• 말소리와 음성과학
• /
• 제1권3호
• /
• pp.151-154
• /
• 2009

The external ear generates resonance gain because of anatomical characteristics. The ear canal resonance is influenced by the length and volume of the ear canal, the pinna, the concha cavity, the body trunk, and the speed of sound wave. This study is focus on the influence of the volume of ear canal. 17-healthy-adult (32 ears) were participated. They did not have any medical and ear disease history. The maximum resonance frequency of the ear canal was 2675 (${\pm}265$) Hz at azimuth $0^{\circ}$ and 2784 (${\pm}268$) Hz at azimuth $45^{\circ}$. The resonance gain was 18.1 (${\pm}3.9$) dB at azimuth $0^{\circ}$ and 17.9 (${\pm}3.8$) dB at azimuth $45^{\circ}$, respectively. The ear canal volume was 0.78 (${\pm}0.2$) cc and 1.32 (${\pm}0.8$) cc including static compliance. The ear canal resonance was changed depending on the ear canal volume. It was also statistically correlated at azimuth $0^{\circ}$ (p=0.038) and $45^{\circ}$ (p=0.013), respectively. The resonance gain was not correlated with the ear canal volume. The change of resonance frequency according to the ear canal volume will be useful information in the field of audiological rehabilitation especially for hearing aids fitting. In addition, we expected this study can provide the basic information for the study of the external ear resonance characteristics.

• 말소리와 음성과학
• /
• 제10권2호
• /
• pp.51-60
• /
• 2018

This study aims to compare the word decoding skills, phonological awareness (PA), rapid automatized naming (RAN) skills, and letter knowledge of first graders with developmental dyslexia (DD) and those who were typically developing (TD). Eighteen children with DD and eighteen TD children, matched by nonverbal intelligence and discourse ability, participated in the study. Word decoding of Korean language-based reading assessment(Pae et al., 2015) was conducted. Phoneme-grapheme correspondent words were analyzed according to whether the word has meaning, whether the syllable has a final consonant, and the position of the grapheme in the syllable. Letter knowledge asked about the names and sounds of 12 consonants and 6 vowels. The children's PA of word, syllable, body-coda, and phoneme blending was tested. Object and letter RAN was measured in seconds. The decoding difficulty of non-words was more noticeable in the DD group than in the TD one. The TD children read the syllable initial and syllable final position with 99% correctness. Children with DD read with 80% and 82% correctness, respectively. In addition, the DD group had more difficulty in decoding words with two patchims when compared with the TD one. The DD group read only 57% of words with two patchims correctly, while the TD one read 91% correctly. There were significant differences in body-coda PA, phoneme level PA, letter RAN, object RAN, and letter-sound knowledge between the two groups. This study confirms the existence of Korean developmental dyslexics, and the urgent need for the inclusion of a Korean-specific phonics approach in the education system.

• 말소리와 음성과학
• /
• 제12권1호
• /
• pp.11-17
• /
• 2020

비성조 모국어 화자가 성조 언어를 학습할 때 모국어에 없는 성조 체계 학습에 어려움을 보인다. 본 연구는 비성조 언어인 한국어 모국어 성인 화자가 중국어를 학습할 때 중국어 성조의 지각과 산출에서 어떤 어려움을 보이는지 규명하고자 하였다. 추가적으로 중국어 분절음이 한국어와 유사하지 않을 경우 초분절음인 성조 지각 및 산출에 어려움을 가중시키는지 알아보기 위해 중국어 자음의 한국어 자음과의 유사성을 조작하여 실험을 진행하였다. 실험 결과, 지각과 산출 간 정확도 및 반응시간이 각 성조 별로 차이를 보였다. 즉, 1성과 2성에서는 지각 과제에서 산출 과제에 비해 어려움을 보인 반면 3성에서는 산출 과제에서 지각 과제보다 어려움을 보였다. 오류 패턴 분석 결과, 지각 과제에서 2성과 3성을 서로 혼동을 하는 반면 산출 과제에서는 2성을 3성으로 산출하는 오류는 줄어들고 3성을 2성으로 산출하는 오류가 증가한 것을 보였다. 마지막으로, 한국어 유사성의 조작은 성조의 지각 및 산출에 영향을 주지 않았다.

• 말소리와 음성과학
• /
• 제6권2호
• /
• pp.55-64
• /
• 2014

This study measured formant bandwidths of profound hearing impaired children and examined the characteristics of their articulation. For this study, 10 cochlear implanted children(CI), 10 hearing aid children(HA) and 10 normal hearing children(NH) were asked to read 7 Korean vowels(/ɑ, ʌ, o, u, ɯ, i, ɛ/). The subjects' readings were recorded by NasalView and analyzed by Praat. The analysis of the formant bandwidths explains the degree of vocal fold opening and the characteristics of radiation. Through the analysis of formant bandwidth, we can see that the hearing-impaired maintain vocal fold tension when they speak high vowels and characteristics of radiation. Narrower B1 means better maintain vocal fold tension, wider B2 means more front and wider B3 means the rounder lips. CI's B1 was widest and NH's was narrowest. And females' B1 was wider than males'. Among vowels, B1 of /a/ was widest, and B1 of /i/ was narrowest. In the case of B2, HA and NH's B2 was wider than CI's. Females' B2 was wider than males'. And B2 of /i/ was widest, and B2 of /ʌ/ was narrowest. In the case of B3, NH's was widest, and CI's was narrowest. Males' was wider than females'. Among vowels, B3 of /o/ was widest, and B3 of /ɛ/ was narrowest. As a result, first, through the analysis of B1, we can find that NH and males could better maintain vocal fold tension than the hearing-impaired or females, and all children articulate /i/ with vocal fold tension than other vowels. Second, through the analysis of B2, NH and HA articulate vowels with the weaker rounded than CI does. And females articulate vowels with the weaker rounded than males do. Third, through the analysis of B3, NH articulate vowels with the rounder than HA or CI do, and males articulate vowels with the rounder than females do. Through the results, we can expect that the analysis of formant bandwidth will be applied to the therapy of articulation for the hearing-impaired with hearing aids or cochlear implant.

• 말소리와 음성과학
• /
• 제6권2호
• /
• pp.37-45
• /
• 2014

The purpose of this study was to determine whether the Korean Singing Voice Handicap Index (K-SVHI) was suitable for singers in other genres than vocal music to assess their vocal problems subjectively. Twenty six college students majoring in vocal music and twenty six students majoring in applied music were included in the study. They were divided into G0 and G1 in voice quality using the GRBAS scale during the tasks of singing. K-SVHI was divided into three sub-areas (Physical, Functional, and Emotional). In the singing task, both groups showed no significant difference between K-SVHI scores by G scale. In the reading task, the vocal music group had significantly higher K-SVHI in G0 than in G1 in K-SVHIs by G scale, while the applied vocal music group had significantly higher K-SVHI in G1 than in G0. Also, the two groups were not significantly different in G0, G1 in the singing task while the vocal music group showed higher K-SVHI than the applied vocal music group in G0 in the reading task. In addition, the vocal music group had higher K-SVHI than the applied vocal music group in G1 in both tasks. As comparing by groups in three sub-areas of K-SVHI, significant differences were found in the Emotional and Functional area. Those results showed that singers felt their voice problems differently by musical genres, which means that K-SVHI may not be a proper tool for evaluating voice handicap of singers in diverse voice music genres.

• 말소리와 음성과학
• /
• 제12권3호
• /
• pp.55-63
• /
• 2020

음성 변환은 다양한 음성 처리 응용에 적용될 수 있으며, 음성 인식을 위한 학습 데이터 증강에도 중요한 역할을 할 수 있다. 기존의 방법은 음성 합성을 이용하여 음성 변환을 수행하는 구조를 사용하여 멜 필터뱅크가 중요한 파라미터로 활용된다. 멜 필터뱅크는 뉴럴 네트워크 학습의 편리성 및 빠른 연산 속도를 제공하지만, 자연스러운 음성파형을 생성하기 위해서는 보코더를 필요로 한다. 또한, 이 방법은 음성 인식을 위한 다양한 데이터를 얻는데 효과적이지 않다. 이 문제를 해결하기 위해 본 논문은 원형 스펙트럼을 사용하여 음성 신호 자체의 변환을 시도하였고, 어텐션 메커니즘으로 스펙트럼 성분 사이의 관계를 효율적으로 찾아내어 변환을 위한 자질을 학습할 수 있는 transformer 네트워크 기반 딥러닝 구조를 제안하였다. 영어 숫자로 구성된 TIDIGITS 데이터를 사용하여 개별 숫자 변환 모델을 학습하였고, 연속 숫자 음성 변환 디코더를 통한 결과를 평가하였다. 30명의 청취 평가자를 모집하여 변환된 음성의 자연성과 유사성에 대해 평가를 진행하였고, 자연성 3.52±0.22 및 유사성 3.89±0.19 품질의 성능을 얻었다.

• 말소리와 음성과학
• /
• 제7권3호
• /
• pp.45-53
• /
• 2015

The current study investigated CVCC syllables in spontaneous American English speech to find out whether such syllables are produced as phonological units with a string of segments, showing a hierarchical structure. Transcribed data from the Buckeye Speech Corpus was used for the analysis in this study. The result of the current study showed that the constituents within a CVCC syllable as a phonological unit may have phonetic variations (namely, the final coda may undergo deletion). First, voiceless alveolar stops were the most frequently deleted when they occurred as the second final coda consonants of a CVCC syllable; this deletion may be an intermediate process on the way from the abstract form CVCC (with the rime VCC) to the actual pronunciation CVC (with the rime VC), a production strategy employed by some individual speakers. Second, in the internal structure of the rime, the proportion of deletion of the final coda consonant depended on the frequency of the word rather than on the position of postvocalic consonants on the sonority hierarchy. Finally, the segment following the consonant cluster proved to have an effect on the reduction of that cluster; more precisely, the following contrast was observed between obstruents and non-obstruents, reflecting the effect of sonority: when the segment following the consonant cluster was an obstruent, the proportion of deletion of the final coda consonant was increased. Among these results, the effect of word frequency played a critical role for promoting the deletion of the second coda consonant for clusters in CVCC syllables in spontaneous speech. The current study implies that the structure of syllables as phonological units can vary depending on individual speakers' lexical representation.

• 말소리와 음성과학
• /
• 제7권3호
• /
• pp.13-26
• /
• 2015

Speech production studies have suggested that bilinguals who are L2-dominant are the most likely to suppress the influence of the first language (L1) on the second language (L2). The voice onset times (VOTs) and fundamental frequencies (f0s) of monolingual and bilingual speakers of English and Korean were examined to address the question whether cross language influences occur particularly in L2 predominant bilinguals and to compare their outcomes with those of L2 proficient bilinguals and monolinguals. A total of 28 speakers participated in this experiment and they produced English and Korean stops in the carrier sentence. In English, for voiceless aspirated and unaspirated stops, L2 predominant bilingual speakers produced VOTs that were significantly shorter than those of monolingual English speakers. The outcome was analogous in Korean speech. For aspirated and lax stops, they produced shorter Korean VOTs than monolingual speakers. The results of f0s were slightly different from those of VOTs. In English, L2 predominant bilinguals produced f0s that were not significantly different from those of monolingual English speakers. In Korean, however, they produced f0s that were significantly different from those of monolingual Korean speakers. Taken VOT and f0 into consideration together, the overall results suggest that, although they tend to show a corresponding pattern of monolinguals, L2 predominant bilinguals had cross language phonetic influences between L1 and L2, similar to L2 proficient bilinguals. Between the two acoustic cues, f0 seemed to be a more reliable cue than VOT to examine the influences.

• 말소리와 음성과학
• /
• 제7권3호
• /
• pp.139-149
• /
• 2015

This paper aims to analyze pronunciation variations of loanwords produced by Korean and improve the performance of pronunciation modeling of loanwords in Korean by using syllable-based segmentation and phonological knowledge. The loanword text corpus used for our experiment consists of 14.5k words extracted from the frequently used words in set-top box, music, and point-of-interest (POI) domains. At first, pronunciations of loanwords in Korean are obtained by manual transcriptions, which are used as target pronunciations. The target pronunciations are compared with the standard pronunciation using confusion matrices for analysis of pronunciation variation patterns of loanwords. Based on the confusion matrices, three salient pronunciation variations of loanwords are identified such as tensification of fricative [s] and derounding of rounded vowel [ɥi] and [$w{\varepsilon}$]. In addition, a syllable-based segmentation method considering phonological knowledge is proposed for loanword pronunciation modeling. Performance of the baseline and the proposed method is measured using phone error rate (PER)/word error rate (WER) and F-score at various context spans. Experimental results show that the proposed method outperforms the baseline. We also observe that performance degrades when training and test sets come from different domains, which implies that loanword pronunciations are influenced by data domains. It is noteworthy that pronunciation modeling for loanwords is enhanced by reflecting phonological knowledge. The loanword pronunciation modeling in Korean proposed in this paper can be used for automatic speech recognition of application interface such as navigation systems and set-top boxes and for computer-assisted pronunciation training for Korean learners of English.

• 말소리와 음성과학
• /
• 제3권1호
• /
• pp.145-151
• /
• 2011

In view of the hypothesis that the effects of Parkinson's disease on voice production can be detected before pharmacological intervention, the prosodic features of patients with idiopathic Parkinson's disease (IPD) and a healthy aging group were diagnostically analyzed with the long term object of establishing, for clinical purposes, early disease-progression biomarkers. Twenty patients (male 8; female 12) with IPD (prior to pharmacological intervention) and a healthy control group of 22 (male 10; female 12) were selected. Ten sentences were recorded with a head-worn microphone. One sentence was chosen for the analysis of this paper. Relevant parameters, i.e. 3-dimensional model (F0, intensity, duration) and pitch and intensity related slopes (maxEnergy, maxF0, meanAbS, semiT, meanEnergy, meanF0), were analyzed by two-group discriminant analysis. The stepwise estimation method of discriminant analysis was performed by gender. The discriminant functions predicted 83.9% of the male test data correctly while the prediction rate was 93.1% for the female group. The results showed that meanF0_slope and semiT_slope were more important parameters than the others for the male group. For the female group, the meanEnergy_slope and maxEnergy_slope were the important ones. These findings indicate that significant parameters are different for the male and female group. Gender lifestyle may be responsible for this difference. Dysprosodic features of IPD show not simultaneously but progressively in terms of F0, intensity and duration.