Improved text-independent speaker identification system for real time applications

AboElenein, Nagwa M.; Amin, Khalid M.; Ibrahim, Mina; Hadhoud, Mohiy M.

doi:10.1109/jec-ecc.2016.7518967

Cited by 15 publications

(9 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results of the FreqCNN model, i-vector and typical CNN models with three-fold cross-validation are given in Table 8. Compared with the traditional method [17] using MFCCs with vector quantization and gaussian mixture model, which obtains an accuracy of 91.00%, our model improves accuracy around +7% and obtains a UAR of 98.05% on average. We also tested i-vector and Table 5 Comparison of two ways of attention methods in CNNs and different numbers of attention-based blocks.…”

Section: Speaker Identificationmentioning

confidence: 88%

“…We further explored the performance of the model using a different number of frequencydistributed set and two attention methods. In the speaker identification task, we compared our model with the traditional method (i.e., MFCCs) [17] and typical CNN models. We also tested the FreqCNN model under different activation functions.…”

Section: Methodsmentioning

confidence: 99%

“…A well-trained text-independent identification model can recognize a speaker from any text, even without training. In this experiment, we tested the FreqCNN model of a text-independent task on the CHAINS corpus [17].…”

Section: Speaker Identificationmentioning

confidence: 99%

“…However, spectrograms also represent the distribution of energy along the change of frequency. In many approaches [16,17], frequency-domain only features (e.g., MFCCs) also obtain good results in audio classification tasks, which have proved the importance of frequencies information. Beside, several works have researched the effectiveness of sub-band spectral feature [18][19][20], which enhanced the representation of speech signals and obtained better robustness in many tasks, such as automatic speech recognition (ASR).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Audio classification using attention-augmented convolutional neural network

Mao

Zhang

2018

Knowledge-Based Systems

View full text Add to dashboard Cite

2018) Audio classification using attention-augmented convolutional neural network. Published by: Elsevier URL: http://dx. A B S T R A C TAudio classification, as a set of important and challenging tasks, groups speech signals according to speakers' identities, accents, and emotional states. Due to the high dimensionality of the audio data, task-specific handcrafted features extraction is always required and regarded cumbersome for various audio classification tasks. More importantly, the inherent relationship among features has not been fully exploited. In this paper, the original speech signal is first represented as spectrogram and later be split along the frequency domain to form frequency-distributed spectrogram. This paper proposes a task-independent model, called FreqCNN, to automaticly extract distinctive features from each frequency band by using convolutional kernels. Further more, an attention mechanism is introduced to systematically enhance the features from certain frequency bands. The proposed FreqCNN is evaluated on three publicly available speech databases thorough three independent classification tasks. The obtained results demonstrate superior performance over the state-of-the-art.

show abstract

Section: Speaker Identificationmentioning

confidence: 88%

Section: Methodsmentioning

confidence: 99%

Section: Speaker Identificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Audio classification using attention-augmented convolutional neural network

Mao

Zhang

2018

Knowledge-Based Systems

View full text Add to dashboard Cite

show abstract

“…To reduce the number of speaker during a test stage, gender detection algorithm is used. Experimental results show that the suggested algorithm reduced the time testing to almost half 0.l0 51sec and gives 91% accuracy in comparison with VQ and GMM gives 88% and take 0.2242sec [15]. www.ijacsa.thesai.org They presented an automatic speaker identification system (SID) based on Gaussian Mixture Model and Support Vector Machines (GMM-SVM), where data set consist of 360 speakers.…”

Section: Related Workmentioning

confidence: 99%

Speaker Identification based on Hybrid Feature Extraction Techniques

AbuAladas¹,

Akram²,

Shaban³

et al. 2019

ijacsa

View full text Add to dashboard Cite

One of the most exciting areas of signal processing is speech processing; speech contains many features or characteristics that can discriminate the identity of the person. The human voice is considered one of the important biometric characteristics that can be used for person identification. This work is concerned with studying the effect of appropriate extracted features from various levels of discrete wavelet transformation (DWT) and the concatenation of two techniques (discrete wavelet and curvelet transform) and study the effect of reducing the number of features by using principal component analysis (PCA) on speaker identification. Backpropagation (BP) neural network was also introduced as a classifier.

show abstract