The purpose of this study was to analyze the relationship between segmental correctness and speech intelligibility in adults with hearing impairment. Segmental correctness was measured by percentage of correct vowels(PCV) and percentage of correct consonants(PCC). The results were shown as follows: First, PCV and PCC could predict speech intelligibility with statistical significance. Second, in consonant classes divided by place and manner of articulation, the PCC of plosives and alveolar sounds were significant predicting variables in each group ($R^{2}=50%;\;59%$). According to this study, the importance of segmental correctness on speech intelligibility of adults with hearing impairment was confirmed. Also correctness of plosive sounds in manner and alveolar sounds in place were significant factors to speech intelligibility.

The present study investigates acoustic cues of russian soft plosive consonants. In previous studies, russian soft consonants are distinguished from hard consonants by F1, F2 of following vowels. The result showed: (1) that F0 of soft plosive consonants in following vowels were lower than those of hard plosive consonants; (2) and that VOT of soft plosive consonants were longer than those of hard plosive consonants. Hence, the present that, in addition to F1, F2, VOT and F0 are detected as acoustic cues that differentiate soft plosive consonants from hard plosive consonant in Russian.

This study investigated the roles of name agreement and word frequency in the English-Korean word translation task. Using the low-frequency homonyms with low name agreement as stimuli, Experiment 1 revealed that the name agreement of materials is a determinant which could modulate times to translate English words into Korean equivalents. On the contrary, Experiment 2 showed that the name agreement of materials does not play a decisive role in the translation task, using the low-frequency homonyms having high name agreement as stimuli. In Experiment 3, we identified that the frequency effects observed from previous two experiments are indeed brought about during the lexical access. Our findings suggest that the word frequencies of materials have a strong influence on English-Korean word translation times, and homonyms are represented independently each other in the lexeme level.

This study focused on the improvement of prosody transplantation technology to be used for effective prosody education. Issues making the technology a less acceptable tool for prosody education were addressed. Instead of merely copying the target pitch onto a learner's utterances, the target pitch was resealed in semitone before the transplantation. In so doing, distortion of a signal was minimized and the transplanted utterance could have the quality of sound not different from the learner's utterances. Instead of manual transplantation, an automatic procedure was proposed to increase the reliability and the consistency of the outcome and enable real time processing. The perceptual performance of the automatic transplantation was evaluated by the perception experiment showing the automatic ransplantation was as good as the manual process.

Discrimination of music and speech from the multimedia signal is an important task in audio coding and broadcast monitoring systems. This paper deals with the problem of feature parameter extraction for discrimination of music and speech. The wavelet transform is a multi-resolution analysis method that is useful for analysis of temporal and spectral properties of non-stationary signals such as speech and audio signals. We propose new feature parameters extracted from the wavelet transformed signal for discrimination of music and speech. First, wavelet coefficients are obtained on the frame-by-frame basis. The analysis frame size is set to 20 ms. A parameter $E_{sum}$ is then defined by adding the difference of magnitude between adjacent wavelet coefficients in each scale. The maximum and minimum values of $E_{sum}$ for period of 2 seconds, which corresponds to the discrimination duration, are used as feature parameters for discrimination of music and speech. To evaluate the performance of the proposed feature parameters for music and speech discrimination, the accuracy of music and speech discrimination is measured for various types of music and speech signals. In the experiment every 2-second data is discriminated as music or speech, and about 93% of music and speech segments have been successfully detected.

In this paper, we propose a scalable audio coder, which has a variable bandwidth from the narrowband speech bandwidth to the audio bandwidth and also has a bit-rate from 8 to 320 kbits/s, in order to cope with the quality of service(QoS) according to the network load. First of all, the proposed scalable coder splits bandwidth of the input audio into narrowband up to around 4 kHz and above. Next, the narrowband signals are compressed by a speech coding method compatible to an existing standard speech coder such as G.729, and the other signals whose bandwidth is above the narrowband are compressed on the basis of a psychoacoustic model. It is shown from the objective quality tests using the signal-to-noise ratio(SNR) and the perceptual evaluation of audio quality(PEAQ) that the proposed scalable audio coder provides a comparable quality to the MPEG-1 Layer III (MP3) audio coder.

A punctuation generation system which combines prosodic information with acoustic and language model information is presented. Experiments have been conducted first for the reference text transcriptions. In these experiments, prosodic information was shown to be more useful than language model information. When these information sources are combined, an F-measure of up to 0.7830 was obtained for adding punctuation to a reference transcription. This method of punctuation generation can also be applied to the 1-best output of a speech recogniser. The 1-best output is first time aligned. Based on the time alignment information, prosodic features are generated. As in the approach applied in the punctuation generation for reference transcriptions, the best sequence of punctuation marks for this 1-best output is found using the prosodic feature model and an language model trained on texts which contain punctuation marks.

A modified speech recogniser have been proposed for automatic capitalisation generation to improve the readability of English speech recognition output. In this modified speech recogniser, every word in its vocabulary is duplicated: once in a de-caplitalised form and again in the capitalised forms. In addition its language model is re-trained on mixed case texts. In order to evaluate the performance of the proposed system, experiments of automatic capitalisation generation were performed for 3 hours of Broadcast News(BN) test data using the modified HTK BN transcription system. The proposed system produced an F-measure of 0.7317 for automatic capitalisation generation with an SER of 48.55, a precision of 0.7736 and a recall of 0.6942.

The environmental differences between training and testing mode are generally considered to be the critical factor for the performance degradation in speaker recognition systems. Especially, general speaker recognition systems try to get as clean speech as possible to train the speaker model, but it's not true in real testing phase due to environmental and channel noise. So in this paper, the new method of weighting the frame-based likelihood according to frame SNR is proposed in order to cope with that problem. That is to make use of the deep correlation between speech SNR and speaker discrimination rate. To verify the usefulness of this proposed method, it is applied to the context dependent speaker identification system. And the experimental results with the cellular phone speech DB which is designed by ETRI for Koran speaker recognition show that the proposed method is effective and increase the identification accuracy by 11% at maximum.