Yong-Wan Roh

Speech Emotion Recognition Using Spectral Entropy

This paper proposes a Gaussian Mixture Model (GMM)---based speech emotion recognition methods using four feature parameters; 1) Fast Fourier Transform(FFT) spectral entropy, 2) delta FFT spectral entropy, 3) Mel-frequency Filter Bank (MFB) spectral entropy, 4) delta MFB spectral entropy. In addition, we use four emotions in a speech database including anger, sadness, happiness, and neutrality. We perform speech emotion recognition experiments using each pre-defined emotion and gender. The experimental results show that the proposed emotion recognition using FFT spectral-based entropy and MFB spectral-based entropy performs better than existing emotion recognition based on GMM using energy, Zero Crossing Rate (ZCR), Linear Prediction Coefficient (LPC), and pitch parameters.

A method of multiple odors detection and recognition

In this paper, we propose a method to detect and recognize multiple odors, and implement a multiple odor recognition system. Multiple odor recognition technology has not yet been developed, since existing odor recognition techniques which have been researched and developed by components analysis and pattern recognition techniques only deal with single odors at a time. Multiple odors represent a dynamic odor change from no odor to a single odor and multiple odors, which is the most common situation in a real-world environment. Therefore, it is necessary to sense and recognize techniques for dynamic odor changes. To recognize multiple odors, the proposed method must include odor data acquisition using a smell sensor array, odor detection using entropy, feature extraction using Principal Component Analysis, recognition candidate selection using Tree Search, and recognition using Euclidean Distance. To verify the validity of this study, a performance evaluation was conducted using a 132 odor database. As a result, the odor detection rate is approximately 95.83% and the odor recognition rate is approximately 88.97%.

An implementation of real time-sentential KSSL recognition system based on the post wearable PC

Korean Standard Sign Language (hereinafter, “KSSL”) is a complex visual-spatial language that is used by the deaf community in the South Korea. Wire communications net and desktop PC-based a traditional study on sign language linguistics with small vocabulary (words) have several restrictions (e.g. limitation of the motion, conditionality in the space) and general problems (e.g. inaccuracy in measuring, necessity of complex computation algorithm) according to using of vision technologies with image capture and video processing system as input module of sign language signals. Consequently, in this paper we propose and implement ubiquitous-oriented wearable PC-based sentential KSSL recognizer that improve efficiency of KSSL input module according to wireless sensor network, recognizes and represents continuous KSSL with flexibility in real time, and analyze and notify definite intention of user more efficiently through correct measurement of KSSL gestures using wireless haptic devices. The experimental result shows an average recognition rate of 93.7% for continuous 44 KSSL sentences.

Performance evaluation of a hand gesture recognition system using fuzzy algorithm and neural network for post PC platform

In this paper, we implement hand gesture recognition system using fuzzy algorithm and neural network for Post PC (the embedded-ubiquitous environment using blue-tooth module, embedded i.MX21 board and smart gate-notebook computer). Also, we propose most efficient and reasonable hand gesture recognition interface for Post PC through evaluation and analysis of performance about each gesture recognition system. The proposed gesture recognition system consists of three modules: 1) gesture input module that processes motion of dynamic hand to input data, 2) Relational Database Management System (hereafter, RDBMS) module to segment significant gestures from input data and 3) 2 each different recognition module: fuzzy max-min and neural network function recognition module to recognize significant gesture of continuous / dynamic gestures. Experimental result shows the average recognition rate of 98.8% in fuzzy max-min module and 96.7% in neural network recognition module about significantly dynamic gestures.

A Study on VoiceXML Application of User-Controlled Form Dialog System

VoiceXML is new markup language which is designed for web resource navigation via voice based on XML. An application using VoiceXML is classified into mutual-controlled and machine-controlled form dialog structure. Such dialog structures can`t construct service which provide free navigation of web resource by user because a scenario is decided by application developer. In this paper, we propose VoiceXML application structure using user-controlled form dialog system which decide service scenario according to user`s intention. The proposed application automatically detects recognition candidates from requested information by user, and then system uses recognition candidate as voice-anchor. Also, system connects each voice-anchor with new voice-node. An example of proposed system, we implement news service with IT term dictionary, and we confirm detection and registration of voice-anchor and make an estimate of hit rate about measurement of an successive offer from information according to user`s intention and response speed. As the experiment result, we confirmed possibility which is more freely navigation of web resource than existing VoiceXML form dialog systems.

An Implementation of Dynamic Gesture Recognizer Based on WPS and Data Glove

WPS(Wearable Personal Station) for next generation PC can define as a core terminal of `Ubiquitous Computing` that include information processing and network function and overcome spatial limitation in acquisition of new information. As a way to acquire significant dynamic gesture data of user from haptic devices, traditional gesture recognizer based on desktop-PC using wire communication module has several restrictions such as conditionality on space, complexity between transmission mediums(cable elements), limitation of motion and incommodiousness on use. Accordingly, in this paper, in order to overcome these problems, we implement hand gesture recognition system using fuzzy algorithm and neural network for Post PC(the embedded-ubiquitous environment using blue-tooth module and WPS). Also, we propose most efficient and reasonable hand gesture recognition interface for Post PC through evaluation and analysis of performance about each gesture recognition system. The proposed gesture recognition system consists of three modules: 1) gesture input module that processes motion of dynamic hand to input data 2) Relational Database Management System(hereafter, RDBMS) module to segment significant gestures from input data and 3) 2 each different recognition modulo: fuzzy max-min and neural network recognition module to recognize significant gesture of continuous / dynamic gestures. Experimental result shows the average recognition rate of 98.8% in fuzzy min-nin module and 96.7% in neural network recognition module about significantly dynamic gestures.

A hybrid warping method approach to speaker warping adaptation

The method of speaker normalization has been known as the successful method for improving the speech recognition at speaker independent speech recognition system. This paper propose a new power spectrum warping approach to making improvement of speaker normalization better than a frequency warping. The power spectrum warping uses Mel-frequency cepstral of Mel filter bank in MFCC. Also, this paper proposes the hybrid VTN combined the power spectrum warping and a frequency warping. Experiment of this paper did a comparative analysis about the recognition performance of the SKKU PBW DB applied each the power spectrum is 3.06%, and hybrid VTN is 4.07% word error rate reduction as word recognition performance of baseline system.

A Research on the Stochastic Model for Spoken Language Understanding

In this paper, we propose a new stochastic model for sentence speech understanding using dictionary and thesaurus. The proposed model searches the dictionary for the same word with input text. If it is not in the dictionary, the proposed model search the high level words in the high level word dictionary based on the thesaurus. We compare the probability of sentence understanding model with threshold probability, and we’ll get the sentence understanding rate. We evaluated the performance of the sentence speech understanding system by applying twenty questions game. As the experiment results, we got sentence speech understanding accuracy of 79.8%. In this case probability (α) of high level word is 0.9 and threshold probability (β) is 0.38.

Language Model based on VCCV and Test of Smoothing Techniques for Sentence Speech Recognition

In this paper, we propose VCCV units as a processing unit of language model and compare them with clauses and morphemes of existing processing units. Clauses and morphemes have many vocabulary and high perplexity. But VCCV units have low perplexity because of the small lexicon and the limited vocabulary. The construction of language models needs an issue of the smoothing. The smoothing technique used to better estimate probabilities when there is an insufficient data to estimate probabilities accurately. This paper made a language model of morphemes, clauses and VCCV units and calculated their perplexity. The perplexity of VCCV units is lower than morphemes and clauses units. We constructed the N-grams of VCCV units with low perplexity and tested the language model using Katz, absolute, modified Kneser-Ney smoothing and so on. In the experiment results, the modified Kneser-Ney smoothing is tested proper smoothing technique for VCCV units.

A study on the Stochastic Model for Sentence Speech Understanding

In this paper, we propose a stochastic model for sentence speech understanding using dictionary and thesaurus. The proposed model extracts words from an input speech or text into a sentence. A computer is sellected category of dictionary database compared the word extracting from the input sentence calculating a probability value to the compare results from stochastic model. At this time, computer read out upper dictionary information from the upper dictionary searching and extracting word compared input sentence caluclating value to the compare results from stochastic model. We compare adding the first and second probability value from the dictionary searching and the upper dictionary searching with threshold probability that we measure the sentence understanding rate. We evaluated the performance of the sentence speech understanding system by applying twenty questions game. As the experiment results, we got sentence speech understanding accuracy of 79.8%. In this case, probability () of high level word is 0.9 and threshold probability () is 0.38.