposed data augmentation approach helps not only in increasing the amount of training data but also in effectively capturing the missing target attributes which helps in boosting the verification performance. A relative improvement of 48.01% in equal error rate (EER) with respect to the baseline system is a testimony of it. Furthermore, the conventionally used Mel-frequency cepstral coefficients (MFCC) are known to average out the higher-frequency components. Prior literary works have shown that a significant amount of relevant acoustic information is available in the higher-frequency region of the children's speech. Therefore, effective preservation of higher-frequency contents in children's speech is of paramount importance which must be appropriately tackled for the development of a reliable and robust children's ASV system. In this regard, frame-level concatenation of the MFCC features with the Inverse-Mel-frequency cepstral coefficient (IMFCC) features is undertaken. The feature concatenation of MFCC and IMFCC is carried out with the sole intention of effectively preserving the higher-frequency contents in the children's speech data. The low canonical correlation existing between the MFCC and the IMFCC feature vectors provides the necessary impetus to go with their feature fusion. The feature concatenation approach, when combined with proposed data augmentation, helps in further improvement of the verification performance. The experimental results testify our claims, wherein we have achieved an overall relative reduction of 50.15% for equal error rate.

음성 비식별화 모델과 방송 음성 변조의 한국어 음성 비식별화 성능 비교

In broadcasts such as news and coverage programs, voice is modulated to protect the identity of the informant. Adjusting the pitch is commonly used voice modulation method, which allows easy voice restoration to the original voice by adjusting the pitch. Therefore, since broadcast voice modulation methods cannot properly protect the identity of the speaker and are vulnerable to security, a new voice modulation method is needed to replace them. In this paper, using the Lightweight speech de-identification model as the evaluation target model, we compare speech de-identification performance with broadcast voice modulation method using pitch modulation. Among the six modulation methods in the Lightweight speech de-identification model, we experimented on the de-identification performance of Korean speech as a human test and EER(Equal Error Rate) test compared with broadcast voice modulation using three modulation methods: McAdams, Resampling, and Vocal Tract Length Normalization(VTLN). Experimental results show VTLN modulation methods performed higher de-identification performance in both human tests and EER tests. As a result, the modulation methods of the Lightweight model for Korean speech has sufficient de-identification performance and will be able to replace the security-weak broadcast voice modulation.

Developing sequentially trained robust Punjabi speech recognition system under matched and mismatched conditions

Development of a native language robust ASR framework is very challenging as well as an active area of research. Although an urge for investigation of effective front-end as well as back-end approaches are required for tackling environment differences, large training complexity and inter-speaker variability in achieving success of a recognition system. In this paper, four front-end approaches: mel-frequency cepstral coefficients (MFCC), Gammatone frequency cepstral coefficients (GFCC), relative spectral-perceptual linear prediction (RASTA-PLP) and power-normalized cepstral coefficients (PNCC) have been investigated to generate unique and robust feature vectors at different SNR values. Furthermore, to handle the large training data complexity, parameter optimization has been performed with sequence-discriminative training techniques: maximum mutual information (MMI), minimum phone error (MPE), boosted-MMI (bMMI), and state-level minimum Bayes risk (sMBR). It has been demonstrated by selection of an optimal value of parameters using lattice generation, and adjustments of learning rates. In proposed framework, four different systems have been tested by analyzing various feature extraction approaches (with or without speaker normalization through Vocal Tract Length Normalization (VTLN) approach in test set) and classification strategy on with or without artificial extension of train dataset. To compare each system performance, true matched (adult train and test—S1, child train and test—S2) and mismatched (adult train and child test—S3, adult + child train and child test—S4) systems on large adult and very small Punjabi clean speech corpus have been demonstrated. Consequently, gender-based in-domain data augmented is used to moderate acoustic and phonetic variations throughout adult and children’s speech under mismatched conditions. The experiment result shows that an effective framework developed on PNCC + VTLN front-end approach using TDNN-sMBR-based model through parameter optimization technique yields a relative improvement (RI) of 40.18%, 47.51%, and 49.87% in matched, mismatched and gender-based in-domain augmented system under typical clean and noisy conditions, respectively.

Investigating a neural all pass warp in modern TTS applications

We present a neural implementation of the all pass warp (APW) previously used for vocal tract length normalisation. This includes an efficient back-propagation, which can easily be integrated in modern neural network frameworks. The APW offers a low-dimensional control to alter the spectrum, which by design generalises over different speakers. We investigate the APW in two tasks required for future dialogue or translation agents, and provide a fairly thorough literature review for both: (1) Zero-shot speaker adaptation to allow keeping the source speaker identity with very small amounts of data. Experiments show increased speaker similarity and prove that the APW increases the generalisability of a multi-speaker model. (2) Emotional speech synthesis to translate or produce affective cues. To the best of our knowledge this is the first attempt on emotional speech synthesis with an APW. While the APW is not able to increase expressiveness or audio quality, our analysis shows that the warping correlates with the level of valence in the emotion. This work should enable future research on emotion translation during machine translation.

A formant modification method for improved ASR of children’s speech

Differences in acoustic characteristics between children’s and adults’ speech degrade performance of automatic speech recognition systems when systems trained using adults’ speech are used to recognize children’s speech. This performance degradation is due to the acoustic mismatch between training and testing. One of the main sources of the acoustic mismatch is the difference in vocal tract resonances (formant frequencies) between adult and child speakers. The present study aims to reduce the mismatch in formant frequencies by modifying formants of children’s speech to better correspond to formants of adults’ speech. This is carried out by warping the linear prediction (LP) spectrum computed from children’s speech. The warped LP spectra computed in a frame-based manner from children’s speech are used with the corresponding LP residuals to synthesize speech whose formant structure is closer to that of adults’ speech. When used in testing of an ASR system trained using adults’ speech, the warping reduces the spectral mismatch in speech between training and testing and improves the system performance in recognition of children’s speech. Experiments were conducted using narrowband (8 kHz) and wideband (16 kHz) speech of adult and child speakers from the WSJCAM0 and PF_STAR databases, respectively, and by recognizing children’s speech using acoustic models trained with adults’ speech. The proposed method gave relative improvements of 24% and 11% for the DNN and TDNN acoustic models, respectively, for narrowband speech. For wideband speech, the technique gave relative improvements of 27% and 13% for the DNN and TDNN acoustic models, respectively. The performance of the proposed method was also compared to two speaker adaptation methods: vocal tract length normalization (VTLN) and speaking rate adaptation (SRA). This comparison showed the best recognition performance for the proposed method. We also combined the proposed method with VTLN and SRA, and found that the combined method gave a further reduction in WER. Moreover, our experiments carried out for noisy speech using various types of additive noise and signal-to-noise ratios showed that the proposed method performs well also for degraded speech.

Fundamental frequency feature warping for frequency normalization and data augmentation in child automatic speech recognition

Effective child automatic speech recognition (ASR) systems have become increasingly important due to the growing use of interactive technology. Due to the lack of publicly available child speech databases, young child ASR systems often rely on older child or adult speech for training data. However, there is a large acoustic mismatch between child and adult speech. This study proposes a novel fundamental frequency (fo)-based frequency warping technique for both frequency normalization and data augmentation to combat this acoustic mismatch and address the lack of available child speech training data. The technique is inspired by the tonotopic distances between formants and fo, developed to model human vowel perception. The tonotopic distances are reformulated as a linear relationship between fo and vowel formants on the Mel scale. This reformulation is verified using fo and formant measurements from child utterances. The relationship is further generalized such that the frequency warping technique only relies on two parameters. The LibriSpeech ASR corpus is used for training, and both the OGI Kids’ Speech and CMU Kids Corpora are used for both training and testing. A single word ASR experiment and a continuous read speech ASR experiment are performed to evaluate the fo-based frequency normalization and data augmentation techniques. In the single word experiment, the system using fo-based frequency normalization significantly improved over the baseline system with no normalization, with a relative improvement of up to 22.3%, when the mismatch between training and testing data was large. In the continuous speech experiment, the combination of fo-based frequency normalization and data augmentation resulted in a relative improvement of 19.3% over the baseline. Additionally, in all experiments, the fo-based techniques outperformed other techniques such as vocal tract length normalization (VTLN) or vocal tract length perturbation (VTLP). Results were validated using Gaussian mixture model (GMM), deep neural network (DNN), and bidirectional long–short term memory (BLSTM) acoustic models.

Feature compensation based on the normalization of vocal tract length for the improvement of emotion-affected speech recognition

The performance of speech recognition systems trained with neutral utterances degrades significantly when these systems are tested with emotional speech. Since everybody can speak emotionally in the real-world environment, it is necessary to take account of the emotional states of speech in the performance of the automatic speech recognition system. Limited works have been performed in the field of emotion-affected speech recognition and so far, most of the researches have focused on the classification of speech emotions. In this paper, the vocal tract length normalization method is employed to enhance the robustness of the emotion-affected speech recognition system. For this purpose, two structures of the speech recognition system based on hybrids of hidden Markov model with Gaussian mixture model and deep neural network are used. To achieve this goal, frequency warping is applied to the filterbank and/or discrete-cosine transform domain(s) in the feature extraction process of the automatic speech recognition system. The warping process is conducted in a way to normalize the emotional feature components and make them close to their corresponding neutral feature components. The performance of the proposed system is evaluated in neutrally trained/emotionally tested conditions for different speech features and emotional states (i.e., Anger, Disgust, Fear, Happy, and Sad). In this system, frequency warping is employed for different acoustical features. The constructed emotion-affected speech recognition system is based on the Kaldi automatic speech recognition with the Persian emotional speech database and the crowd-sourced emotional multi-modal actors dataset as the input corpora. The experimental simulations reveal that, in general, the warped emotional features result in better performance of the emotion-affected speech recognition system as compared with their unwarped counterparts. Also, it can be seen that the performance of the speech recognition using the deep neural network-hidden Markov model outperforms the system employing the hybrid with the Gaussian mixture model.

Developing children’s speech recognition system for low resource Punjabi language

Building an automatic speech recognition (ASR) system for children is a very challenging problem especially when the domain-specific data for training is absent or insufficient. In this paper, we present our efforts towards developing a children’s ASR system in Punjabi which a low-resourced language. To begin with, since speech data from children in the case of the Punjabi language is unavailable, we first created a small speech corpus consisting of data from both adult and child speakers. Next, an ASR system was developed on a mix of adults’ and children’s speech and tested on children’s speech. Due to the differences in acoustic attributes such as formant frequency, pitch, and speaking-rate differences between adults’ and children’s speech, the developed ASR system is observed to result in a highly degraded recognition rate. To reduce the acoustic mismatch, we have explored vocal-tract length normalization (VTLN), explicit pitch, and duration modification. All the three explored approaches are observed to be highly effective. To deal with training data scarcity, the role of prosody-modification-based out-of-domain data augmentation is studied. For that purpose, the pitch and speaking-rate of adults’ speech training set are explicitly changed to render it similar to children’s speech. The original and prosody modified data are then pooled together before learning the acoustic models. Significantly reduced error rates are observed by prosody-modification-based out-of-domain data augmentation. In addition to these, we have also studied the effect of varying the number of senones, the number of hidden nodes, and hidden layers as well as early stopping resulting in 32.1% of Relative Improvement (RI) in comparison to the baseline system with varied senones.

Noise robust in-domain children speech enhancement for automatic Punjabi recognition system under mismatched conditions

Success of any commercial Automatic Speech Recognition (ASR) system depends upon availability of its training data. Although, it's performance gets degraded due to absence of enough signal processing characteristics in less resource language corpora. Development of Punjabi Children speech system is one such challenge where zero resource conditions and variabilities in children speech occurs due to speaking speed and vocal tract length than that of adult speech. In this paper, efforts have been made to build Punjabi Children ASR system under mismatched conditions using noise robust approaches like Mel Frequency Cepstral Coefficient (MFCC) or Gammatone Frequency Cepstral Coefficient (GFCC). Consequently, acoustic and phonetic variations among adult and children speech are handled using gender based in-domain training data augmentation and later acoustic variability among speakers in training and testing sets are normalised using Vocal Tract Length Normalization (VTLN). We demonstrate that inclusion of pitch features with test normalized children dataset has significantly enhanced system performance over different environment conditions i.e clean or noisy. The experimental results show a relative improvement of 30.94% using adult female voice pooled with limited children speech over adult male corpus on noise based training data augmentation respectively.

When Automatic Voice Disguise Meets Automatic Speaker Verification

The technique of transforming voices in order to hide the real identity of a speaker is called voice disguise, among which automatic voice disguise (AVD) by modifying the spectral and temporal characteristics of voices with miscellaneous algorithms are easily conducted with softwares accessible to the public. AVD has posed great threat to both human listening and automatic speaker verification (ASV). In this paper, we have found that ASV is not only a victim of AVD but could be a tool to beat some simple types of AVD. Firstly, three types of AVD, pitch scaling, vocal tract length normalization (VTLN) and voice conversion (VC), are introduced as representative methods. State-of-the-art ASV methods are subsequently utilized to objectively evaluate the impact of AVD on ASV by equal error rates (EER). Moreover, an approach to restore disguised voice to its original version is proposed by minimizing a function of ASV scores w.r.t. restoration parameters. Experiments are then conducted on disguised voices from Voxceleb, a dataset recorded in real-world noisy scenario. The results have shown that, for the voice disguise by pitch scaling, the proposed approach obtains an EER around 7% comparing to the 30% EER of a recently proposed baseline using the ratio of fundamental frequencies. The proposed approach generalizes well to restore the disguise with nonlinear frequency warping in VTLN by reducing its EER from 34.3% to 18.5%. However, it is difficult to restore the source speakers in VC by our approach, where more complex forms of restoration functions or other paralinguistic cues might be necessary to restore the nonlinear transform in VC. Finally, contrastive visualization on ASV features with and without restoration illustrate the role of the proposed approach in an intuitive way.

The ΔF method of vocal tract length normalization for vowels

Given the acoustic consequences of physiological differences between talkers, there is a practical need for effective and theoretically motivated procedures of vowel normalization to facilitate comparison of speech produced by people who differ by dialect or language. In addition, there is a question whether listeners might utilize a normalization procedure during speech perception. This paper reports the results of two studies that explore these questions—with particular focus on vocal tract length normalization. Drawing on research in speech engineering, where accurate estimates of vocal tract length are needed in some approaches to automatic speech recognition and speaker verification, a new model of vowel normalization is introduced. The model uses a direct measure of average formant spacing (the ΔF) which can be used to measure vocal tract length. The acoustic consequences of vocal tract length differences are removed from vowel measurements by scaling vowel formant measurements by ΔF. Study 1 found that this method is comparable to Nearey’s (1978) uniform normalization method, while providing an explicit vocal tract length interpretation, and a rationalized unit of measure. Study 2 found that uniform normalization measures (which let each formant serve as a noisy estimator of ΔF) improve vowel classification even with only a couple of randomly selected vowel tokens. This suggests that vocal tract length normalization could be involved in speech perception.

Unsupervised speaker adaptation for speaker independent acoustic to articulatory speech inversion.

Speech inversion is a well-known ill-posed problem and addition of speaker differences typically makes it even harder. Normalizing the speaker differences is essential to effectively using multi-speaker articulatory data for training a speaker independent speech inversion system. This paper explores a vocal tract length normalization (VTLN) technique to transform the acoustic features of different speakers to a target speaker acoustic space such that speaker specific details are minimized. The speaker normalized features are then used to train a deep feed-forward neural network based speech inversion system. The acoustic features are parameterized as time-contextualized mel-frequency cepstral coefficients. The articulatory features are represented by six tract-variable (TV) trajectories, which are relatively speaker invariant compared to flesh point data. Experiments are performed with ten speakers from the University of Wisconsin X-ray microbeam database. Results show that the proposed speaker normalization approach provides an 8.15% relative improvement in correlation between actual and estimated TVs as compared to the system where speaker normalization was not performed. To determine the efficacy of the method across datasets, cross speaker evaluations were performed across speakers from the Multichannel Articulatory-TIMIT and EMA-IEEE datasets. Results prove that the VTLN approach provides improvement in performance even across datasets.

Vocal Tract Length Normalization using a Gaussian mixture model framework for query-by-example spoken term detection

A speech spectrum is known to be changed by the variations in the length of the vocal tract of a speaker. This is because of the fact that speech formants are inversely related to the vocal tract length (VTL). The process of compensating spectral variation due to the length of the vocal tract is known as Vocal Tract Length Normalization (VTLN). VTLN is a very important speaker normalization technique for speech recognition and related tasks. In this paper, we used Gaussian Posteriorgram (GP) of VTL-warped spectral features for a Query-by-Example Spoken Term Detection (QbE-STD) task. This paper presents the use of a Gaussian Mixture Model (GMM) framework for VTLN warping factor estimation. In particular, the presented GMM framework does not require phoneme-level transcription. We observed the correlation between the VTLN warping factor estimates obtained via a supervised HMM-based approach and an unsupervised GMM-based approach. In addition, a phoneme recognition and speaker de-identification tasks were conducted using GMM-based VTLN warping factor estimates. For QbE-STD, we considered three spectral features, namely, Mel Frequency Cepstral Coefficients (MFCC), Perceptual Linear Prediction (PLP), and MFCC-TMP (which uses Teager Energy Operator (TEO) to exploit implicitly magnitude and phase information in the MFCC framework). Linear frequency scaling variations for VTLN warping factor are incorporated into these three cepstral representations for the QbE-STD task. Similarly, VTL-warped Gaussian posteriorgram improved the Maximum Term Weighted Value by 0.021 (i.e., 2.1%), and 0.015 (i.e., 1.5%), for MFCC and PLP feature sets, respectively, on the evaluation set of the MediaEval SWS 2013 corpus. The better performance is primarily due to VTLN warping factor estimation using unsupervised GMM framework. Finally, the effectiveness of the proposed VTL-warped GP is presented to rescore using various detection sources, such as depth of detection valley, Self-Similarity Matrix, Pseudo Relevance Feedback and weighted mean features.

Role of Linear, Mel and Inverse-Mel Filterbanks in Automatic Recognition of Speech from High-Pitched Speakers

In the context of automatic speech recognition (ASR), the power spectrum is generally warped to the Mel-scale during front-end speech parameterization. This is motivated by the fact that human perception of sound is nonlinear. The Mel-filterbank provides better resolution for low-frequency contents, while a greater degree of averaging happens in the high-frequency range. The work presented in this paper aims at studying the role of linear, Mel and inverse-Mel-filterbanks in the context of ASR. When speech data are from high-pitched speakers like children, there is a significant amount of relevant information in the high-frequency region. Hence, down-sampling the information in that range through Mel-filterbank reduces the recognition performance. On the other hand, employing inverse-Mel or linear-filterbanks is expected to be more effective in such cases. The same has been experimentally validated in this work. For that purpose, an ASR system is developed on adults’ speech and tested using data from adult as well as child speakers. Significantly improved recognition rates are noted for children’s as well adult females’ speech when linear or inverse-Mel-filterbank is used. The use of linear filters results in a relative improvement of $$21\%$$ over the baseline. To further boost the performance, vocal-tract length normalization, explicit pitch scaling and pitch-adaptive spectral estimation are also explored on top of linear filterbank.

Improving children’s mismatched ASR using structured low-rank feature projection

The work presented in this paper explores the issues in automatic speech recognition (ASR) of children’s speech on acoustic models trained on adults’ speech. In such contexts, due to a large acoustic mismatch between training and test data, highly degraded recognition rates are noted. Even with the use of vocal tract length normalization (VTLN), the mismatched case recognition performance is still much below that for the matched case. Our earlier studies have shown that, for commonly used mel-filterbank-based cepstral features, the acoustic mismatch is exacerbated by insufficient smoothing of pitch harmonics for child speakers. To address this problem, a structured low-rank projection of the features vectors prior to learning the acoustic models as well as before decoding is proposed in this paper. To accomplish this, first a low-rank transform is learned on the training data (adults’ speech). Any dimensionality reduction technique which depends on the variance of the training data may be used for this purpose. In this work, principal component analysis and heteroscedastic linear discriminant analysis have been explored for the same. When the derived low-rank projection is applied in the mismatched testing case, it alleviates the pitch-dependent mismatch. The proposed approach provides a relative recognition performance improvement of 35% over the VTLN included baseline for the children’s mismatched ASR employing acoustic modeling based on hidden Markov models (HMM) with observation densities modeled using Gaussian mixture models (GMM). In addition to that, other acoustic modeling approaches based on subspace GMM (SGMM) and deep neural networks (DNN) have also been explored. Projecting the data to a lower-dimensional subspace is found to be effective in those frameworks as well. In the case of SGMM and DNN-based systems, the proposed approach is noted to result in relative recognition performance improvements of 33% and 21%, respectively, over their corresponding baselines.

A computationally efficient approach for acoustic class specific VTLN using regression tree

In this paper a novel frame-work for acoustic class specific vocal tract length normalization (VTLN) is developed. Unlike the computationally expensive grid search involved in conventional VTLN, the proposed technique works in the joint paradigm of linear transform VTLN and the txpectation maximization algorithm, and uses Regression class tree for robustness. Experimental results are demonstrated on two wall street journal (WSJ) test sets Nov92 eval and Dev-93 with the acoustic model being trained on the WSJ-284 set. It is found that the proposed acoustic class specific VTLN provides consistent improvements in word accuracies in comparison to the conventional VTLN which uses single warp-factor for spectral warping.

Studying the role of pitch-adaptive spectral estimation and speaking-rate normalization in automatic speech recognition

In the context of automatic speech recognition (ASR) systems, the front-end acoustic features should not be affected by signal periodicity (pitch period). Motivated by this fact, we have studied the role of pitch-synchronous spectrum estimation approach, referred to as TANDEM STRAIGHT, in this paper. TANDEM STRAIGHT results in a smoother spectrum devoid of pitch harmonics to a large extent. Consequently, the acoustic features derived using the smoothed spectra outperform the conventional Mel-frequency cepstral coefficients (MFCC). The experimental evaluations reported in this paper are performed on speech data from a wide range of speakers belonging to different age groups including children. The proposed features are found to be effective for all groups of speakers. To further improve the recognition of children's speech, the effect of vocal-tract length normalization (VTLN) is studied. The inclusion of VTLN further improves the recognition performance. We have also performed a detailed study on the effect of speaking-rate normalization (SRN) in the context of children's speech recognition. An SRN technique based on the anchoring of glottal closure instants estimated using zero-frequency filtering is explored in this regard. SRN is observed to be highly effective for child speakers belonging to different age groups. Finally, all the studied techniques are combined for effective mismatch reduction. In the case of children's speech test set, the use of proposed features results in a relative improvement of 21.6% over the MFCC features even after combining VTLN and SRN.

An Experimental Study on the Significance of Variable Frame-Length and Overlap in the Context of Children’s Speech Recognition

It is well known that the recognition performance of an automatic speech recognition (ASR) system is affected by intra-speaker as well inter-speaker variability. The differences in the geometry of vocal organs, pitch and speaking-rate among the speakers are some such inter-speaker variabilities affecting the recognition performance. A mismatch between the training and test data with respect to any of those aforementioned factors leads to increased error rates. An example of acoustically mismatched ASR is the task of transcribing children’s speech on adult data-trained system. A large number of studies have been reported earlier that present a myriad of techniques for addressing acoustic mismatch arising from differences in pitch and dimensions of vocal organs. At the same time, only a few works on speaking-rate adaptation employing timescale modification have been reported. Furthermore, those studies were performed on ASR systems developed using Gaussian mixture models. Motivated by these facts, speaking-rate adaptation is explored in this work in the context of children’s ASR system employing deep neural network-based acoustic modeling. Speaking-rate adaptation is performed by changing the frame-length and overlap during front-end feature extraction process. Significant reductions in errors are noted by speaking-rate adaptation. In addition to that, we have also studied the effect of combining speaking-rate adaptation with vocal-tract length normalization and explicit pitch modification. In both the cases, additive improvements are obtained. To summarize, relative improvements in 15–20% over the baselines are obtained by varying the frame-length and frame-overlap.

AL-QURAN RECITATION SPEECH SIGNALS TIME SERIES SEGMENTATION FOR SPEAKER ADAPTATION USING DYNAMIC TIME WARPING

The transformation of the whole traditional process of Al-Quran Recitation into automated system application could raise the issues of robustness and integrity of recitation and correction quality and acceptance. There are several existing variations especially involve of rhythm, tone and length of recitation for different speaker. Dynamic Time warping (DTW) is used as dynamic programming to normalize the recitation speech length of speaker which warp the speech spectrum amplitude in time series for the experts and learners. DTW is used to gain the same length of recitation which warp the amplitudes, rhythm and tone into same length in time series segments based on formant frequency frame for different recitation speakers. There are 8 experts and 10 users from the Malay Muslim community had three selected surahs for Al-Quran recitation session. The aim of the paper is to normalize Al-Quran recitation speech signals as speaker adaption between experts and learner then represent each recitation speech signal at same vocal tract formant frequency that can be used in robust Automatic Recitation Recognition (ARR) to evaluate the performance of recitation evaluation. Keywords: Al-Quran speech recitation, Automatic Recitation Recognition, speaker adaptation, vocal tract length normalization