Digit recognition using wavelet and SVM in Brazilian Portuguese

Bresolin, Adriano de Andrade; Neto, Adrião Duarte Dória; Alsina, Pablo Javier

doi:10.1109/icassp.2008.4517917

Cited by 13 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Nowadays it has getting a great interest in the various scientific communities, machine learning community, regression and learning [8,9]. The theory of SVM was first introduced by Vapnik [8].…”

Section: A Support Vector Machinementioning

confidence: 99%

“…By mapping the input samples into a high dimensional space, SVM learns the boundary regions between samples which belong to two classes and after that it seek a separating hyper plane or set of hyper planes are used for classification, regression and this is chosen in such a way that it maximizes its distance to the closest training samples [8,10]. …”

Section: A Support Vector Machinementioning

confidence: 99%

See 1 more Smart Citation

Speaker Identification System for Hindi And Marathi Languages using Wavelet and Support Vector Machine

Hibare¹

2015

IJRITCC

View full text Add to dashboard Cite

Abstract-In this paper, a speaker identification system using speech processing for Hindi and Marathi languages is developed. Database of common words between Hindi and Marathi languages whose script is common but pronunciation is different is created. Here feature extraction is performed by using Wavelet Packet Decomposition (WPD) and classification is performed by using Support Vector Machine (SVM). As compared to the conventional feature extraction techniques wavelet transform is very much suitable for processing speech signals which are non-stationary in nature because of its efficient time frequency localizations and multi-resolution characteristics. Also SVM is well suitable for addressing speaker identification task. Recognition accuracy of 99.77% is obtained whereas real time recognition accuracy of 84.66% is obtained in identical condition using this hybrid architecture of WPD and SVM. In noisy conditions recognition accuracy of 60% is obtained.

show abstract

Section: A Support Vector Machinementioning

confidence: 99%

Section: A Support Vector Machinementioning

confidence: 99%

Speaker Identification System for Hindi And Marathi Languages using Wavelet and Support Vector Machine

Hibare¹

2015

IJRITCC

View full text Add to dashboard Cite

show abstract

“…Other model architecture like Long Short-Term Memory Recurrent Neural Networks [43] which, in contrast to conventional Recurrent Neural Networks, consider longrange dependencies between the observations was recently proven to be well suited for speech recognition [44]. Even static classifiers like Support Vector Machines have been successfully applied in isolated word recognition tasks [45], where a warping of the observation sequence is less essential than in continuous speech recognition.…”

Section: 4mentioning

confidence: 99%

Recognition of Noisy Speech: A Comparative Survey of Robust Model Architecture and Feature Enhancement

Schuller¹,

Wöllmer²,

Moosmayr³

et al. 2009

EURASIP Journal on Audio, Speech, and Music Processing

View full text Add to dashboard Cite

Performance of speech recognition systems strongly degrades in the presence of background noise, like the driving noise inside a car. In contrast to existing works, we aim to improve noise robustness focusing on all major levels of speech recognition: feature extraction, feature enhancement, speech modelling, and training. Thereby, we give an overview of promising auditory modelling concepts, speech enhancement techniques, training strategies, and model architecture, which are implemented in an in-car digit and spelling recognition task considering noises produced by various car types and driving conditions. We prove that joint speech and noise modelling with a Switching Linear Dynamic Model (SLDM) outperforms speech enhancement techniques like Histogram Equalisation (HEQ) with a mean relative error reduction of 52.7% over various noise types and levels. Embedding a Switching Linear Dynamical System (SLDS) into a Switching Autoregressive Hidden Markov Model (SAR-HMM) prevails for speech disturbed by additive white Gaussian noise.

show abstract

“…O reconhecimento de dígitos em língua portuguesaé abordado no trabalho dede Andrade Bresolin et al (2008). Esses autores utilizam uma base proprietária, composta por vozes masculinas.…”

unclassified

Reconhecimento de Dígitos de Medidores de Energia por meio da Voz no Contexto de um Aplicativo de Autoleitura

Mendes

Santos²,

Rocha

et al. 2019

Anais Do 14º Simpósio Brasileiro De Automação Inteligente

View full text Add to dashboard Cite

According to Brazilian Electricity Regulatory Agency (ANEEL), non-technical losses are related to consumption reading hindrances. In order to reduce these failures, a feasible and low cost alternative would be the reading done by customers which is called selfreading. This process includes the use of digital platforms and the customer would register and send consumption information. An important step of this process is the digits automatic recognition in energy meters through voice. This work proposes a method to perform this task using audio processing and machine learning. Mel-frequency Cepstral Coefficients (MFCC) and MelSpectrogram are used for feature extraction. The proposed method obtains 94.74% Recall, 94.91% Precision, 94,72% F1 score and 0,9419 Kappa Index using Support Vector Machine (SVM) classifier. Resumo: A Agência Nacional de Energia Elétrica (ANEEL) destaca que perdas não-técnicas estão relacionadas a entraves no processo de leitura de consumo. Para a redução dessas falhas, uma alternativa factível e de menor custo seria a leitura realizada pelo próprio consumidor, denominada de autoleitura. Este processo leva em consideração o uso de plataformas digitais, através das quais o consumidor registraria e enviaria as informações de consumo. Uma etapa importante desse processoé o reconhecimento automático de dígitos de medidores por meio da voz. Este trabalho, portanto, propõe um método para a realização dessa tarefa, que utiliza processamento deáudio e inteligência computacional. Para a extração de características deáudio, utiliza-se Mel-frequency Cepstral Coefficients (MFCC) e MelSpectrogram de forma combinada. O método apresenta Recall de 94,74%; Precision de 94,91%; F1 score de 94,72% e 0,9419 deíndice Kappa utilizando-se o classificador Support Vector Machine (SVM).

show abstract

Digit recognition using wavelet and SVM in Brazilian Portuguese

Cited by 13 publications

References 11 publications

Speaker Identification System for Hindi And Marathi Languages using Wavelet and Support Vector Machine

Speaker Identification System for Hindi And Marathi Languages using Wavelet and Support Vector Machine

Recognition of Noisy Speech: A Comparative Survey of Robust Model Architecture and Feature Enhancement

Reconhecimento de Dígitos de Medidores de Energia por meio da Voz no Contexto de um Aplicativo de Autoleitura

Contact Info

Product

Resources

About