• Speech Sciences
• /
• v.2
• /
• pp.9-23
• /
• 1997

Through iterative analysis and synthesis experiments on Korean monosyllables, the Korean text-to-speech system was implemented using the phoneme-based formant synthesis method. Since the formants of initial and final consonants in this system showed many variations depending on the medial vowels, the database for each phoneme was made up of formants depending on the medial vowels as well as duration information of transition region. These techniques were needed to improve the intelligibility of synthetic speech. This paper investigates also methods of concatenating the synthesis units to improve the quality of synthetic speech.

• Speech Sciences
• /
• v.8 no.1
• /
• pp.53-66
• /
• 2001

This paper aims to investigate the acoustic characteristics of the vowels attested in Korean and French and to seek a way of understanding them from a phonological point of view. We first compare the two vowel systems by measuring the actual frequencies of the formants using CSL. It is shown that the first and second formants vary in wider range in French compared to Korean. In order to understand the two vowel systems from a phonological point of view, we apply the theory of Licensing Constraints, proposed and developed by Kaye (1994), and Charette and Kaye (1994). We propose the licensing constraints placed upon the vowels both in Korean and French. For Korean, we propose the licensing constraints such that both elements I and U must be heads. For French, we claim the following licensing constraints: U in a headed expression must be head, A cannot be head, and Nothing can only license an expression A in it.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1994.06a
• /
• pp.971-976
• /
• 1994

A speech display system is developed for the evaluation and the training of speech utterance. The speech is analyzed by linear predictive technique every 5 ms and the frequencies of the lowest two spectral local peaks P1 and P2 are extracted. The vowel trakectory is displayed using those frequencies on th P1-P2 plane. In most cases, P1 and P2 correspond to the first and the second formants, but in the case of indistinct utterance, the correspondence between the local spectral peaks and the formants tends to fall into disorder. And the system is considered to be useful for the evaluation of speech quality. The examples of some words uttered by normal speakers and some patients with difficulty in utterance are compared each other for the discussion of the effectiveness of the system.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.25 no.12
• /
• pp.1503-1512
• /
• 1988

As the basic analysis to solve the within-and cross-speaker variability in phoneme based speech recognition, changes in pitch and formant frequencies of 8 Korean vowels with age and sex of speaker has been investigated by analyzing a large number fo samples. Conclusions obtained are as follows: 1) Changes in pitch frequency with age and sex of speaker for children are hard to distinguish and the difference of before and after the voice change is analyzed approximately 0.2 oct. for female an 0.9 oct. for male. 2) While most of the formants of vowel considerably change with the age of speaker, the change becomes smaller as the age becomes older. 3) While there is an indirect correlation between pitch and formant with change in age, it is hard to see a direct correlation. 4) When the objects of the recognition experiment by pitch and formants are various speakers in each age and sex, pitch also works as an efficient recognition parameter.

• Phonetics and Speech Sciences
• /
• v.2 no.4
• /
• pp.11-18
• /
• 2010

This study compared classical operatic singing and applied music singing using the vocal assessment software, Dr. Speech and SPEAD from Lx Speech Studio. Participants in this study included: eight female operatic singers (mean 22.6 yrs, average career 7.5 yrs); eight male operatic singers (mean 25.6 yrs, average career 7.3 yrs); eight female applied music singers (mean 25.1 yrs, average career 6.1 yrs); and eight male applied music singers (mean 27.6 yrs, average career 6.8 yrs). The results demonstrated significantly higher closed quotient values in female applied music singers during singing (p<.05). In addition, higher closed quotient values in speaking were presented in male classical singers and longer MPT was obtained in female operatic singers (p<.05). Furthermore, singer's formants were identified in all male operatic singers and in three female operatic singers. In contrast, only one applied music male and one female singer showed singer's formants while singing.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.2E
• /
• pp.31-38
• /
• 1999

This paper focuses on extracting formants from transfer function, derived from linear prediction analysis of speech signal. The second derivative of the log magnitude spectrum of the transfer function, the first and third derivatives of the phase spectrum of the transfer function in the z-plane are discussed. Their resolutions of detecting formants are analyzed and some comparisons are given. Theoretical analyses and experimental results show that the third derivative of the phase spectrum decays more rapidly around the formant locations than the first derivative of the phase spectrum and the second derivative of the log magnitude spectrum. Compared with the second derivative of the log spectrum and the first derivative of the phase spectrum, the third derivative of the phase spectrum has higher resolution in frequency domain and provides more accurate formant extraction.

• Proceedings of the KSPS conference
• /
• 2005.11a
• /
• pp.97-100
• /
• 2005

The aim of the present study is to compare the vowel formants between generations in Daegu dialect. 20 Daegu dialect speakers were participated in this study; 10 were in their 40's, the other 10were in their 20's. As a result, the distance of /ㅣ/ and /ㅐ/, and, /ㅡ/ and /ㅓ/ in 20's is further than 40's, while the distance of /ㅗ/ and in 20's is closer than 40's. It seems reasonable to conclude that vowels in Daegu dialect change to have their own stable space, but /ㅗ/ and /ㅜ/ does not.

• Journal of the Korean Society of Laryngology, Phoniatrics and Logopedics
• /
• v.16 no.1
• /
• pp.10-14
• /
• 2005

Background and Objectives : Vowels are characterized on the basis of formant patterns. The first formant(F1) is determined by high-low placement of the tongue, and the second formant (F2) by front-back placement of the tongue. The fundamental frequency(F0) of a soprano often exceed the normal frequency of the first formant. And the vocal intensity is boosted when F0 is high and a harmonic coincides with a formant. This is called a formant tuning. Experienced singers thus learned how to tune their formants over a resonable range by lowering the tongue to maximize their vocal intensity. So, the current study aimed to identify the formant tuning in one experienced soprano by comparing the first formants of vowel [i] in three different voice production : speech, ascending scale, and vocalize. Materials and Method : All voices recordings of vowel [i] in speech, ascending scale (from F4 note to A4 note), and vocalize(:Ridente la calam") were made with digital audio tape-corder in a sound treated room. And the captured data were analyzed by the long term average(LTA) power spectrum using the FFT algorithm of the Computerized Speech Lab(CSL, Kay elementrics, Model, 4300B). Results : Although the first formant of vowel [i] in speech was 238Hz, those of ascending scale [i] were 377Hz, 405Hz, 453Hz respectively in F4(349z), G4(392Hz), A4(440Hz) note, and 722Hz, 820Hz, 918Hz respectively in F5 (698Hz), G5(784Hz), A5(880Hz) note. In vocalize, first formants of [i] were 380Hz, 398Hz, 453Hz respectively in F4, G4, A4 note, and 720Hz, 821Hz, 890Hz respectively in F5, G5, A5 note. Conclusion : These results showed that the first formant of ascending scale and vocalize sustained higher frequency than fundamental frequency in high pitch. This finding implicates that the formant tuning of vowel [i] in ascending scale was also noted in vocalize.

• Phonetics and Speech Sciences
• /
• v.11 no.4
• /
• pp.97-108
• /
• 2019

An automated Praat script was implemented to measure optimal formant frequencies for adults. Optimal formant analysis could be interpreted to show that the deviation of formant frequency that resulted from the two variously combined setting parameters (maximum formant and number of formants) was minimal. To increase the reliability of formant analysis, LPC order should be set differently, based on the gender or vowel type. Praat recommends 5,000 Hz and 5,500 Hz as maximum formant settings and, at the same time, recommends 5 as the number of formants for males and females. However, verification is needed to determine whether these recommended settings are valid for Korean vowels. Statistical analysis showed that formant frequencies significantly varied across the adapted scripts, especially with respect to the data on females. Formant plots and statistical results showed that linear_script and qtone_script are much more reliable in formant measurements. Among four kinds of scripts, the linear and qtone_scripts proved to be more stable and reliable. While the linear_script was designed to have a linearly increased formant step in for-loop, the increment of formant step in the qtone_script was arranged by quarter tone scale (base frequency×common ratio ($\sqrt[24]{2}$)). When looking at the tendency of the formant setting drawn by the two referred algorithms in the context of front vowel [i, e], the maximum formant was set higher; and the number of formants set at a lower value than recommended by Praat. The back vowel [o, u], on the contrary, has a lower maximum formant and a higher number of formants than the standard setting.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.1042-1045
• /
• 2003

In this paper, we propose the speaker identification system that uses vowel that has speaker's characteristic. System is divided to speech feature extraction part and speaker identification part. Speech feature extraction part extracts speaker's feature. Voiced speech has the characteristic that divides speakers. For vowel extraction, formants are used in voiced speech through frequency analysis. Vowel-a that different formants is extracted in text. Pitch, formant, intensity, log area ratio, LP coefficients, cepstral coefficients are used by method to draw characteristic. The cpestral coefficients that show the best performance in speaker identification among several methods are used. Speaker identification part distinguishes speaker using Neural Network. 12 order cepstral coefficients are used learning input data. Neural Network's structure is MLP and learning algorithm is BP (Backpropagation). Hidden nodes and output nodes are incremented. The nodes in the incremental learning neural network are interconnected via weighted links and each node in a layer is generally connected to each node in the succeeding layer leaving the output node to provide output for the network. Though the vowel extract and incremental learning, the proposed system uses low learning data and reduces learning time and improves identification rate.