• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.14 no.8
• /
• pp.3992-3998
• /
• 2013

This paper is related to the method of adding a emotional speech corpus to a high-quality large corpus based speech synthesizer, and generating various synthesized speech. We made the emotional speech corpus as a form which can be used in waveform concatenated speech synthesizer, and have implemented the speech synthesizer that can be generated various synthesized speech through the same synthetic unit selection process of normal speech synthesizer. We used a markup language for emotional input text. Emotional speech is generated when the input text is matched as much as the length of intonation phrase in emotional speech corpus, but in the other case normal speech is generated. The BIs(Break Index) of emotional speech is more irregular than normal speech. Therefore, it becomes difficult to use the BIs generated in a synthesizer as it is. In order to solve this problem we applied the Variable Break[3] modeling. We used the Japanese speech synthesizer for experiment. As a result we obtained the natural emotional synthesized speech using the break prediction module for normal speech synthesize.

• MALSORI
• /
• no.48
• /
• pp.93-105
• /
• 2003

Since telephone channel has bandlimited frequency characteristics, speech signal over the telephone channel shows degraded speech quality. In this paper, we propose an algorithm using neural network to reconstruct wideband speech from its narrowband version. Although single neural network is a good tool for direct mapping, it has difficulty in training for vast and complicated data. To alleviate this problem, we modularize the neural networks based on appropriate clustering of the acoustic space. We also introduce fuzzy computing to compensate for probable misclassification at the cluster boundaries. According to our simulation, the proposed algorithm showed improved performance over the single neural network and conventional codebook mapping method in both objective and subjective evaluations.

• Speech Sciences
• /
• v.10 no.2
• /
• pp.7-25
• /
• 2003

Predicting the right prosodic elements is a key factor in improving the quality of synthesized speech. Prosodic elements include break, pitch, duration and loudness. Pitch, which is realized by Fundamental Frequency (F0), is the most important element relating to the quality of the synthesized speech. However, the previous method for predicting the F0 appears to reveal some problems. If voiced and unvoiced sounds are not correctly classified, it results in wrong prediction of pitch, wrong unit of triphone in synthesizing the voiced and unvoiced sounds, and the sound of click or vibration. This kind of feature is usual in the case of the transformation from the voiced sound to the unvoiced sound or from the unvoiced sound to the voiced sound. Such problem is not resolved by the method of grammar, and it much influences the synthesized sound. Therefore, to steadily acquire the correct value of pitch, in this paper we propose a new model for predicting and classifying the voiced and unvoiced sounds using the CART tool.

• Speech Sciences
• /
• v.11 no.2
• /
• pp.227-235
• /
• 2004

In this paper, we proposed a new practical implementation method of the lattice transversal joint (LTJ) adaptive filter using speech codec's information. And it was applied to the adaptive echo cancellation problem to verify the efficiency of the proposed method. Realtime implementation of the LTJ adaptive filter is very difficult due to high computational complexity for the filter coefficients compensation. However, in case of using speech codec, complexity can be reduced since linear predictive coding (LPC) coefficients are updated each frame or sub-frame instead of every sample. Furthermore, LPC coefficients can be acquired from speech decoder and transformed to the reflection coefficients. Therefore, the computational complexity for updates of the reflection coefficients can be reduced. The effectiveness of the proposed LTJ adaptive filter was verified by the experiments about convergence and tracking performance of the adaptive echo canceller.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.10d
• /
• pp.302-305
• /
• 2006

In this paper, we have proposed a novel method using quantization of wavelet coefficients for secret data communication. First, speech signal is partitioned into small time frames and the frames are transformed into frequency domain using a WT(Wavelet Transform). We quantize the wavelet coefficients and embedded secret data into the quantized wavelet coefficients. The destination regard quantization errors of received speech as seceret dat. As most speech watermark techniques have a trade off between noise robustness and speech quality, our method also have. However we solve the problem with a partial quantization and a noise level dependent threshold. In additional, we improve the speech quality with de-noising method using wavelet transform. Since the signal is processed in the wavelet domain, we can easily adapt the de-noising method based on wavelet transform. Simulation results in the various noisy environments show that the proposed method is reliable for secret communication.

• Speech Sciences
• /
• v.8 no.3
• /
• pp.123-132
• /
• 2001

In recent years research on HCI technology has been very active and speech recognition is being used as its typical method. Its recognition, however, is deteriorated with the increase of surrounding noise. To solve this problem, studies concerning the multimodal HCI are being briskly made. This paper describes automated lipreading for bimodal speech recognition on the basis of image- and speech information. It employs audio-visual DB containing 1,074 words from 70 voice and tri-viseme as a recognition unit, and state tied HMM as a recognition model. Performance of automated recognition of 22 to 1,000 words are evaluated to achieve word recognition of 60.5% in terms of 22word recognizer.

• Journal of Information Processing Systems
• /
• v.19 no.5
• /
• pp.641-651
• /
• 2023

Malicious hate speech and gender bias comments are common in online communities, causing social problems in our society. Gender bias and hate speech detection has been investigated. However, it is difficult because there are diverse ways to express them in words. To solve this problem, we attempted to detect malicious comments in a Korean hate speech dataset constructed in 2020. We explored bidirectional encoder representations from transformers (BERT)-based deep learning models utilizing hyperparameter tuning, data sampling, and logits ensembles with a label distribution. We evaluated our model in Kaggle competitions for gender bias, general bias, and hate speech detection. For gender bias detection, an F1-score of 0.7711 was achieved using an ensemble of the Soongsil-BERT and KcELECTRA models. The general bias task included the gender bias task, and the ensemble model achieved the best F1-score of 0.7166.

• Journal of the Korean Society of Laryngology, Phoniatrics and Logopedics
• /
• v.30 no.1
• /
• pp.39-47
• /
• 2019

Background and Objectives : Pediatric vocal health problems are relatively common. However, it is not yet well studied whether uniform diagnosis and treatment is done properly in South Korea. The purpose of this study was to investigate the methods that the Korean speech therapists use to diagnose and treat pediatric voice problem. Materials and Method : An anonymous online questionnaire was administered to 32 speech language therapists registered at the Korean laryngeal speech linguistics society detailing demographics, employment institution, general management of pediatric patients with vocal problem including assessment and treatment procedures. Results : Current practice patterns were analyzed on 32 speech language therapists providing services in South Korea mostly working at tertiary university hospital. One third of pediatric patients were assessed without proceeding to treatment. One fifth of patients were treated without assessment. Perceptual assessment was the main pretreatment assessment methods used. Treatment was done in the following order : Voice rest, SOVT, yawn-sigh and resonant voice. Post-treatment evaluation was used in the following order : Instrumental assessment, clinical judgment, and recording comparison. Conclusion : Speech language therapists practice in South Korea mostly follows the ASHA practice guidelines. However, there are still great amount of cases in which only the evaluation was done without appropriate treatment. Further research is needed to make SPLs more systematic and efficient for evaluating and treating pediatric vocal patients.

• The Journal of the Korea Contents Association
• /
• v.8 no.1
• /
• pp.328-338
• /
• 2008

Most speech recognition systems that have a large capacity and high recognition rates are isolated word speech recognition systems. In order to extend the scope of recognition, it is necessary to increase the number of words that are to be searched. However, it shows a problem that exhibits a decrease in the system performance according to the increase in the number of words. This paper defines the context information that affects speech recognition in a ubiquitous environment to solve such a problem and develops user localization method using inertial sensor and RFID. Also, we develop a new speech recognition system that demonstrates better performances than the existing system by establishing a word model domain of a speech recognition system by context information. This system shows operation without decrease of recognition rate in smart home environment.

• The Journal of the Acoustical Society of Korea
• /
• v.29 no.7
• /
• pp.468-474
• /
• 2010

The major factor that disturbs practical use of speech recognition is distortion by the ambient and channel noises. Generally, the ambient noise drops the performance and restricts places to use. DSR (Distributed Speech Recognition) based speech recognition also has this problem. Various noise cancelling algorithms are applied to solve this problem, but loss of spectrum and remaining noise by incorrect noise estimation at low SNR environments cause drop of recognition rate. This paper proposes methods for speech enhancement. This method uses MMSE-STSA for noise cancelling and ideal binary mask to compensate damaged spectrum. According to experiments at noisy environment (SNR 15 dB ~ 0 dB), the proposed methods showed better spectral results and recognition performance.