Speech Recognition of Moroccan Dialect Using Hidden Markov Models

Bezoui, Mouaz; Abderrahim, Beni Hssane; Elmoutaouakkil, Abdelmajid

doi:10.1016/j.procs.2019.04.138

Cited by 21 publications

(11 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The ML technique is still utilized in conjunction. The classic approaches, such as the HMM, recognize four Moroccan dialect speakers using 20 speakers; this research achieved a 90% accuracy rate for speaker recognition ( Mouaz, Abderrahim & Abdelmajid, 2019 ).…”

Section: Related Workmentioning

confidence: 97%

“…Therefore, artificial intelligence (AI) has been used in numerous fields of the latest studies. Previously, speech recognition studies utilizing ML achieved a high degree of precision by using the Gaussian mixture model (GMM) technique ( Marufo da Silva, Evin & Verrastro, 2016 ; Maghsoodi et al, 2019 ; Mouaz, Abderrahim & Abdelmajid, 2019 ), and the hidden Markov model (HMM) technique ( Veena & Mathew, 2015 ; Bao & Shen, 2016 ; Chakroun et al, 2016 ; Maurya, Kumar & Agarwal, 2018 ). However, as the data increases, the level of accuracy with these techniques drops rapidly, to the point where these traditional ML approaches suffer from low accuracy and generalization issues ( Xie et al, 2018 ).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Data augmentation and deep neural networks for the classification of Pakistani racial speakers recognition

Amjad

Khan

Chang

2022

PeerJ Computer Science

View full text Add to dashboard Cite

Speech emotion recognition (SER) systems have evolved into an important method for recognizing a person in several applications, including e-commerce, everyday interactions, law enforcement, and forensics. The SER system’s efficiency depends on the length of the audio samples used for testing and training. However, the different suggested models successfully obtained relatively high accuracy in this study. Moreover, the degree of SER efficiency is not yet optimum due to the limited database, resulting in overfitting and skewing samples. Therefore, the proposed approach presents a data augmentation method that shifts the pitch, uses multiple window sizes, stretches the time, and adds white noise to the original audio. In addition, a deep model is further evaluated to generate a new paradigm for SER. The data augmentation approach increased the limited amount of data from the Pakistani racial speaker speech dataset in the proposed system. The seven-layer framework was employed to provide the most optimal performance in terms of accuracy compared to other multilayer approaches. The seven-layer method is used in existing works to achieve a very high level of accuracy. The suggested system achieved 97.32% accuracy with a 0.032% loss in the 75%:25% splitting ratio. In addition, more than 500 augmentation data samples were added. Therefore, the proposed approach results show that deep neural networks with data augmentation can enhance the SER performance on the Pakistani racial speech dataset.

show abstract

Section: Related Workmentioning

confidence: 97%

Section: Introductionmentioning

confidence: 99%

Data augmentation and deep neural networks for the classification of Pakistani racial speakers recognition

Amjad

Khan

Chang

2022

PeerJ Computer Science

View full text Add to dashboard Cite

show abstract

“…Video merupakan salah satu multimedia paling populer yang digunakan masyarakat umum [3]. Jadi sangat penting untuk membuat informasi berupa suara dalam video agar dapat dipahami oleh orang-orang yang memiliki masalah pendengaran yaitu dengan cara paling alami terletak pada penggunaan subtitle.…”

Section: A Videounclassified

Pengenalan Lirik Lagu Otomatis Pada Video Lagu Indonesia Menggunakan Hidden Markov Model Yang Dilengkapi Music Removal

Tirta

Santoso

Setyati

2022

INSIGHT

View full text Add to dashboard Cite

Video sangat penting untuk membuat informasi berupa suara dalam video agar dapat dipahami oleh semua kalangan masyarakat, dan orang-orang yang memiliki masalah pendengaran yaitu dengan cara paling alami terletak pada penggunaan subtitle. Oleh karena itu, peneliti membuat pengenalan lirik lagu otomatis pada video lagu Indonesia menggunakan Hidden Markov Model yang dilengkapi music removal. Dalam pengenalan suara lebih akurat dilakukan dengan menggunakan model HMM yang dilengkapi oleh MFCC (kata yang cocok 81% dan WER 19%) dibandingkan dengan model LDA + MFCC (kata yang cocok 71% dan WER 29%) dan DWT + MFCC (kata yang cocok 61% dan WER 39%). Jumlah kata dan sample suara pada library Bahasa Indonesia yang digunakan cukup sangat mempengaruhi MFCC dan CMU Sphinx-4, Nada pada inputan lagu yang akan diproses CMU Sphinx-4 juga sangat berpengaruh pada tingkat keberhasilan, dikarenakan CMU Sphinx-4 sangat sensitif dengan nada yang terlalu tinggi dan noise yang ada pada inputan lagu tersebut sehingga peneliti menambahkan fitur ekstraksi pada suara yaitu menggunakan MFCC. Dalam hal ini menggunakan dataset kecil terlebih dahulu untuk memastikan metode Hidden Markov Model yang dilengkapi MFCC dan CMU Sphinx-4 dapat berjalan dengan baik, Dari penelitian beberapa peneliti sebelumnya, maka hasil akhir yang diperoleh dengan menggunakan metode HMM yang dilengkapi oleh MFCC dan CMU Sphinx-4 dalam penelitian ini mendapatkan hasil akurasi training 78% dan testing 81% kecocokan kata pada video lagu.

show abstract

“…Since the mid-1980s, MFCCs are the most widely used feature extraction method in the field of ASR. Most of the works concerning Moroccan Darija Speech recognition have used the MFCC as a feature extraction method [7], [8]. The main purpose of this feature extraction method is to mimic the human ear.…”

Section: Speech Features Extraction Techniquesmentioning

confidence: 99%

Automatic Speech Recognition Features Extraction Techniques: A Multi-criteria Comparison

Labied¹,

Belangour²

2021

IJACSA

View full text Add to dashboard Cite

Features extraction is an important step in Automatic Speech Recognition, which consists of determining the audio signal components that are useful for identifying linguistic content while removing background noise and irrelevant information. The main objective of features extraction is to identify the discriminative and robust features in the acoustic data. The derived feature vector should possess the characteristics of low dimensionality, long-time stability, nonsensitivity to noise, and no correlation with other features, which makes the application of a robust feature extraction technique a significant challenge for Automatic Speech Recognition. Many comparative studies have been carried out to compare different speech recognition feature extraction techniques, but none of them have evaluated the criteria to be considered when applying a feature extraction technique. The objective of this work is to answer some of the questions that may arise when considering which feature extraction techniques to apply, through a multicriteria comparison of different features extraction techniques using the Weighted Scoring Method.

show abstract

Speech Recognition of Moroccan Dialect Using Hidden Markov Models

Cited by 21 publications

References 8 publications

Data augmentation and deep neural networks for the classification of Pakistani racial speakers recognition

Data augmentation and deep neural networks for the classification of Pakistani racial speakers recognition

Pengenalan Lirik Lagu Otomatis Pada Video Lagu Indonesia Menggunakan Hidden Markov Model Yang Dilengkapi Music Removal

Automatic Speech Recognition Features Extraction Techniques: A Multi-criteria Comparison

Contact Info

Product

Resources

About