An enhanced automatic speech recognition system for Arabic

Menacer, Mohamed Amine; Mella, Odile; Fohr, Dominique; Jouvet, Denis; Langlois, David; Smaïli, Kamel

doi:10.18653/v1/w17-1319

Cited by 15 publications

(5 citation statements)

References 25 publications

(18 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent work has been conducted for MSA automatic speech recognition utilizing weighted finite state transducer structure in the Kaldi ASR system [21]. Finite state transducer has also been utilized for MSA morphological analysis and diacritization [22].…”

Section: Literature Reviewmentioning

confidence: 99%

Concatenative Speech Recognition using Morphemes

Jafri¹

2021

IJACSA

View full text Add to dashboard Cite

This paper adopts a novel sub-lexical approach to construct viable continuous speech recognition systems with scalable vocabulary that use the components of words to form the elements of pronunciation dictionaries and recognition lattices. The proposed Concatenative ASR family utilizes combination rules between morphemes (prefixes, stems, and suffixes), along with their theoretical grammatical categories. The constrained structure reduces invalid words by using grammar rules governing agglutination of affixes with stems, while having a large vocabulary space and hence fewer out-of-vocabulary words. In pursuing this approach, the project develops automatic speech recognition (ASR) parameterized models, designs parameter values, constructs and implements ASR systems, and analyzes the characteristics of these systems. The project designs parameter values in the context of Arabic to yield a subset hierarchy of vocabularies of the ASR systems facilitating meaningful analysis. It investigates the characteristics of the ASR systems with respect to vocabulary, recognition lattice, dictionary, and word error rate (WER). In the experiments, the standard Word ASR model has the best characteristics for vocabulary of up to five thousand words and the Concatenative ASR family is most appropriate for vocabulary of up to half a million words. The paper shows that the approach used encompasses fundamentally different processes of word formation and thus is applicable to languages that exhibit concatenative word-formation processes.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Concatenative Speech Recognition using Morphemes

Jafri¹

2021

IJACSA

View full text Add to dashboard Cite

show abstract

“…Transcriptions for interviews add further burden to the researcher (Stewart, Shamdasani, & Rook, 2007). Because Arabic speech recognition systems, that transform speech to text, are still facing many challenges (AbuZeina & Elshafei, 2012;Menacer et al, 2017), the researcher decided to avoid transcription and instead record summary notes, which may cause bias in summarisation, but the researcher will check the audio recordings to obtain clarifications and augment the notes.…”

Section: Research Limitationsmentioning

confidence: 99%

Enhancing Employability and e-Business Capacities for Arabic-Speaking Residents of Australia through START Online Training

Hanna

Conner

Sweeney

2020

ssp

View full text Add to dashboard Cite

Background: Arabic minority groups in Australia face language barriers and shortage of computer skills, which cause unemployment and/or an inability to establish their own businesses. The unemployment rate for this group is ~ 20.5%, which is 3-times higher than the average unemployment rate in Australia (~7%). The unemployment will get worse due to COVID-19 pandemic. The current provision of computer and language training in Australia is in English, which results in longer training times and higher chances of non-completion. Objective: The Smart Training for Arabic Residents on Technology (START) is an interventional online bi-lingual training that assists Arabic-speaking residents of Australia to establish an online business with minimum resources (money, space, and infrastructure) or at least help them find suitable employment. Methods: START uses Design-Based Research DBR, as it has its own progressive refinement approach. Both qualitative methods (skills assessment interviews, semi-guided observation, and final follow-up interviews) and quantitative methods (practical tests, log analysis/learning analytics, feedback surveys) contribute to evaluation and improvement cycles. Discussion: DBR has not been applied to vocational immigrant education previously. This research project contributes to a better understanding of the relationships between educational theory, designed learning and outcomes, to help advance learning and teaching environments by refining critical factors that lead to success for trainees. Practically, Arabic residents are provided opportunities to master computer and English skills for establishing their own online businesses. This research, however, has some limitations. Usually a team of teacher, learning designer, and researcher is recommended for DBR, but that is not possible in this PhD study. It is also acknowledged that although this study aims for optimal refinement of the START program, through multiple cycles of improvement, realistically it will be difficult to “recreate” the exact learning environment in future programs.

show abstract

“…The probabilistic algorithms (DNN, HMM, SVM etc. ), as used to cluster the acoustic model, are directly applied to this set of frames which forms the training base [2,4,19]. Even with hybrid methods, the estimation of the acoustic model is always done in the same manner.…”

Section: Proposed Speech Recognition Systemmentioning

confidence: 99%

Recognition of isolated digits using DNN–HMM and harmonic noise model

Guerid¹,

Houacine

2019

IET signal process.

View full text Add to dashboard Cite

Speech recognition is an area that is constantly developing. In this study, the authors present a new system of speech recognition applied to the Arabic language. The system proposed here is based on the harmonic plus noise model (HNM). This model is rather used in speech synthesis tasks and is known for providing excellent speech production quality. Thus, their contribution lies in replacing the conventional mel-frequency cepstrum coefficients (MFCC) parameters with a set of acoustic parameters, extracted through the HNM estimation process. The HNM model allows development of a more adapted processing by distinguishing voiced and unvoiced speech frames and by characterising the harmonic property of speech. As common, their system consists of both training and recognition phases. Deep neural networks and hidden Markov models (DNN-HMM) are used for modelling the voiced frames corresponding to the harmonic part. The DNN model is estimated with static and dynamic parameters. Moreover, the unvoiced frames, which represent the noise part of the HNM, are clustered with an HMM model. The spoken Arabic digits are used to measure the performance of the proposed recognition system and a comparison with the MFCC-based approach is performed.

show abstract

An enhanced automatic speech recognition system for Arabic

Cited by 15 publications

References 25 publications

Concatenative Speech Recognition using Morphemes

Concatenative Speech Recognition using Morphemes

Enhancing Employability and e-Business Capacities for Arabic-Speaking Residents of Australia through START Online Training

Recognition of isolated digits using DNN–HMM and harmonic noise model

Contact Info

Product

Resources

About