Automatic Dialect Identification Research Articles

Robust automatic accent identification based on the acoustic evidence

Robust automatic accent identification based on the acoustic evidence

Under-resourced dialect identification in Ao using source information.

This paper reports the findings of an automatic dialect identification (DID) task conducted on Ao speech data using source features. Considering that Ao is a tone language, in this study for DID, the gammatonegram of the linear prediction residual is proposed as a feature. As Ao is an under-resourced language, data augmentation was carried out to increase the size of the speech corpus. The results showed that data augmentation improved DID by 14%. A perception test conducted on Ao speakers showed better DID by the subjects when utterance duration was 3 s. Accordingly, automatic DID was conducted on utterances of various duration. A baseline DID system with the Slms feature attained an average F1-score of 53.84% in a 3 s long utterance. Inclusion of source features, Silpr and S, improved the F1-score to 60.69%. In a final system, with a combination of Silpr, S, Slms, and Mel frequency cepstral coefficient features, the F1-score increased to 61.46%.

Analysis and modeling of dialect information in Ao, a low resource language.

Ao is a Tibeto-Burman language spoken in Nagaland, India. It is a low resource, tonal language with three lexical tones, namely, high, mid, and low. However, tone assignment on lexical words may differ among the three dialects of Ao, namely, Chungli, Mongsen, and Changki. In this work, an acoustic study is conducted on the three tones in the three dialects of Ao. It was found that the acoustic characteristics of the tones in the Changki dialect are markedly different from that of the Chungli and the Mongsen dialects. Hence, in the latter part of the work, automatic dialect identification (DID) in the Ao dialects is attempted with Mel Frequency Cepstral Coefficients, Shifted Delta Cepstral coefficients, and F0 features using the Gaussian Mixture models. It is confirmed that in both text-dependent and text-independent DID, the F0 features improve the accuracy of classification.

Automatic accent identification as an analytical tool for accent robust automatic speech recognition

We present a novel study of relationships between automatic accent identification (AID) and accent-robust automatic speech recognition (ASR), using i-vector based AID and deep neural network, hidden Markov Model (DNN-HMM) based ASR. A visualization of the AID i-vector space and a novel analysis of the accent content of the WSJCAM0 corpus are presented. Accents that occur at the periphery of AID space are referred to as “extreme”. We demonstrate a negative correlation, with respect to accent, between AID and ASR accuracy, where extreme accents exhibit the highest AID and lowest ASR performance. These relationships between accents inform a set of ASR experiments in which a generic training set (WSJCAM0) is supplemented with a fixed amount of accented data from the ABI (Accents of the British Isles) corpus. The best performance across all accents, a 32% relative reduction in errors compared with the baseline ASR system, is obtained when the supplementary data comprises extreme accented speech, even though this accent accounts for just 14% of the test data. We conclude that i-vector based AID analysis provides a principled approach to the selection of training material for accent robust ASR. We speculate that this may generalize to other detection technologies and other types of variability, such as Speaker Identification (SI) and speaker variability.

Spoken Arabic dialect recognition using X-vectors

AbstractThis paper describes our automatic dialect identification system for recognizing four major Arabic dialects, as well as Modern Standard Arabic. We adapted the X-vector framework, which was originally developed for speaker recognition, to the task of Arabic dialect identification (ADI). The training and development ADI VarDial 2018 and VarDial 2017 were used to train and test all of our ADI systems. In addition to the introduced X-vectors, other systems use the traditional i-vectors, bottleneck features, phonetic features, words transcriptions, and GMM-tokens. X-vectors achieved good performance (0.687) on the ADI 2018 Discriminating between Similar Languages shared task testing dataset, outperforming other systems. The performance of the X-vector system is slightly improved (0.697) when fused with i-vectors, bottleneck features, and word uni-gram features.

Automatic dialect identification system for Kannada language using single and ensemble SVM algorithms

In this paper, an automatic dialect identification (ADI) system is proposed by extracting spectral and prosodic features for Kannada language. A new dialect dataset is collected from native speakers of Kannada language (A Dravidian language). This dataset includes five distinct dialects of Kannada language representing five geographical regions of Karnataka state. Investigation of the significance of spectral and prosodic variations on five Kannada dialects is carried out. Mel-frequency cepstral coefficients (MFCCs), spectral flux, and entropy are used as representatives of spectral features. Besides, pitch and energy features are extracted as representatives of prosodic parameters for identification of dialects. These raw feature vectors are further processed to get a new derived feature vectors by using statistical processing. In this paper, a single classifier based multi-class support vector machine (SVM) and multiple classifier based ensemble SVM (ESVM) techniques are employed for classification of dialects. The effectiveness and performance evaluation of the explored features are carried out on newly collected Kannada speech corpus, with five Kannada dialects and internationally known standard Intonation Variation in English (IViE) dataset with nine British English dialects. Experimental results have demonstrated that the derived feature vectors performs better when compared to raw feature vectors. However, ESVM technique has demonstrated better performance over a single SVM. Spectral and prosodic features have resulted individually with the dialect recognition performance of 83.12% and 44.52% respectively. Further, the complementary nature of both spectral and prosodic features is evaluated by combining both feature vectors for dialect recognition. However, an increase in dialect recognition performance of about 86.25% is observed. This indicates the existence of complementary dialect specific evidence with spectral and prosodic features. The experiments conducted on standard IViE corpus have shown a higher recognition rate of 91.38% using ESVM. Proposed ADI systems with derived features have shown better performance over the state-of-the-art i-vector feature based systems on both datasets.

Extraction of Prosodic Features to Automatically Recognize Tamil Dialects

In speech processing applications, there is a great demand for recognizing the speech correctly. Automatic Dialect Identification has attracted researchers in the field of speech signal processing. Speaker’s phonemes, pronunciation, and traits such as tonality, loudness and nasality can be recognized with the help of Dialects. Dialect can be defined as the language characteristics of a specific community. Prosodic features are features that appear, when the sounds are put together in connected speech. Now a day’s prosodic features are used in most emotion recognition algorithms. Prosodic features are relatively simple in their structures and known for their effectiveness in some recognition tasks. This paper proposes a language model to identify these parameters and then to recognize the regional speaker dialects to identify the speaker using various feature extraction techniques for Tamil language.

The effect of pitch tracking on automatic dialect identification

Pitch tracking is one of the most important research topics in the recognition and identification area. This study concerns the effect of the pitch tracking technique used on the accuracy and speed of automatic dialect identification. This effort was carried out using the TIMIT database. The pitch tracking procedures investigated are the Boersma algorithm, the iterative adaptive inverse filtering approach, and the summation of residual harmonics method. All else being equal, the summation of residual harmonics provided the highest accuracy as well as the fastest performance of the three methods.

Unsupervised accent classification for deep data fusion of accent and language information

Automatic Dialect Identification (DID) has recently gained substantial interest in the speech processing community. Studies have shown that the variation in speech due to dialect is a factor which significantly impacts speech system performance. Dialects differ in various ways such as acoustic traits (phonetic realization of vowels and consonants, rhythmical characteristics, prosody) and content based word selection (grammar, vocabulary, phonetic distribution, lexical distribution, semantics). The traditional DID classifier is usually based on Gaussian Mixture Modeling (GMM), which is employed as baseline system. We investigate various methods of improving the DID based on acoustic and text language sub-systems to further boost the performance. For acoustic approach, we propose to use i-Vector system. For text language based dialect classification, a series of natural language processing (NLP) techniques are explored to address word selection and grammar factors, which cannot be modeled using an acoustic modeling system. These NLP techniques include: two traditional approaches, including N-Gram modeling and Latent Semantic Analysis (LSA), and a novel approach based on Term Frequency–Inverse Document Frequency (TF-IDF) and logistic regression classification. Due to the sparsity of training data, traditional text approaches do not offer superior performance. However, the proposed TF-IDF approach shows comparable performance to the i-Vector acoustic system, which when fused with the i-Vector system results in a final audio-text combined solution that is more discriminative. Compared with the GMM baseline system, the proposed audio-text DID system provides a relative improvement in dialect classification performance of +40.1% and +47.1% on the self-collected corpus (UT-Podcast) and NIST LRE-2009 data, respectively. The experiment results validate the feasibility of leveraging both acoustic and textual information in achieving improved DID performance.

GMM-Based Maghreb Dialect IdentificationSystem

While Modern Standard Arabic is the formal spoken and written language of the Arab world; dialects are the major communication mode for everyday life. Therefore, identifying a speaker`s dialect is critical in the Arabic-speaking world for speech processing tasks, such as automatic speech recognition or identification. In this paper, we examine two approaches that reduce the Universal Background Model (UBM) in the automatic dialect identification system across the five following Arabic Maghreb dialects: Moroccan, Tunisian, and 3 dialects of the western (Oranian), central (Algiersian), and eastern (Constantinian) regions of Algeria. We applied our approaches to the Maghreb dialect detection domain that contains a collection of 10-second utterances and we compared the performance precision gained against the dialect samples from a baseline GMM-UBM system and the ones from our own improved GMM-UBM system that uses a Reduced UBM algorithm. Our experiments show that our approaches significantly improve identification performance over purely acoustic features with an identification rate of 80.49%.

Fusion of multi-stream speech features for dialect classification

Current research in the area of voice recognition has entered a new stage. It does not only concentrate on the correct evaluation of linguistic information embodied in the speech signal, it also works towards identification of variations naturally present in speech. Undoubtedly, the focus is to enhance the accuracy and precision of the developed technique. Speaker’s accent due to his native dialect is one of the major source of variability. Prior knowledge of the spoken dialect will help in the creation of multi-model speech recognition system and can enhance its recognition performance. This paper focusses on applying some of the established dialect identification techniques to identify speaker’s spoken dialect among dialects of Hindi. Fusion of multiple streams obtained as a combination of phonetic and prosodic features is implemented to exploit the acoustic information. The work presented here also exploits the ability of AANN to capture the distribution of data points in a reduced number and further to classify them into groups. System performance for different level of fusion is recorded for Hindi dialect classification. It is observed that Duration as prosodic feature is an important cue for automatic dialect identification systems.

The effect of code-mixing on accent identification accuracy

We investigate whether accent identification is more effective for English utterances embedded in a different language as part of a mixed code than for English utterances that are part of a monolingual dialogue. Our focus is on Xhosa and Zulu, two South African languages for which code-mixing with English is very common. In order to carry out our investigation, we extract English utterances from mixed-code Xhosa and Zulu speech corpora, as well as comparable utterances from an English-only corpus by Xhosa and Zulu mother-tongue speakers. Experiments using automatic accent identification systems show that identification is substantially more accurate for the utterances originating from the mixed-code speech. These findings are supported by a corresponding set of perceptual experiments in which human subjects were asked to identify the accents of recorded utterances. We conclude that accent identification is more successful for these utterances because accents are more pronounced for English embedded in mother-tongue speech than for English spoken as part of a monolingual dialogue by non-native speakers. Furthermore we find that this is true for human listeners as well as for automatic identification systems.

Incorporating vocalic segment memories in automatic dialect identification

Systematic pronunciation differences among speakers of regional varieties of U.S. English are recognizable to other native speakers to varying degrees. This has often been demonstrated through experiments wherein listeners were asked to match a talker to his or her dialect region. Machines have also been able to identify the regional origin of a speaker to some degree, although attempts to this end have typically not been as successful as efforts to automatically identify the language of a speaker. In order to refine the dialect discrimination ability of a machine, this paper draws methodological inspiration from the area of musical artist classification, and from linguistic notions that vowels contribute more heavily than consonants to regional differences. Using this insight, an automatic dialect identification system is developed that first recognizes the more vowel-like slices of the signal, and then updates a vocalic segment memory component with Mel-frequency cepstral coefficient, formant, and pitch information from the current frame. Besides providing a means to analyze MFCC and formant trajectories, the segment memory enriches the representation of vocalic events by allowing the system to explicitly model prosodic aspects such as duration and tilt.