Speech Enhancement Algorithm Using Recursive Wavelet Shrinkage

Lee, Gihyoun; Na, Sung Dae; Seong, Ki Woong; Cho, Jin‐Ho; Kim, Myoung Nam

doi:10.1587/transinf.2015edl8251

Cited by 6 publications

(5 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the evolution, the high-frequency side of the received audio signal will drop by 6 dB per octave after it is generated, while the noise signal is the opposite. This makes the low-noise signal of the speech signal larger and the low-frequency signal reaching the speaker is smaller, which makes the transmission difficult [ 29 ]. In order to solve this problem, advanced technology will be used in the early stage of the audio signal to increase the high-frequency component of the audio signal to compensate for the loss.…”

Section: Research Methods Of Speech Recognition Technology In Music S...mentioning

confidence: 99%

Improvement of Speech Recognition Technology in Piano Music Scene Based on Deep Learning of Internet of Things

Bai

2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

The main goal of speech recognition technology is to use computers to convert human analog speech signals into computer-generated signals, such as behavior patterns or binary codes. Different from speaker identification and speaker confirmation, the latter attempts to identify or confirm the speaker who uttered the speech rather than the lexical content contained in it. The short-term idea is that it should be able to record the musical sound played by the user with a certain musical instrument, then extract the note and duration information from it, and finally generate the corresponding MID file according to the MIDI standard, which can set the type of musical instrument in advance to complete the function of musical sound transformation, such as playing with a harmonica, and playing the MID at the end is the piano sound. With the rapid development of the mobile Internet, fields such as machine learning, electronic communication, and navigation have placed high demands on real-time and standard text recognition technology. This paper merges the sound of visual music into text-based data set training, uses the exported scanner features for model training, uses the model to extract features, then uses the features for prior training, and then uses pretraining. DNN results show that the combined training of target prevention and expansion plans, by replacing long-term and short-term memory networks, end-to-end speech recognition programs, and behavioral tests organized by mobile devices, can provide a larger receptive field combined with expanded convolution instead of long and short periods. The experimental results show that when the input sampling point is 2400, it can be seen that the convergence speed of the model becomes slower with more than 90 iterations and the loss of the model on the verification set increases with the increase in the number of iterations. This shows that the model in this paper can fully meet the needs of speech recognition in piano music scenes.

show abstract

Section: Research Methods Of Speech Recognition Technology In Music S...mentioning

confidence: 99%

Improvement of Speech Recognition Technology in Piano Music Scene Based on Deep Learning of Internet of Things

Bai

2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

show abstract

“…It is possible to resolve high-frequency components within a small time window of the signal [ 32 ]. Generally, WT is employed to decompose a signal by transforming a wavelet packet into time–frequency wavelet coefficients of multiple sub-bands [ 33 , 34 ]. In this study, WT decomposition was designed to represent the time–frequency form of the fNIRS signal using the Daubechies6 wavelet basis.…”

Section: Methodsmentioning

confidence: 99%

Motion Artifact Correction of Multi-Measured Functional Near-Infrared Spectroscopy Signals Based on Signal Reconstruction Using an Artificial Neural Network

Lee

Jin

2018

Sensors

Self Cite

View full text Add to dashboard Cite

In this paper, a new motion artifact correction method is proposed based on multi-channel functional near-infrared spectroscopy (fNIRS) signals. Recently, wavelet transform and hemodynamic response function-based algorithms were proposed as methods of denoising and detrending fNIRS signals. However, these techniques cannot achieve impressive performance in the experimental environment with lots of movement such as gait and rehabilitation tasks because hemodynamic responses have features similar to those of motion artifacts. Moreover, it is difficult to correct motion artifacts in multi-measured fNIRS systems, which have multiple channels and different noise features in each channel. Thus, a new motion artifact correction method for multi-measured fNIRS is proposed in this study, which includes a decision algorithm to determine the most contaminated fNIRS channel based on entropy and a reconstruction algorithm to correct motion artifacts by using a wavelet-decomposed back-propagation neural network. The experimental data was achieved from six subjects and the results were analyzed in comparing conventional algorithms such as HRF smoothing, wavelet denoising, and wavelet MDL. The performance of the proposed method was proven experimentally using the graphical results of the corrected fNIRS signal, CNR that is a performance evaluation index, and the brain activation map.

show abstract

“…24 The structure of the critical bands in WPD is optimized to compartmentalize fNIRS signal bands. For a given wavelet packet level j , the wavelet packet transform decomposes the fNIRS signal y ( n ) into 2 j sub-bands corresponding to wavelet coefficient sets w j,m ( k ) 25 as …”

Section: Theory and Methodsmentioning

confidence: 99%

Robust functional near infrared spectroscopy denoising using multiple wavelet shrinkage based on a hemodynamic response model

Lee

Jin

et al. 2018

Journal of Near Infrared Spectroscopy

Self Cite

View full text Add to dashboard Cite

Functional near infrared spectroscopy can measure hemodynamic signals, and the results are similar to functional magnetic resonance imaging of blood-oxygen-level-dependent signals. Thus, functional near infrared spectroscopy can be employed to investigate brain activity by measuring the absorption of near infrared light through an intact skull. Recently, a general linear model, which is a standard method for functional magnetic resonance imaging, was applied to functional near infrared spectroscopy imaging analysis. However, the general linear model fails when functional near infrared spectroscopy signals retain noise, such as that caused by the subject's movement during measurement. Although wavelet-based denoising and hemodynamic response function smoothing are popular denoising methods for functional near infrared spectroscopy signals, these methods do not exhibit impressive performances for very noisy environments and a specific class of noise. Thus, this paper proposes a new denoising algorithm that uses multiple wavelet shrinkage and a multiple threshold function based on a hemodynamic response model. Through the experiments, the performance of the proposed algorithm is verified using graphic results and objective indexes, and it is compared with existing denoising algorithms.

show abstract

Speech Enhancement Algorithm Using Recursive Wavelet Shrinkage

Cited by 6 publications

References 4 publications

Improvement of Speech Recognition Technology in Piano Music Scene Based on Deep Learning of Internet of Things

Improvement of Speech Recognition Technology in Piano Music Scene Based on Deep Learning of Internet of Things

Motion Artifact Correction of Multi-Measured Functional Near-Infrared Spectroscopy Signals Based on Signal Reconstruction Using an Artificial Neural Network

Robust functional near infrared spectroscopy denoising using multiple wavelet shrinkage based on a hemodynamic response model

Contact Info

Product

Resources

About