Accent Classification of the Three Major Nigerian Indigenous Languages Using 1D CNN LSTM Network Model

Salau, Ayodeji Olalekan; Olowoyo, Tilewa David; Akinola, Solomon Oluwole

doi:10.1007/978-981-15-2620-6_1

Cited by 15 publications

(9 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In most cases, an inverse proportional relationship exists between Precision and Recall, and therefore, a harmonic mean of these two metrics gives a better estimate of the model's performance. This metric is known as F 1 score or F-measure which is calculated according to Equation (7).…”

Section: Evaluation Metricsmentioning

confidence: 99%

See 1 more Smart Citation

How many Mel‐frequency cepstral coefficients to be utilized in speech recognition? A study with the Bengali language

Hasan

Hossain

2021

The Journal of Engineering

View full text Add to dashboard Cite

Speech‐related research has a wide range of applications. Most speech‐related researches employ Mel‐frequency cepstral coefficients (MFCCs) as acoustic features. However, finding the optimum number of MFCCs is an active research question. MFCC‐based speech classification was performed for both vowels and words in the Bengali language. As for the classification model, deep neural network (DNN) with Adam optimizer was used. The performances were measured with five different performance metrics, namely confusion matrix, classification accuracy, area under curve of receiver operating characteristic (AUC‐ROC), F1 score, and Cohen's Kappa with four‐fold cross‐validations at different number of MFCCs. All performance metrics gave the best score for 24/25 MFCCs; hence it is suggested that the optimum number of MFCCs should be 25, although many existing studies use only 13 MFCCs. Furthermore, it is verified that increasing the number of MFCCs yields better classification metrics with lower computational burden than the increment of hidden layers. Lastly, the optimum number of MFCCs obtained from this study was used in a more improved DNN model, from which 99% and 90% accuracies were achieved for vowel and word classification, respectively, and the vowel classification score outperformed state‐of‐the‐art results.

show abstract

Section: Evaluation Metricsmentioning

confidence: 99%

“…Furthermore, Salau et al. [7] performed accent classification of three Nigerian languages using MFCC features. From their comparison with similar studies, it is evident that MFCC is the most accepted feature.…”

Section: Introductionmentioning

confidence: 99%

How many Mel‐frequency cepstral coefficients to be utilized in speech recognition? A study with the Bengali language

Hasan

Hossain

2021

The Journal of Engineering

View full text Add to dashboard Cite

show abstract

“…Por ser uma das línguas mais faladas no mundo (Shi et al, 2021), vários trabalhos estudam os sotaques na língua inglesa (Ahmed et al, 2019;Wang et al, 2020;Zhang et al, 2021). Encontram-se artigos sobre sotaque árabe (Biadsy et al, 2009), sotaque francês (Lazaridis et al, 2014), sotaque em mandarim (Weninger et al, 2019), dialetos da Nigéria Salau et al (2020), dentre outros. Em português, destacam-se o trabalho realizado por Ynoguti (1999) e por Batista et al (2019).…”

Section: Introdu ç ãOunclassified

“…Tal como no trabalho de Batista et al (2019), o problema investigado neste trabalho é a classificação dos sinais de áudio nos sete diferentes sotaques brasileiros. Diferente do trabalho anterior, contudo, a proposta é usar arquiteturas de redes neurais para a tarefa, uma vez que este tipo de modelo têm apresentado boa performance em artigos recentes de classificação de sotaques (Badhon et al, 2021;Weninger et al, 2019;Wang et al, 2020;Salau et al, 2020;Ahmed et al, 2019;Wu et al, 2018).…”

Section: Introdu ç ãOunclassified

See 1 more Smart Citation

Classificação de Sotaques Brasileiros usando Redes Neurais Profundas

Tostes

Boldt

Komati

et al. 2021

Procedings Do XV Simpósio Brasileiro De Automação Inteligente

View full text Add to dashboard Cite

The automatic classification of accents has several potential applications, for instance, the identification and authentication of users, forensic investigation tools and the selection of specialized models in text-to-speech and speech-to-text systems. In this work, we propose and evaluate several architectures of artificial neural networks for accent classification. The performance of these architectures in the Braccent dataset was compared with the methods GMM-UBM, GMM-SVM and iVector. Experimental results show that 4 out of 6 architectures achieve better values of accuracy, precision and recall than the previous methods. The best architecture reached 90% of accuracy, with precision, recall and F1-score of 0.92, 0.84 and 0.87, respectively. Resumo: A classificação automática de sotaques possui diversas aplicações potenciais como a identificação e autenticação de usuários, ferramentas de investigação forense e a seleção de modelos especializados para text-to-speech e speech-to-text. Neste trabalho, propomos e avaliamos diversas arquiteturas de redes neurais artificiais para classificação de sotaques. A performance das arquiteturas na base de dados Braccent foi comparada com os métodos GMM-UBM, GMM-SVM e iVector. Resultados experimentais mostram que 4 das 6 arquiteturas alcançam valores melhores de acurácia, precisão e revocação que os métodos anteriores. A melhor arquitetura alcançou 90% de acerto, com precisão, revocação e F1-score de 0.92, 0.84 e 0.87, respectivamente.

show abstract

Bengali Accent Classification from Speech Using Different Machine Learning and Deep Learning Techniques

Badhon

Rahaman

Rupon

et al. 2020

Advances in Intelligent Systems and Computing

View full text Add to dashboard Cite

Accent Classification of the Three Major Nigerian Indigenous Languages Using 1D CNN LSTM Network Model

Cited by 15 publications

References 21 publications

How many Mel‐frequency cepstral coefficients to be utilized in speech recognition? A study with the Bengali language

How many Mel‐frequency cepstral coefficients to be utilized in speech recognition? A study with the Bengali language

Classificação de Sotaques Brasileiros usando Redes Neurais Profundas

Bengali Accent Classification from Speech Using Different Machine Learning and Deep Learning Techniques

Contact Info

Product

Resources

About