Discrete Wavelet Transforms and Artificial Neural Networks for Recognition of Isolated Spoken Words

Sunny, Sonia; S, David Peter; Jacob, K. Poulose

doi:10.5120/4634-6871

Cited by 11 publications

(5 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The extraction of characteristics was carried out by means of the Wavelet transform method. Later, they carried out the training of an Artificial Neural Network with an accuracy of 89% [26].…”

Section: Related Workmentioning

confidence: 99%

Implementación de un sistema de reconocimiento de voz en el lenguaje Nasa Yuwe basado en Redes Neuronales Convolusionales

Burbano¹,

Gómez²,

Caicedo³

2023

ing. Solidar

View full text Add to dashboard Cite

Introducción: Este artículo presenta la Implementación de un algoritmo para el reconocimiento de voz en el lenguaje Nasa Yuwe basado en Redes Neuronales Convolusionales (RNC), desarrollado en la Universidad del Cauca en el año 2022. Problema: La riqueza fonética del lenguaje Nasa Yuwe es grande, al poseer 32 vocales, y 34 consonantes, lo que lleva a confusiones en la pronunciación y por lo tanto a dificultades en el reconocimiento de patrones de voz. Objetivo: Implementar un algoritmo de reconocimiento de voz, para el lenguaje Nasa Yuwe soportado en RNC. Metodología: Se realizó el preprocesamiento de las señales de audio para posteriormente obtener las características por medio de los escalogramas de los coeficientes de Mel. Finalmente se propone una arquitectura de la RNC para el proceso de clasificación. Resultados: Se construye un DataSet a partir de los escalograma de los patrones de voz, y se realiza el proceso de entrenamiento de la RNC. Conclusión: La implementación de un SRV basado RNC, proporciona bajos márgenes de error en el proceso de clasificación de palabras del lenguaje Nasa Yuwe. Originalidad: El sistema de reconocimiento de voz planteado es el primero y único en su clase que se ha realizado hasta el momento, con el propósito de colaborar en el proceso de enseñanza, conservación y aprendizaje del lenguaje Nasa Yuwe. Limitaciones: Se requiere aumentar el número de patrones de voz aportados por hablantes nativos, y se plantea la necesidad de implementar otras herramientas tecnologías que permitan la conservación y difusión del lenguaje Nasa Yuwe.

show abstract

Section: Related Workmentioning

confidence: 99%

Implementación de un sistema de reconocimiento de voz en el lenguaje Nasa Yuwe basado en Redes Neuronales Convolusionales

Burbano¹,

Gómez²,

Caicedo³

2023

ing. Solidar

View full text Add to dashboard Cite

show abstract

“…The combination of DWT and ANN proved effective in feature extraction from speech signals for automatic speech recognition, significantly reducing computational complexity and feature vector size. The result was an impressive recognition accuracy of 90% [8].…”

Section: Literature Reviewmentioning

confidence: 99%

Enhancing Emotion Classification in Malayalam Accented Speech: An In-Depth Clustering Approach

Thandil,

K.P

2023

JAZ

View full text Add to dashboard Cite

Accurate emotion classification in accented speech for the Malayalam language poses a unique challenge in the realm of speech recognition. In this study, we explore the application of various clustering algorithms to this specific dataset, evaluating their effectiveness using the Silhouette Score as a measure of cluster quality. Our findings reveal significant insights into the performance of these algorithms. Among the clustering methods, Affinity Propagation emerged as the frontrunner, achieving the highest Silhouette Score of 0.5255. This result indicates a superior cluster quality characterized by well-defined and distinct groups. OPTICS and Mean Shift Clustering also demonstrated strong performance with scores of 0.4029 and 0.2511, respectively, indicating the presence of relatively distinct and well-formed clusters. In addition, we introduced Ensemble Clustering (Majority Voting), which achieved a score of 0.2399, indicating moderate cluster distinction. These findings provide a valuable perspective on the potential advantages of ensemble methods in this context. Our experiment results shed light on the effectiveness of various clustering methods in the context of emotion classification in accented Malayalam speech. This study contributes to the advancement of speech recognition technology and lays the groundwork for further research in this area..

show abstract

“…While conventional computers use a very fast & complex central processor with explicit program instructions and locally addressable memory, by contrast the human brain uses a massively parallel collection of slow & simple processing elements (neurons), densely connected by weights (synapses) whose strengths are modified with experience, directly supporting the integration of multiple const raints, and providing a distributed form of associative memory. ANN is adaptive in nature where learning by examples replaces programming in solving problems [17]. It can process information in parallel, at a very high speed, and in a distributed manner.…”

Section: Artificial Neural Networkmentioning

confidence: 99%

Classification of the Spoken Hindi Partially Reduplicated Words using Artificial Neural Network

Gupta¹,

Sharma²

2014

IJCA

View full text Add to dashboard Cite

The most ordinary way of information exchange is Speech. It provides an efficient way of man-machine communication using speech interfacing. Speech interfacing involves two process, speech synthesis and speech recognition. Speech recognition allows a computer to identify the words that a person speaks to a microphone or telephone. The two main mechanism, used in speech recognition, are signal processing mechanism at front-end and pattern matching mechanism at back-end. In this paper, a setup for recognition of Spoken Hindi Partially Reduplicated Words (SHPRW), that uses Mel frequency cepstral coefficients at front-end and artificial neural networks at back-end has been developed to perform the experiment.

show abstract

Discrete Wavelet Transforms and Artificial Neural Networks for Recognition of Isolated Spoken Words

Cited by 11 publications

References 13 publications

Implementación de un sistema de reconocimiento de voz en el lenguaje Nasa Yuwe basado en Redes Neuronales Convolusionales

Implementación de un sistema de reconocimiento de voz en el lenguaje Nasa Yuwe basado en Redes Neuronales Convolusionales

Enhancing Emotion Classification in Malayalam Accented Speech: An In-Depth Clustering Approach

Classification of the Spoken Hindi Partially Reduplicated Words using Artificial Neural Network

Contact Info

Product

Resources

About