• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2000.11a
• /
• pp.113-117
• /
• 2000

수작업으로 이루어지고 있는 환경의 영향이나 작업의 영향에 따른 정신피로나 신체피로의 주관적인 평가를 자동화하기 위한 방법에 대하여 논하였다. 사람의 가장 자연스러운 의사소통인 평가어를 척도로 하여 평가가 이루어지는 음성인식기술을 응용한 주관평가법에 대하여 연구하였다. 주관평가의 자동화를 위하여 우선, 평가어에 대한 음성 인식을 한 후 인식된 평가 결과 데이터를 이용하여 설문지를 자동 생성시킴과 동시에 파일 형태로 저장시켰다. 음성 인식 알고리즘으로는 DTW(Dynamic Time Warping)인식 알고리즘을 사용하였고. 설문지 질의 내용은 집중도 평가를 이용하였다. 인식실험은 설문에 대한 응답에 필요한 평가어를 대상으로 하였다.

• The Journal of the Acoustical Society of Korea
• /
• v.17 no.7
• /
• pp.22-27
• /
• 1998

본 논문에서는 ETRI에서 개발한 가변 어휘 음성 인식기의 어휘 독립 음향 모델링 방법을 기술하고, 이 모델의 어휘 종속, 어휘 독립 및 어휘적응 성능을 평가하기 위하여 다 양한 고립단어 및 연속음성 DB에 대하여 실험한 결과를 분석하였다. 평가를 위하여 사용한 음성 DB로는 고립단어 음성으로 POW(Phonetically Optimized Words) 3848, PBW(Phonetically Balanced Words) 445, PBW 452, 호텔예약 244 단어, 게임 제어용 단어 등이며, 연속음성으로 일반 문장 음성 및 연속 숫자음을 이용하였다. 성능 분석 결과 40개 음소 모델만으로도 비교적 높은 인식률을 보여 주었지만, 어휘독립의 경우는 어휘종속에 비 하여 성능이 크게 낮았고, 특히 대상 어휘가 숫자음, 알파벳, 연속음 등의 경우에는 POW 데이터나 PBW 데이터만 가지고는 우수한 가변 어휘 음성 인식기를 구현하기에 한계가 있 음을 알 수 있다. 또한, 훈련 데이터의 어휘와 평가데이터의 어휘가 비슷할 경우에는 변이음 모델을 사용하면 음소 모델만을 사용할 경우에 비하여 그 성능이 우수하였지만, 일반적인 어휘독립의 상황에서는 효과가 별로 없음을 알 수 있었다.

• 재활복지
• /
• v.22 no.1
• /
• pp.157-168
• /
• 2018

The purpose of this study is to investigate the impact of semi-closed vocal training-based Vocal Aerobic Treatment on the voice improvement of soprano. Study subject was one soprano who appealed to the suffering of her voice problem due to vocal cord nodule. A study method of conducting pre/post acoustic evaluation and subjective voice evaluation to compare the measures was used; Vocal Aerobic Treatment was carried out twice a week for a total of 32 session. In the acoustic evaluation, MDVP (multi-dimensional voice program) and VRP (voice range profile) were used to evaluate the pitch, voice quality, and voice range; in the subjective voice evaluation, SVHI (singing voice handicap index) was used to assess voice satisfaction. As a result of the pitch evaluation, the soprano maintained a proper Fo. As a result of the voice quality evaluation, the jitter, shimmer, and the noise harmonic ratio numbers decreased compared to the numbers shown before the treatment. As a result of the voice range evaluation, the scope of the range was broadened, with the number of semitone increasing from 30 to 35. As for the subjective voice evaluation, the result of the total score obtained after the survey report divided by the number of questions showed a decrease from 3.6 to 0.6. The soprano herself reported of having a minor extent of a voice problem. The summary of the above results reflects that Vocal Aerobic Treatment is useful in the voice improvement of vocalists However, as this study is case research regarding the Vocal Aerobic Treatment effect on one soprano, further research on the treatment effect covering many other vocalists is necessary. Also, there is a need for follow-up studies regarding voice management and voice treatment program on not only the vocalists but also the voice users in many other professions.

• Information and Communications Magazine
• /
• v.11 no.9
• /
• pp.79-92
• /
• 1994

음성인식 및 합성 시스템으로 대표되는 음성입출력장치의 개발 및 실용화에 따라 이들의 성능을 객관적으로 평가할 수 있는 방법이 중요하게 되었다. 그러나 국내에서의 이분야에 대한 체계적인 연구는 맞지 않다. 본고에서는 이와 관련한 지금까지의 국내외 연구예들을 검토하고 우리 실정에 맞는 평가법 작성을 위한 방법을 모색하고자 한다. 음성입출력장치의 객관적인 평가방법이 확립되면 연구 개발자는 여러방식의 우열을 객관적으로 파악할 수 있고 신뢰성있는 시스템을 효율적으로 개발할 수 있으며 관련제품의 사용자입장에서는 제품간의 성능비교가 가능하게 될 것이다.

• Phonetics and Speech Sciences
• /
• v.13 no.4
• /
• pp.67-74
• /
• 2021

This study examines the effectiveness of semi-occluded vocal tract exercise (SOVTE) conducted through telepractice for 10 female teachers who have experienced vocal discomfort while working in early childhood education facilities (childcare centers, kindergartens). The effects of SOVTE conducted through telepractice were evaluated based on the Korean voice handicap index (KVHI), the Korean version of the voice activity and participation profile (K-VAPP), vocal effort, and auditory perception evaluation by using the grade, roughness, breathiness, asthenia, and strain (GRBAS) scale. The results show that total, functional, and physical scores of KVHI significantly reduced after SOVTE. The total score in K-VAPP significantly reduced after SOVTE. Moreover, vocal effort significantly decreased after SOVTE. However, statistically significant differences were not noted in GRB scales before and after SOVTE. In conclusion, early childhood teachers experienced reduced vocal discomfort SOVTE conducted through telepractice. The study results indicate that voice therapy conducted through telepractice is an effective method for reducing vocal discomfort in early childhood teachers.

• Journal of Software Assessment and Valuation
• /
• v.15 no.2
• /
• pp.73-83
• /
• 2019

Voice Recognition has recently been improved with AI(Artificial Intelligence) and has greatly improved the false recognition rate and provides an effective and efficient Human Machine Interface (HMI). This trend has also been applied in the defense industry, particularly in the aviation, F-35. However, for the quality evaluation of Voice Recognition, the defense industry, especially the aircraft, requires measurable quantitative models. In this paper, the quantitative evaluation model is proposed for applying Voice Recognition to aircraft. For the proposal, the evaluation items are identified from the Voice Recognition technology and ISO/IEC 25000(SQuaRE) quality attributes. Using these two perspectives, the quantitative evaluation model is proposed under aircraft operation condition and confirms the evaluation results.

• Proceedings of the Korean Society for Emotion and Sensibility Conference
• /
• 2001.05a
• /
• pp.276-279
• /
• 2001

환경이나 작업부하 등이 인간에게 주는 피로나, stress 또는 쾌, 불쾌감 등의 감성을 평가하기 위한 정신물리학적인 방법의 하나으로 설문지에 의한 주관적인 평가법이 많이 사용되고 있다. 기존의 수작업으로 이루어지던 설문 방식을 자동화하여 PC 기반으로 설문양식을 자동 생성하고 음성을 통해 응답할 수 있는 프로그램을 개발하였다. 주관평가 자동화 시스템은 주관평가 데이터를 효율적으로 처리를 할 수 있고 음성을 이용함으로써 피험자의 정신적 부담을 경감시키며 생리신호와 주관평가와의 경시적인 변화를 효과적으로 평가할 수 있다. 설문 형식으로는 5점 척도와 7점 척도를 선택하였으며 평가어는 “매우 아니다”∼“매우 그렇다”로 구성되었다. 평가어를 인식함에 있어 좋은 인식률을 얻기 위한 특징벡터의 치수와 기본 프레임 개수를 대상으로 인식실험을 하였다.

• Proceedings of the KSLP Conference
• /
• 2015.03a
• /
• pp.12-12
• /
• 2015

• Phonetics and Speech Sciences
• /
• v.16 no.1
• /
• pp.67-76
• /
• 2024

Recent advances in text-to-speech (TTS) technology have significantly improved the quality of synthesized speech, reaching a level where it can closely imitate natural human speech. Especially, TTS models offering various voice characteristics and personalized speech, are widely utilized in fields such as artificial intelligence (AI) tutors, advertising, and video dubbing. Accordingly, in this paper, we propose a one-shot multi-speaker TTS system that can ensure acoustic diversity and synthesize personalized voice by generating speech using unseen target speakers' utterances. The proposed model integrates a speaker encoder into a TTS model consisting of the FastSpeech2 acoustic model and the HiFi-GAN vocoder. The speaker encoder, based on the pre-trained RawNet3, extracts speaker-specific voice features. Furthermore, the proposed approach not only includes an English one-shot multi-speaker TTS but also introduces a Korean one-shot multi-speaker TTS. We evaluate naturalness and speaker similarity of the generated speech using objective and subjective metrics. In the subjective evaluation, the proposed Korean one-shot multi-speaker TTS obtained naturalness mean opinion score (NMOS) of 3.36 and similarity MOS (SMOS) of 3.16. The objective evaluation of the proposed English and Korean one-shot multi-speaker TTS showed a prediction MOS (P-MOS) of 2.54 and 3.74, respectively. These results indicate that the performance of our proposed model is improved over the baseline models in terms of both naturalness and speaker similarity.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.2
• /
• pp.127-134
• /
• 2009

In this study, possibility of the laser eavesdropping is investigated on the window glasses with various thicknesses, Glass windows are excited by maximum length sequency (MLS) signal and the vibration sound is detected by a laser doppler vibrometer. From the detected sound, speech intelligibility is objectively estimated. Speech transmission index (STI), which is based on the modulation transfer function (MTF). is calculated for the estimation. Finally, disturbing wave effect on the speech intelligibility is analysed by using an outside speaker and a window shaker attached on the glass window. The purpose of the study is to estimate the possibility of remote eavesdropping by the laser sensor and to evaluate the performance of the homemade window shaker to protect from the remote eavesdropping.