SlideShare a Scribd company logo
STATE-CLUSTERING BASED MULTIPLE DEEP NEURAL NETWORKS MODELING
APPROACH FOR SPEECH RECOGNITION
ABSTRACT
The hybrid deep neural network (DNN) and hidden Markov model (HMM) has recently
achieved dramatic performance gains in automatic speech recognition (ASR). The DNN-based
acoustic model is very powerful but its learning process is extremely time-consuming. In this
paper, we propose a novel DNN-based acoustic modeling framework for speech recognition,
where the posterior probabilities of HMM states are computed from multiple DNNs (mDNN),
instead of a single large DNN, for the purpose of parallel training towards faster turnaround. In
the proposed mDNN method all tied HMM states are first grouped into several disjoint clusters
based on data-driven methods. Next, several hierarchically structured DNNs are trained
separately in parallel for these clusters using multiple computing units (e.g. GPUs). In decoding,
the posterior probabilities of HMM states can be calculated by combining outputs from multiple
DNNs. In this work, we have shown that the training procedure of the mDNN under popular
criteria, including both frame-level cross-entropy and sequence-level discriminative training, can
be parallelized efficiently to yield significant speedup. The training speedup is mainly attributed
to the fact that multiple DNNs are parallelized over multiple GPUs and each DNN is smaller in
size and trained by only a subset of training data. We have evaluated the proposed mDNN
method on a 64-hour Mandarin transcription task and the 320-hour Switchboard task. Compared
to the conventional DNN, a 4-cluster mDNN model with similar size can yield comparable
recognition performance in Switchboard (only about 2% performance degradation) with a greater
than 7 times speed improvement in CE training and a 2.9 times improvement in sequence
training, when 4 GPUs are used.

More Related Content

What's hot

Deep belief networks for spam filtering
Deep belief networks for spam filteringDeep belief networks for spam filtering
Deep belief networks for spam filtering
SOYEON KIM
 
Icacci2017 poster template
Icacci2017 poster templateIcacci2017 poster template
Icacci2017 poster template
vinaykumar R
 
Cnn
CnnCnn
An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding   An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding
IJCERT
 
Lecture 7: Recurrent Neural Networks
Lecture 7: Recurrent Neural NetworksLecture 7: Recurrent Neural Networks
Lecture 7: Recurrent Neural Networks
Sang Jun Lee
 
HRNET : Deep High-Resolution Representation Learning for Human Pose Estimation
HRNET : Deep High-Resolution Representation Learning for Human Pose EstimationHRNET : Deep High-Resolution Representation Learning for Human Pose Estimation
HRNET : Deep High-Resolution Representation Learning for Human Pose Estimation
taeseon ryu
 
Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)
Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)
Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)
Universitat Politècnica de Catalunya
 
Applying Deep Learning Machine Translation to Language Services
Applying Deep Learning Machine Translation to Language ServicesApplying Deep Learning Machine Translation to Language Services
Applying Deep Learning Machine Translation to Language Services
Yannis Flet-Berliac
 
Deep Belief Networks
Deep Belief NetworksDeep Belief Networks
Deep Belief Networks
Hasan H Topcu
 

What's hot (11)

Deep belief networks for spam filtering
Deep belief networks for spam filteringDeep belief networks for spam filtering
Deep belief networks for spam filtering
 
Icacci2017 poster template
Icacci2017 poster templateIcacci2017 poster template
Icacci2017 poster template
 
Data comparation
Data comparationData comparation
Data comparation
 
Cnn
CnnCnn
Cnn
 
An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding   An Image representation using Compressive Sensing and Arithmetic Coding
An Image representation using Compressive Sensing and Arithmetic Coding
 
Lecture 7: Recurrent Neural Networks
Lecture 7: Recurrent Neural NetworksLecture 7: Recurrent Neural Networks
Lecture 7: Recurrent Neural Networks
 
hara-san's research
hara-san's researchhara-san's research
hara-san's research
 
HRNET : Deep High-Resolution Representation Learning for Human Pose Estimation
HRNET : Deep High-Resolution Representation Learning for Human Pose EstimationHRNET : Deep High-Resolution Representation Learning for Human Pose Estimation
HRNET : Deep High-Resolution Representation Learning for Human Pose Estimation
 
Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)
Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)
Recurrent Neural Networks (D2L2 2017 UPC Deep Learning for Computer Vision)
 
Applying Deep Learning Machine Translation to Language Services
Applying Deep Learning Machine Translation to Language ServicesApplying Deep Learning Machine Translation to Language Services
Applying Deep Learning Machine Translation to Language Services
 
Deep Belief Networks
Deep Belief NetworksDeep Belief Networks
Deep Belief Networks
 

Viewers also liked

Download here
Download hereDownload here
Download here
Hung Nguyen Manh
 
Presentation overview of neural & kernel based clustering
Presentation overview of neural & kernel based clustering Presentation overview of neural & kernel based clustering
Presentation overview of neural & kernel based clustering
Shubham Vijay Vargiy
 
Introduction to Neural networks (under graduate course) Lecture 7 of 9
Introduction to Neural networks (under graduate course) Lecture 7 of 9Introduction to Neural networks (under graduate course) Lecture 7 of 9
Introduction to Neural networks (under graduate course) Lecture 7 of 9
Randa Elanwar
 
Unsupervised Feature Learning
Unsupervised Feature LearningUnsupervised Feature Learning
Unsupervised Feature LearningAmgad Muhammad
 
Autoencoders for image_classification
Autoencoders for image_classificationAutoencoders for image_classification
Autoencoders for image_classification
Cenk Bircanoğlu
 
Artificial Neural Networks Lect3: Neural Network Learning rules
Artificial Neural Networks Lect3: Neural Network Learning rulesArtificial Neural Networks Lect3: Neural Network Learning rules
Artificial Neural Networks Lect3: Neural Network Learning rules
Mohammed Bennamoun
 
Neural network & its applications
Neural network & its applications Neural network & its applications
Neural network & its applications
Ahmed_hashmi
 
Artificial neural network
Artificial neural networkArtificial neural network
Artificial neural networkDEEPASHRI HK
 

Viewers also liked (8)

Download here
Download hereDownload here
Download here
 
Presentation overview of neural & kernel based clustering
Presentation overview of neural & kernel based clustering Presentation overview of neural & kernel based clustering
Presentation overview of neural & kernel based clustering
 
Introduction to Neural networks (under graduate course) Lecture 7 of 9
Introduction to Neural networks (under graduate course) Lecture 7 of 9Introduction to Neural networks (under graduate course) Lecture 7 of 9
Introduction to Neural networks (under graduate course) Lecture 7 of 9
 
Unsupervised Feature Learning
Unsupervised Feature LearningUnsupervised Feature Learning
Unsupervised Feature Learning
 
Autoencoders for image_classification
Autoencoders for image_classificationAutoencoders for image_classification
Autoencoders for image_classification
 
Artificial Neural Networks Lect3: Neural Network Learning rules
Artificial Neural Networks Lect3: Neural Network Learning rulesArtificial Neural Networks Lect3: Neural Network Learning rules
Artificial Neural Networks Lect3: Neural Network Learning rules
 
Neural network & its applications
Neural network & its applications Neural network & its applications
Neural network & its applications
 
Artificial neural network
Artificial neural networkArtificial neural network
Artificial neural network
 

Similar to STATE-CLUSTERING BASED MULTIPLE DEEP NEURAL NETWORKS MODELING APPROACH FOR SPEECH RECOGNITION

Sequence learning and modern RNNs
Sequence learning and modern RNNsSequence learning and modern RNNs
Sequence learning and modern RNNs
Grigory Sapunov
 
Et25897899
Et25897899Et25897899
Et25897899
IJERA Editor
 
Literature Review
Literature ReviewLiterature Review
Literature Review
Sandeep Karthikeyan
 
Decentralized eigenvalue algorithms for distributed signal detection in wirel...
Decentralized eigenvalue algorithms for distributed signal detection in wirel...Decentralized eigenvalue algorithms for distributed signal detection in wirel...
Decentralized eigenvalue algorithms for distributed signal detection in wirel...ieeeprojectsbangalore
 
Convolutional Neural Network and Feature Transformation for Distant Speech Re...
Convolutional Neural Network and Feature Transformation for Distant Speech Re...Convolutional Neural Network and Feature Transformation for Distant Speech Re...
Convolutional Neural Network and Feature Transformation for Distant Speech Re...
IJECEIAES
 
Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...
Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...
Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...Pierre Laffitte
 
deeplearning
deeplearningdeeplearning
deeplearning
huda2018
 
Conformer review
Conformer reviewConformer review
Conformer review
June-Woo Kim
 
Three classes of deep learning networks
Three classes of deep learning networksThree classes of deep learning networks
Three classes of deep learning networks
Venkat Chaithanya Chintha
 
M phil-computer-science-signal-processing-projects
M phil-computer-science-signal-processing-projectsM phil-computer-science-signal-processing-projects
M phil-computer-science-signal-processing-projects
Vijay Karan
 
Signal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 ProjectsSignal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 Projects
Vijay Karan
 
Signal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 ProjectsSignal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 Projects
Vijay Karan
 
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Willy Marroquin (WillyDevNET)
 
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Willy Marroquin (WillyDevNET)
 
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
taeseon ryu
 
Occurrence Prediction_NLP
Occurrence Prediction_NLPOccurrence Prediction_NLP
Occurrence Prediction_NLP
Guttenberg Ferreira Passos
 
Implemetation of parallelism in HMM DNN based state of the art kaldi ASR Toolkit
Implemetation of parallelism in HMM DNN based state of the art kaldi ASR ToolkitImplemetation of parallelism in HMM DNN based state of the art kaldi ASR Toolkit
Implemetation of parallelism in HMM DNN based state of the art kaldi ASR ToolkitShubham Verma
 
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITIONQUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
ijma
 
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITIONQUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
ijma
 

Similar to STATE-CLUSTERING BASED MULTIPLE DEEP NEURAL NETWORKS MODELING APPROACH FOR SPEECH RECOGNITION (20)

Deep leaning Vincent Vanhoucke
Deep leaning Vincent VanhouckeDeep leaning Vincent Vanhoucke
Deep leaning Vincent Vanhoucke
 
Sequence learning and modern RNNs
Sequence learning and modern RNNsSequence learning and modern RNNs
Sequence learning and modern RNNs
 
Et25897899
Et25897899Et25897899
Et25897899
 
Literature Review
Literature ReviewLiterature Review
Literature Review
 
Decentralized eigenvalue algorithms for distributed signal detection in wirel...
Decentralized eigenvalue algorithms for distributed signal detection in wirel...Decentralized eigenvalue algorithms for distributed signal detection in wirel...
Decentralized eigenvalue algorithms for distributed signal detection in wirel...
 
Convolutional Neural Network and Feature Transformation for Distant Speech Re...
Convolutional Neural Network and Feature Transformation for Distant Speech Re...Convolutional Neural Network and Feature Transformation for Distant Speech Re...
Convolutional Neural Network and Feature Transformation for Distant Speech Re...
 
Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...
Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...
Deep Neural Networks for Automatic detection of Sreams and Shouted Speech in ...
 
deeplearning
deeplearningdeeplearning
deeplearning
 
Conformer review
Conformer reviewConformer review
Conformer review
 
Three classes of deep learning networks
Three classes of deep learning networksThree classes of deep learning networks
Three classes of deep learning networks
 
M phil-computer-science-signal-processing-projects
M phil-computer-science-signal-processing-projectsM phil-computer-science-signal-processing-projects
M phil-computer-science-signal-processing-projects
 
Signal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 ProjectsSignal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 Projects
 
Signal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 ProjectsSignal Processing IEEE 2015 Projects
Signal Processing IEEE 2015 Projects
 
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
 
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
Facial Expression Recognition Using Enhanced Deep 3D Convolutional Neural Net...
 
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
[딥논읽] Meta-Transfer Learning for Zero-Shot Super-Resolution paper review
 
Occurrence Prediction_NLP
Occurrence Prediction_NLPOccurrence Prediction_NLP
Occurrence Prediction_NLP
 
Implemetation of parallelism in HMM DNN based state of the art kaldi ASR Toolkit
Implemetation of parallelism in HMM DNN based state of the art kaldi ASR ToolkitImplemetation of parallelism in HMM DNN based state of the art kaldi ASR Toolkit
Implemetation of parallelism in HMM DNN based state of the art kaldi ASR Toolkit
 
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITIONQUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
 
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITIONQUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
QUALITATIVE ANALYSIS OF PLP IN LSTM FOR BANGLA SPEECH RECOGNITION
 

More from I3E Technologies

Add
AddAdd
Design of a low voltage low-dropout regulator
Design of a low voltage low-dropout regulatorDesign of a low voltage low-dropout regulator
Design of a low voltage low-dropout regulator
I3E Technologies
 
An efficient constant multiplier architecture based on vertical horizontal bi...
An efficient constant multiplier architecture based on vertical horizontal bi...An efficient constant multiplier architecture based on vertical horizontal bi...
An efficient constant multiplier architecture based on vertical horizontal bi...
I3E Technologies
 
Aging aware reliable multiplier design with adaptive hold logic
Aging aware reliable multiplier design with adaptive hold logicAging aware reliable multiplier design with adaptive hold logic
Aging aware reliable multiplier design with adaptive hold logic
I3E Technologies
 
A high performance fir filter architecture for fixed and reconfigurable appli...
A high performance fir filter architecture for fixed and reconfigurable appli...A high performance fir filter architecture for fixed and reconfigurable appli...
A high performance fir filter architecture for fixed and reconfigurable appli...
I3E Technologies
 
A generalized algorithm and reconfigurable architecture for efficient and sca...
A generalized algorithm and reconfigurable architecture for efficient and sca...A generalized algorithm and reconfigurable architecture for efficient and sca...
A generalized algorithm and reconfigurable architecture for efficient and sca...
I3E Technologies
 
A combined sdc sdf architecture for normal i o pipelined radix-2 fft
A combined sdc sdf architecture for normal i o pipelined radix-2 fftA combined sdc sdf architecture for normal i o pipelined radix-2 fft
A combined sdc sdf architecture for normal i o pipelined radix-2 fft
I3E Technologies
 
Reverse converter design via parallel prefix adders novel components, method...
Reverse converter design via parallel prefix adders  novel components, method...Reverse converter design via parallel prefix adders  novel components, method...
Reverse converter design via parallel prefix adders novel components, method...
I3E Technologies
 
Pre encoded multipliers based on non-redundant radix-4 signed-digit encoding
Pre encoded multipliers based on non-redundant radix-4 signed-digit encodingPre encoded multipliers based on non-redundant radix-4 signed-digit encoding
Pre encoded multipliers based on non-redundant radix-4 signed-digit encoding
I3E Technologies
 
Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...
Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...
Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...
I3E Technologies
 
Variable form carrier-based pwm for boost-voltage motor driver with a charge-...
Variable form carrier-based pwm for boost-voltage motor driver with a charge-...Variable form carrier-based pwm for boost-voltage motor driver with a charge-...
Variable form carrier-based pwm for boost-voltage motor driver with a charge-...
I3E Technologies
 
Ultrasparse ac link converters
Ultrasparse ac link convertersUltrasparse ac link converters
Ultrasparse ac link converters
I3E Technologies
 
Single inductor dual-output buck–boost power factor correction converter
Single inductor dual-output buck–boost power factor correction converterSingle inductor dual-output buck–boost power factor correction converter
Single inductor dual-output buck–boost power factor correction converter
I3E Technologies
 
Ripple minimization through harmonic elimination in asymmetric interleaved mu...
Ripple minimization through harmonic elimination in asymmetric interleaved mu...Ripple minimization through harmonic elimination in asymmetric interleaved mu...
Ripple minimization through harmonic elimination in asymmetric interleaved mu...
I3E Technologies
 
Resonance analysis and soft switching design of isolated boost converter with...
Resonance analysis and soft switching design of isolated boost converter with...Resonance analysis and soft switching design of isolated boost converter with...
Resonance analysis and soft switching design of isolated boost converter with...
I3E Technologies
 
Reliability evaluation of conventional and interleaved dc–dc boost converters
Reliability evaluation of conventional and interleaved dc–dc boost convertersReliability evaluation of conventional and interleaved dc–dc boost converters
Reliability evaluation of conventional and interleaved dc–dc boost converters
I3E Technologies
 
Power factor corrected zeta converter based improved power quality switched m...
Power factor corrected zeta converter based improved power quality switched m...Power factor corrected zeta converter based improved power quality switched m...
Power factor corrected zeta converter based improved power quality switched m...
I3E Technologies
 
Pfc cuk converter fed bldc motor drive
Pfc cuk converter fed bldc motor drivePfc cuk converter fed bldc motor drive
Pfc cuk converter fed bldc motor drive
I3E Technologies
 
Optimized operation of current fed dual active bridge dc dc converter for pv ...
Optimized operation of current fed dual active bridge dc dc converter for pv ...Optimized operation of current fed dual active bridge dc dc converter for pv ...
Optimized operation of current fed dual active bridge dc dc converter for pv ...
I3E Technologies
 
Online variable topology type photovoltaic grid-connected inverter
Online variable topology type photovoltaic grid-connected inverterOnline variable topology type photovoltaic grid-connected inverter
Online variable topology type photovoltaic grid-connected inverter
I3E Technologies
 

More from I3E Technologies (20)

Add
AddAdd
Add
 
Design of a low voltage low-dropout regulator
Design of a low voltage low-dropout regulatorDesign of a low voltage low-dropout regulator
Design of a low voltage low-dropout regulator
 
An efficient constant multiplier architecture based on vertical horizontal bi...
An efficient constant multiplier architecture based on vertical horizontal bi...An efficient constant multiplier architecture based on vertical horizontal bi...
An efficient constant multiplier architecture based on vertical horizontal bi...
 
Aging aware reliable multiplier design with adaptive hold logic
Aging aware reliable multiplier design with adaptive hold logicAging aware reliable multiplier design with adaptive hold logic
Aging aware reliable multiplier design with adaptive hold logic
 
A high performance fir filter architecture for fixed and reconfigurable appli...
A high performance fir filter architecture for fixed and reconfigurable appli...A high performance fir filter architecture for fixed and reconfigurable appli...
A high performance fir filter architecture for fixed and reconfigurable appli...
 
A generalized algorithm and reconfigurable architecture for efficient and sca...
A generalized algorithm and reconfigurable architecture for efficient and sca...A generalized algorithm and reconfigurable architecture for efficient and sca...
A generalized algorithm and reconfigurable architecture for efficient and sca...
 
A combined sdc sdf architecture for normal i o pipelined radix-2 fft
A combined sdc sdf architecture for normal i o pipelined radix-2 fftA combined sdc sdf architecture for normal i o pipelined radix-2 fft
A combined sdc sdf architecture for normal i o pipelined radix-2 fft
 
Reverse converter design via parallel prefix adders novel components, method...
Reverse converter design via parallel prefix adders  novel components, method...Reverse converter design via parallel prefix adders  novel components, method...
Reverse converter design via parallel prefix adders novel components, method...
 
Pre encoded multipliers based on non-redundant radix-4 signed-digit encoding
Pre encoded multipliers based on non-redundant radix-4 signed-digit encodingPre encoded multipliers based on non-redundant radix-4 signed-digit encoding
Pre encoded multipliers based on non-redundant radix-4 signed-digit encoding
 
Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...
Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...
Energy optimized subthreshold vlsi logic family with unbalanced pull up down ...
 
Variable form carrier-based pwm for boost-voltage motor driver with a charge-...
Variable form carrier-based pwm for boost-voltage motor driver with a charge-...Variable form carrier-based pwm for boost-voltage motor driver with a charge-...
Variable form carrier-based pwm for boost-voltage motor driver with a charge-...
 
Ultrasparse ac link converters
Ultrasparse ac link convertersUltrasparse ac link converters
Ultrasparse ac link converters
 
Single inductor dual-output buck–boost power factor correction converter
Single inductor dual-output buck–boost power factor correction converterSingle inductor dual-output buck–boost power factor correction converter
Single inductor dual-output buck–boost power factor correction converter
 
Ripple minimization through harmonic elimination in asymmetric interleaved mu...
Ripple minimization through harmonic elimination in asymmetric interleaved mu...Ripple minimization through harmonic elimination in asymmetric interleaved mu...
Ripple minimization through harmonic elimination in asymmetric interleaved mu...
 
Resonance analysis and soft switching design of isolated boost converter with...
Resonance analysis and soft switching design of isolated boost converter with...Resonance analysis and soft switching design of isolated boost converter with...
Resonance analysis and soft switching design of isolated boost converter with...
 
Reliability evaluation of conventional and interleaved dc–dc boost converters
Reliability evaluation of conventional and interleaved dc–dc boost convertersReliability evaluation of conventional and interleaved dc–dc boost converters
Reliability evaluation of conventional and interleaved dc–dc boost converters
 
Power factor corrected zeta converter based improved power quality switched m...
Power factor corrected zeta converter based improved power quality switched m...Power factor corrected zeta converter based improved power quality switched m...
Power factor corrected zeta converter based improved power quality switched m...
 
Pfc cuk converter fed bldc motor drive
Pfc cuk converter fed bldc motor drivePfc cuk converter fed bldc motor drive
Pfc cuk converter fed bldc motor drive
 
Optimized operation of current fed dual active bridge dc dc converter for pv ...
Optimized operation of current fed dual active bridge dc dc converter for pv ...Optimized operation of current fed dual active bridge dc dc converter for pv ...
Optimized operation of current fed dual active bridge dc dc converter for pv ...
 
Online variable topology type photovoltaic grid-connected inverter
Online variable topology type photovoltaic grid-connected inverterOnline variable topology type photovoltaic grid-connected inverter
Online variable topology type photovoltaic grid-connected inverter
 

Recently uploaded

TOP 10 B TECH COLLEGES IN JAIPUR 2024.pptx
TOP 10 B TECH COLLEGES IN JAIPUR 2024.pptxTOP 10 B TECH COLLEGES IN JAIPUR 2024.pptx
TOP 10 B TECH COLLEGES IN JAIPUR 2024.pptx
nikitacareer3
 
Unbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptxUnbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptx
ChristineTorrepenida1
 
[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf
[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf
[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf
awadeshbabu
 
Tutorial for 16S rRNA Gene Analysis with QIIME2.pdf
Tutorial for 16S rRNA Gene Analysis with QIIME2.pdfTutorial for 16S rRNA Gene Analysis with QIIME2.pdf
Tutorial for 16S rRNA Gene Analysis with QIIME2.pdf
aqil azizi
 
Modelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdfModelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdf
camseq
 
spirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptxspirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptx
Madan Karki
 
Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024
Massimo Talia
 
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressionsKuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
Victor Morales
 
Technical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prismsTechnical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prisms
heavyhaig
 
Swimming pool mechanical components design.pptx
Swimming pool  mechanical components design.pptxSwimming pool  mechanical components design.pptx
Swimming pool mechanical components design.pptx
yokeleetan1
 
14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application
SyedAbiiAzazi1
 
一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理
一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理
一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理
dxobcob
 
Fundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptxFundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptx
manasideore6
 
6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)
ClaraZara1
 
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理
zwunae
 
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
obonagu
 
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdfGoverning Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
WENKENLI1
 
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
Mukeshwaran Balu
 
ACEP Magazine edition 4th launched on 05.06.2024
ACEP Magazine edition 4th launched on 05.06.2024ACEP Magazine edition 4th launched on 05.06.2024
ACEP Magazine edition 4th launched on 05.06.2024
Rahul
 
Literature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptxLiterature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptx
Dr Ramhari Poudyal
 

Recently uploaded (20)

TOP 10 B TECH COLLEGES IN JAIPUR 2024.pptx
TOP 10 B TECH COLLEGES IN JAIPUR 2024.pptxTOP 10 B TECH COLLEGES IN JAIPUR 2024.pptx
TOP 10 B TECH COLLEGES IN JAIPUR 2024.pptx
 
Unbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptxUnbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptx
 
[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf
[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf
[JPP-1] - (JEE 3.0) - Kinematics 1D - 14th May..pdf
 
Tutorial for 16S rRNA Gene Analysis with QIIME2.pdf
Tutorial for 16S rRNA Gene Analysis with QIIME2.pdfTutorial for 16S rRNA Gene Analysis with QIIME2.pdf
Tutorial for 16S rRNA Gene Analysis with QIIME2.pdf
 
Modelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdfModelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdf
 
spirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptxspirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptx
 
Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024Nuclear Power Economics and Structuring 2024
Nuclear Power Economics and Structuring 2024
 
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressionsKuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
 
Technical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prismsTechnical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prisms
 
Swimming pool mechanical components design.pptx
Swimming pool  mechanical components design.pptxSwimming pool  mechanical components design.pptx
Swimming pool mechanical components design.pptx
 
14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application
 
一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理
一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理
一比一原版(Otago毕业证)奥塔哥大学毕业证成绩单如何办理
 
Fundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptxFundamentals of Electric Drives and its applications.pptx
Fundamentals of Electric Drives and its applications.pptx
 
6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)6th International Conference on Machine Learning & Applications (CMLA 2024)
6th International Conference on Machine Learning & Applications (CMLA 2024)
 
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单专业办理
 
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
 
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdfGoverning Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
 
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
 
ACEP Magazine edition 4th launched on 05.06.2024
ACEP Magazine edition 4th launched on 05.06.2024ACEP Magazine edition 4th launched on 05.06.2024
ACEP Magazine edition 4th launched on 05.06.2024
 
Literature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptxLiterature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptx
 

STATE-CLUSTERING BASED MULTIPLE DEEP NEURAL NETWORKS MODELING APPROACH FOR SPEECH RECOGNITION

  • 1. STATE-CLUSTERING BASED MULTIPLE DEEP NEURAL NETWORKS MODELING APPROACH FOR SPEECH RECOGNITION ABSTRACT The hybrid deep neural network (DNN) and hidden Markov model (HMM) has recently achieved dramatic performance gains in automatic speech recognition (ASR). The DNN-based acoustic model is very powerful but its learning process is extremely time-consuming. In this paper, we propose a novel DNN-based acoustic modeling framework for speech recognition, where the posterior probabilities of HMM states are computed from multiple DNNs (mDNN), instead of a single large DNN, for the purpose of parallel training towards faster turnaround. In the proposed mDNN method all tied HMM states are first grouped into several disjoint clusters based on data-driven methods. Next, several hierarchically structured DNNs are trained separately in parallel for these clusters using multiple computing units (e.g. GPUs). In decoding, the posterior probabilities of HMM states can be calculated by combining outputs from multiple DNNs. In this work, we have shown that the training procedure of the mDNN under popular criteria, including both frame-level cross-entropy and sequence-level discriminative training, can be parallelized efficiently to yield significant speedup. The training speedup is mainly attributed to the fact that multiple DNNs are parallelized over multiple GPUs and each DNN is smaller in size and trained by only a subset of training data. We have evaluated the proposed mDNN method on a 64-hour Mandarin transcription task and the 320-hour Switchboard task. Compared to the conventional DNN, a 4-cluster mDNN model with similar size can yield comparable recognition performance in Switchboard (only about 2% performance degradation) with a greater than 7 times speed improvement in CE training and a 2.9 times improvement in sequence training, when 4 GPUs are used.