Speech and Computer

Name: Speech and Computer | 21st International Conference, SPECOM 2019, Istanbul, Turkey, August 20-25, 2019, Proceedings
Brand: Springer
Price: 53.49 EUR
Availability: OnlineOnly

21st International Conference, SPECOM 2019, Istanbul, Turkey, August 20-25, 2019, Proceedings

Albert Ali Salah Alexey Karpov Rodmonga Potapova(Editor)

Springer (Publisher)

Published on 9. August 2019

XIII, 580 pages

E-Book

PDF with digital watermarking

System requirements

978-3-030-26061-3 (ISBN)

€53.49incl. 7% vat

System requirements

for PDF with digital watermarking

E-Book Single Licence

Available for download

Description

Alles über E-Books | Antworten auf Fragen rund um E-Books, Kopierschutz und Dateiformate finden Sie in unserem Info- & Hilfebereich.

Alles über E-Books, Kopierschutz & Dateiformate finden Sie in unserem Info- & Hilfebereich.

This book constitutes the proceedings of the 21st International Conference on Speech and Computer, SPECOM 2019, held in Istanbul, Turkey, in August 2019.
The 57 papers presented were carefully reviewed and selected from 86 submissions. The papers present current research in the area of computer speech processing including audio signal processing, automatic speech recognition, speaker recognition, computational paralinguistics, speech synthesis, sign language and multimodal processing, and speech and language resources.

More details

Other editions

Content

Intro
SPECOM 2019 Preface
Organization
Contents
The Representation of Speech and Its Processing in the Human Brain and Deep Neural Networks
Abstract
1 Introduction
2 Speech Representations
3 Adaptation to Non-standard Speech
3.1 Does a DNN Show Human-Like Adaptation to Nonstandard Speech?
3.2 Are Nonstandard Sounds Processed Similarly in Human Listeners and DNNs?
4 Concluding Remarks
Acknowledgments
References
A Detailed Analysis and Improvement of Feature-Based Named Entity Recognition for Turkish
1 Introduction
2 Previous Work
3 Datasets
4 Methodology
4.1 Model
4.2 Features
4.3 Evaluation Metrics
5 Results and Discussion
6 Conclusion and Future Work
References
A Comparative Study of Classical and Deep Classifiers for Textual Addressee Detection in Human-Human-Machine Conversations
1 Introduction
2 Related Work
3 Classifiers
3.1 Classical Models
3.2 Deep Models
3.3 Fusion
4 Corpora
5 Experimental Results
6 Conclusions
References
Acoustic Event Mixing to Multichannel AMI Data for Distant Speech Recognition and Acoustic Event Classification Benchmarking
1 Introduction
2 Preliminary Information
2.1 Revision of Available Corpora
2.2 Problem Formulation
3 Proposed Approach to Acoustic Event Mixing
3.1 Acoustic Event Spatial Reverberation
3.2 Acoustic Event Mixing
3.3 Parameter Specification for the AMI Corpus
4 Evaluation and Results
4.1 Evaluation Methods and Metrics
4.2 Evaluation Results
5 Conclusion
References
Speech-Based L2 Call System for English Foreign Speakers
1 Introduction
2 Dataset Description
3 Features Prediction
3.1 Features Produced by Universal Sentence Encoder
3.2 Features Using Python English Grammar Checker Toolkit
3.3 Part-of-Speech (POS) Features
3.4 Response Embedding to Binary Features
4 Evaluation
4.1 Evaluation Metric
4.2 Experiments and Results
4.3 Fusion of Multiple Models
4.4 Comparison and Discussion
5 Conclusion and Future Work
References
A Pattern Mining Approach in Feature Extraction for Emotion Recognition from Speech
1 Introduction
2 Our Approach
2.1 Dimensionality Reduction
2.2 Discretization
2.3 Pattern Mining
2.4 Feature Extraction
2.5 Classification
3 Dataset and Experimental Results
3.1 Dataset
3.2 Experimental Results
4 Conclusion
References
Towards a Dialect Classification in German Speech Samples
1 Introduction
2 Methods
2.1 Speech Data
2.2 Models
3 Baseline Experiments
4 Results and Discussion
5 Conclusions
References
Classification of Regional Accent Using Speech Rhythm Metrics
Abstract
1 Introduction
2 MSA Language
3 Rhythm Modeling
4 Participants
5 Experiments and Results
5.1 Rhythm Metrics Dataset
5.2 Classifier Design
5.3 Experiments
6 Conclusion
References
PocketEAR: An Assistive Sound Classification System for Hearing-Impaired
1 Introduction
1.1 Motivation
2 System Architecture
2.1 Client Application
2.2 Servers
2.3 Operator's Console
3 Environmental Sound Classification
3.1 Classifier Architecture
3.2 Classifier Training
4 Results
References
Time-Continuous Emotion Recognition Using Spectrogram Based CNN-RNN Modelling
1 Introduction
2 Related Work
3 Data and Preprocessing
3.1 Database
3.2 Labels
3.3 The Spectrogram
3.4 The Waveform
4 Experiments and Results
4.1 The Spectrogram
4.2 The Waveform
4.3 Comparison of Segmenting Window Length
4.4 Comparison of the Proposed Approach to the Existing One
5 Conclusion
References
Developmental Disorders Manifestation in the Characteristics of the Child's Voice and Speech: Perceptual and Acoustic Study
Abstract
1 Introduction
2 Methods
3 Results
3.1 Experiment 1. Recognition of the Child's State, Age, Gender by Pediatric Students
3.2 Experiment 2. Recognition of the Child's State by Psychiatric Students and Psychiatrists
3.3 Acoustic Data
3.4 Experiment 3. Description of Child's Speech Material by Specialists (Researchers)
4 Discussion
5 Conclusions
Acknowledgements
References
RUSLAN: Russian Spoken Language Corpus for Speech Synthesis
1 Introduction
2 Speech Corpus
2.1 Text Preprocessing
2.2 Recording Process
3 Neural Network for Speech Synthesis
3.1 Neural Network Architecture
3.2 Training
4 Evaluation
4.1 Mean Opinion Score
5 Conclusion
References
Differentiating Laughter Types via HMM/DNN and Probabilistic Sampling
1 Introduction
2 The Recordings Used
3 DNN Training by Probabilistic Sampling
4 Classification Experiments
4.1 DNN Parameters
4.2 Probabilistic Sampling
4.3 Evaluation
4.4 Results
5 Experiments with a Hidden Markov Model
5.1 Evaluation Metrics
5.2 Results
6 Conclusions
References
Word Discovering in Low-Resources Languages Through Cross-Lingual Phonemes
1 Introduction
2 System Overview
3 Autoencoders
4 Corpus and Evaluation Measures
5 Tuning Process
6 Experiments
7 Conclusions
References
Semantic Segmentation of Historical Documents via Fully-Convolutional Neural Network
1 Introduction
2 Data
3 Method
4 Experiments and Results
4.1 Two-Class Classification
4.2 Single-Character Classification
5 Conclusion and Future Work
References
A New Approach of Adaptive Filtering Updating for Acoustic Echo Cancellation
Abstract
1 Introduction
2 Adaptive Filtering Based Acoustic Echo Cancellation
3 Proposed Approach of Adaptive Filtering Updating
4 Simulation Results and Discussions
5 Conclusion
References
Code-Switching Language Modeling with Bilingual Word Embeddings: A Case Study for Egyptian Arabic-English
1 Introduction
2 Related Work
3 Data
4 Bilingual Word Embeddings
5 Language Modeling
6 Intrinsic Evaluations
7 Discussion
8 Conclusion
References
Identity Extraction from Clusters of Multi-modal Observations
1 Introduction
2 Definitions and Methods
2.1 Identity
2.2 Persona
2.3 Multi-modal Observation
2.4 Time Correlation Functions
2.5 Multi-modal Distances
3 Identity Extraction Using Agglomerative Clustering
3.1 Hard Conditions
3.2 Leftover Clusters
4 Experiments
5 Conclusion
References
Don't Talk to Noisy Drones - Acoustic Interaction with Unmanned Aerial Vehicles
1 Introduction
2 UAV-Based Communication Scenarios and Challenges
3 Sound and Communication Experiments
3.1 Acoustic Characteristics of the Sample UAV
3.2 Speech Command and POLQA Tests
3.3 Environmental Noise Measurement Setup
4 Results and Discussion
4.1 Generic Acoustic Characteristics
4.2 Speech Command Analyses
4.3 Environmental Noise Measurement Analyses
5 Lessons Learned for the UAV Research in Progress
6 Conclusions
References
Method for Multimodal Recognition of One-Handed Sign Language Gestures Through 3D Convolution and LSTM Neural Networks
Abstract
1 Introduction
2 Related Work
3 Database
4 Method
5 Experiment
6 Conclusion
Acknowledgments
References
LSTM-Based Kazakh Speech Synthesis
1 Introduction
2 Corpus and Language
2.1 Language
2.2 Corpus
3 Prosodic Processing
3.1 Pause Place Prediction
3.2 Phoneme Duration Prediction
4 Acoustic Model
4.1 Acoustic Model Based on LSTM
4.2 Training Acoustic Model
5 TTS
6 Evaluation
7 Discussion
8 Conclusion
References
Combination of Positions and Angles for Hand Pose Estimation
1 Introduction
2 Methods
3 Experiments
4 Results and Discussion
5 Conclusion
References
LSTM-Based Language Models for Very Large Vocabulary Continuous Russian Speech Recognition System
Abstract
1 Introduction
2 Related Works
3 LSTM Language Models for Russian
4 Experiments
4.1 Experimental Setup
5 Conclusions and Future Work
Acknowledgements
References
Svarabhakti Vowel Occurrence and Duration in Rhotic Clusters in French Lyric Singing
Abstract
1 Introduction
1.1 Rhotics in French Lyric Singing
1.2 Consonant Clusters vs. Syllable Structure
1.3 Svarabhakti Vowel in Singing
2 Material and Method
2.1 Performers and Repertoire
2.2 Acoustic and Statistic Analysis
3 Results
3.1 CR vs. RC Clusters
3.2 Consonantal Environment Impact
3.3 Svarabhakti Occurrence vs. Duration vs. Musical Tempo
4 Discussion
5 Conclusion
References
The Evaluation Process Automation of Phrase and Word Intelligibility Using Speech Recognition Systems
Abstract
1 Introduction
2 Proposed Approach
2.1 Applied Assessment Method
2.2 Description of Speech Recognition Systems
2.3 Description of the Database and Experiment Methodology
3 Results
4 Conclusion
Acknowledgments
References
Detection of Overlapping Speech for the Purposes of Speaker Diarization
1 Introduction
1.1 Problems with Data
2 Overlap Detector
3 Data
3.1 Synthetic Training Data
3.2 Test Data
4 Evaluation
4.1 Results
5 Conclusion
References
Exploring Hybrid CTC/Attention End-to-End Speech Recognition with Gaussian Processes
1 Introduction
2 Background
2.1 Location-Aware Attention-Based Encoder and Decoder
2.2 Frame-Discriminative CTC Network
2.3 Hybrid CTC/Attention Multi-objective Training
2.4 Joint One-Pass Beam Search Decoding
2.5 Gaussian Processes Optimization
3 Experiment Setup
4 Results and Evaluation
4.1 Observations on Certain Parameter Groups
4.2 Feedback Loops Caused by Unexpected Letter Hypotheses
4.3 CTC as Aligning Regularizer
5 Conclusion
References
Estimating Aggressiveness of Russian Texts by Means of Machine Learning
Abstract
1 Introduction
2 Related Work
2.1 Methods and Systems
2.2 Datasets
3 Processing Scheme
4 Experiments
5 Conclusions
Acknowledgment
References
Software Subsystem Analysis of Prosodic Signs of Emotional Intonation
1 Introduction
2 Visual Representation of Emotional Intonation Features
3 Numerical Evaluation of Signs of Emotional Intonation
4 Preliminary Testing of the Developed Signs of Emotional Intonation
5 Conclusions
References
Assessing Alzheimer's Disease from Speech Using the i-vector Approach
1 Introduction
2 Data
3 Methods
3.1 Feature Extraction
3.2 The i-vector Approach
4 Experiments and Results
4.1 i-vectors Extraction
4.2 Evaluation
5 Conclusions and Future Work
References
AD-Child.Ru: Speech Corpus for Russian Children with Atypical Development
Abstract
1 Introduction
2 Speech Database for Children with Atypical Development
2.1 Data Collection
2.2 Database Structure
3 Data Analysis
4 Experimental Results
4.1 Experiment 1. Word's Meaning and State (Normally Development or Disorder) Recognition by Listeners via Speech of TD Children and Children with ASD and DS
4.2 Experiment 2. Word's Meaning, Age, and Gender of TD and ASD Children Recognition by Listeners
5 Discussion
6 Conclusions
Acknowledgements
References
Building a Pronunciation Dictionary for the Kabyle Language
Abstract
1 Introduction
2 Methodology
2.1 Text Corpus
2.2 Phone Set
2.3 Algorithm
3 Results
4 Conclusion
References
Speech-Based Automatic Assessment of Question Making Skill in L2 Language
1 Introduction
2 Related Works
3 System Description
3.1 Speech Recognition System
3.2 WH-Question Word Rule-Based Component
3.3 Python English Grammar Checker
3.4 Machine Learning Based Grammar/Language Checker
4 Data Collection and Description
5 Experiments and Results
6 Conclusion
References
Automatic Recognition of Speaker Age and Gender Based on Deep Neural Networks
Abstract
1 Introduction
2 Related Work
3 Speech Corpus
4 Proposed System for Age and Gender Recognition
5 Experimental Results
6 Conclusions and Future Work
Acknowledgements
References
Investigating Joint CTC-Attention Models for End-to-End Russian Speech Recognition
1 Introduction
2 Main Related Works
3 Model Architecture
3.1 Attention-Based Encoder-Decoder Model
3.2 Proposed Recognition Model
3.3 Beam Search Prunning Method
3.4 Using of Sparsemax Function
3.5 Highway BLSTM Encoder
4 Experimental Setup
4.1 Training Dataset
4.2 Data Augmentation
5 Results of the Proposed Model
6 Conclusion
References
Author Clustering with and Without Topical Features
Abstract
1 Introduction
2 Related Work
2.1 Unmasking for Cross-Domain Authorship Attribution
2.2 Author Clustering
2.3 Unmasking for Author Clustering in Russian
3 Experiment
3.1 Dataset Description
3.2 Methods
4 Results and Discussion
4.1 Unmasking Topical Features
4.2 Author Clustering Performance
5 Conclusions and Future Work
Acknowledgment
References
Assessment of Syllable Intelligibility Based on Convolutional Neural Networks for Speech Rehabilitation After Speech Organs Surgical Interventions
Abstract
1 Introduction
2 Description of the Proposed Approach
3 Data Preprocessing
3.1 Analysis of Phonetic Composition of the Audio Data Given, Splitting Recordings into Frames, and Labeling
3.2 Calculating MFCCs of Phoneme Audio Files
4 Construction and Training a Neural Network
4.1 Building CNN
4.2 Training the CNN
4.3 Phoneme Recognition in the Syllables
5 Evaluation of the Phoneme Recognition by the CNN
6 Working with Signals After Surgery
7 Conclusion
Acknowledgments
References
Corpus Study of Early Bulgarian Onomatopoeias in the Terms of CHILDES
1 Introduction
2 The Data
3 Onomatopeias in the Early Speech of Two Bulgarian Children
4 Conclusion
References
EEG Investigation of Brain Bioelectrical Activity (Regarding Perception of Multimodal Polycode Internet Discourse)
Abstract
1 Introduction
2 Method, Results
3 Conclusion
Acknowledgments
References
Some Peculiarities of Internet Multimodal Polycode Corpora Annotation
Abstract
1 Introduction
2 Background and Method
3 Results and Discussion
4 Conclusion
Acknowledgments
References
New Perspectives on Canadian English Digital Identity Based on Word Stress Patterns in Lexicon and Spoken Corpus
Abstract
1 Introduction
2 Methods
3 Results
3.1 Overall Quantitative and Structural Analysis of the Lexicon
3.2 Spoken Corpus and the Dictionary Data Compared
3.3 Regional Identity in Ontario and Quebec
4 Discussion and Conclusions
References
Automatic Speech Recognition for Kreol Morisien: A Case Study for the Health Domain
Abstract
1 Introduction
2 Literature Review
2.1 Kreol Morisien
2.2 Automatic Speech Recognition for Under-Resourced Languages
3 Implementation of Acoustic Model
3.1 Data Collection
3.2 Building of Phonetic Dictionary
3.3 Building of Language Model
3.4 Preparation of Transcript Files
3.5 Training the Acoustic Model
4 User Evaluation
4.1 User Study 1
4.2 User Study 2
5 Conclusion and Future Work
References
Script Selection Using Convolutional Auto-encoder for TTS Speech Corpus
1 Introduction
2 Methodology
2.1 Information Extraction
2.2 Embedding Space
2.3 Utterance Selection
3 Experiments and Results
3.1 Experimental Setup
3.2 Best Configuration Selection
3.3 Coverage Rate and KLD Evaluation
3.4 Subjective Evaluation
4 Conclusion
References
Pragmatic Markers Distribution in Russian Everyday Speech: Frequency Lists and Other Statistics for Discourse Modeling
Abstract
1 Introduction
2 Research Data
2.1 Dialogue Everyday Speech
2.2 Monologue Speech
2.3 Data Annotation
3 Frequency Lists of Pragmatic Markers in Monologue and Dialogue Speech
4 The Functions of Pragmatic Markers in Monologue and Dialogue Speech
5 Top Frequency Lists of Russian PMs for Speakers of Different Gender and Age Groups
6 Conclusion
Acknowledgements
References
Curriculum Learning in Sentiment Analysis
1 Introduction
2 Related Work
3 Experiment Setup
3.1 Curriculum Epochs Design
3.2 Architecture
3.3 The Data Set
4 Experiments Results
5 Future Work
6 Conclusion
References
First Minute Timing in American Telephone Talks: A Cognitive Approach
Abstract
1 Introduction
2 Methodology
2.1 Data
2.2 Measurements
3 Results
4 Discussion and Conclusions
References
Syntactic Segmentation of Spontaneous Speech: Psychological and Cognitive Aspects
1 Introduction
2 Data and Experimental Design
2.1 Participants
2.2 Experimental Stimuli
2.3 Personality Inventory
2.4 Measuring Working Memory Capacity
2.5 Processing Speed Tasks
2.6 Dichotic Listening
3 Data Analysis
3.1 Descriptive Analysis of the Data
3.2 Inter-annotator Agreement
3.3 Mixed Linear Regression Modelling
4 Discussion and Conclusions
References
Dual-Microphone Speech Enhancement System Attenuating both Coherent and Diffuse Background Noise
1 Introduction
2 Dual Microphone System
3 The Separation of Sounds Using Dual-Microphone Array
4 Speech Enhancement Using Adaptive MVDR
5 Experiments and Results
5.1 Experiments with Artificial Mixture
5.2 Experiment in Anechoic Chamber
5.3 Experiments in Reverberant Room
6 Conclusions
References
Reducing the Inter-speaker Variance of CNN Acoustic Models Using Unsupervised Adversarial Multi-task Training
1 Introduction
2 Multi-task and Adversarial Multi-task Training
3 Experimental Set-Up
4 Results with Supervised Adversarial Training
5 Unsupervised Training Without Speaker Labels
5.1 Conventional Speaker Clustering
5.2 Clustering Using a Siamese Multi-task Network
6 Summary
References
Estimates of Transmission Characteristics Related to Perception of Bone-Conducted Speech Using Real Utterances and Transcutaneous Vibration on Larynx
1 Introduction
2 Analysis of the Long-Term Average Spectrum
2.1 Speakers
2.2 Apparatus and Procedure
2.3 Analysis
2.4 Results and Discussions
3 Transfer Function Measurement for BC Speech Transmission
3.1 Participants
3.2 Apparatus
3.3 Procedure
3.4 Analysis
3.5 Results and Discussion
4 General Discussions
5 Conclusion
References
Singing Voice Database
Abstract
1 Introduction
2 Previous Work
3 Singing Voice Database Content
4 Singing Voice Database Recording
5 Singing Voice Database Processing
6 Conclusions
References
How Dysarthric Prosody Impacts Naïve Listeners' Recognition
1 Introduction
2 Methods
2.1 Data Collection
2.2 Participants for Perceptual Experiment
2.3 Stimuli
2.4 Procedure
2.5 Analysis
3 Results
4 Discussion
References
Light CNN Architecture Enhancement for Different Types Spoofing Attack Detection
1 Introduction
2 Deep Learning Approach
2.1 Light CNN Classifier
2.2 LCNN System Modifications
2.3 Angular Margin Based Softmax Activation
2.4 Front-End
3 Experimental Setup
3.1 Datasets
3.2 Experiments
4 Results and Discussion
5 Conclusion
References
Deep Neural Network Quantizers Outperforming Continuous Speech Recognition Systems
1 Introduction
2 Proposed Method
2.1 Deep Neural Network Quantizer Training
2.2 Discrete Hidden Markov Model Training
3 Experimental Setup
4 Results
4.1 Number of Clusters
4.2 Number of Spliced Frames
5 Conclusion
References
Speaking Style Based Apparent Personality Recognition
1 Introduction and Related Works
1.1 The Big-Five Model
1.2 Audio Based AAPR
1.3 Neural Style Transfer
2 System Description
3 Experiments and Results
3.1 Dataset
3.2 Low Level Feature Extraction
3.3 Overall Settings
3.4 Our Baseline
3.5 Results and Discussion
4 Conclusion and Future Work
References
Diarization of the Language Consulting Center Telephone Calls
1 Introduction
2 Archive Data Description
3 Speaker Diarization
3.1 Segmentation
3.2 Segment Description
3.3 Clustering
3.4 Resegmentation
4 Speaker Identification
4.1 Training the Speakers GMMs
5 Modified Resegmentation Step
6 Kaldi Diarization System
7 Experiments
7.1 Training Data
7.2 Results
8 Discussion
9 Conclusion
References
NN-Based Czech Sign Language Synthesis
1 Introduction
2 Related Work
3 Skeleton Model Restoration
4 Sign Language Synthesis
4.1 Implicit Sign Language Translation
4.2 Sign Language Production
5 Experiments and Results
6 Conclusion
References
Re-evaluation of Words Used in Speech Audiometry
Abstract
1 Introduction
2 Methodology
2.1 Hypotheses
2.2 The Testing Tool
2.3 Subjects
2.4 Procedure
3 Results
4 Discussion
5 Conclusions
Acknowledgements
References
Author Index

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

Speech and Computer

Description

More details

Other editions

Additional editions

Content

System requirements