• Proceedings of the KAIS Fall Conference
• /
• 2010.05a
• /
• pp.245-248
• /
• 2010

일정한 스펙트럼 민감도와 낮은 스펙트럼 왜곡을 보이고 선형보간이 용이하다는 장점을 갖는 LSP 파라미터는 음성코덱(codec)이나 인식기에서 음성신호를 분석하여 전송형이나 저장형 파라미터로 변환되어, 주로 저전송률 음성부호화기에 사용된다. 그러나 LPC 계수를 LSP로 변환하는 방법이 복잡하여 계산시간이 많이 소요된다는 단점이 있다. 기존의 LSP 변환 방법 중 음성 부호화기에서 주로 사용하는 real root 방법은 근을 구하기 위해 주파수 영역을 순차적으로 검색하기 때문에 계산시간이 많이 소요되는 단점을 갖는다. 본 논문에서 제안하는 알고리즘은 LSP 분포 특성에 따라 검색구간의 순서와 검색간격을 달리하며, 제1 포만트와 제2 포만트의 연관성을 고려하여 검색구간을 조절한다. 기존의 real root 방법과 제안한 방법을 비교한 결과 검색시간이 평균 48.13% 단축되었다.

• The Journal of Engineering Research
• /
• v.1 no.1
• /
• pp.65-70
• /
• 1997

To transmit the degraded voice signal within various environment surrounding acoustic noises, we extract lip i the face and then compare lip edge features with prestoring DB having features such as mouth height, width, area, and rate. It provides high security and is not affected by acoustic noise because it is not necessary to transmit the actual utterance.

• Proceedings of the KSLP Conference
• /
• 1997.11a
• /
• pp.253-253
• /
• 1997

비폐색이 있는 경우 음성이 변하는 것을 쉽게 느낄 수 있지만, 비폐색 때의 음향학적인 특성에 대하여 알려진 바를 찾기는 쉽지 않다. 저자들은 인위적으로 비폐색을 유발하여 비폐색 부위에 따른 비강자음의 음향학적 변화특성을 파악하고 비음도의 변화 정도를 비교하고자 하였다. 정상비음도를 보이는 성인남녀 각 10명을 대상으로, 2ml의 부피를 갖도록 수술용장갑에 Merocel$^{\circledR}$을 넣은 뒤, 이를 이용하여 ostiomeatal unit(OMU)을 중심으로 전후상하 4부위의 인위적인 비폐색을 유발하여, 비폐색 전과 후의 부위에 빠른 차이틀 비교하였다. /나나/의 발성을 각 조건에서 3회 실시하여, 모음중간의 /ㄴ/중 (CVCV) 안정된 spectorgram소견을 보이는 부위를 선택하여, 해당구간의 제1, 제2, 제3 음형대와 각각의 bandwidth 평균값을 남녀별로 비교하였고, 표준비음비율이 알려진 rabbit, baby, mama 문장을 이용하여 비음비율을 비교하였다. 남녀모두 비폐색전에 비하여 OMU의 앞쪽부위를 막은 경우에 제1음형대가 가장 뚜렷하게 감소되었으며, 비음비율의 유의한 감소를 보였다. 비폐색이 있는 경우, 비강자음 /ㄴ/은 제1음형대를 중심으로 주요변화가 일어남을 알 수 있었으며, 비폐색 부위에 따라 비음비율이나 제1음형대 감소의 정도가 다름을 알 수 있었다.

• Proceedings of the KAIS Fall Conference
• /
• 2006.05a
• /
• pp.364-367
• /
• 2006

음성신호를 관찰하였을 때 성문특성으로 인해서 고주파 쪽 특성이 약화되는 경향이 있다. 약화된 고주파 특성을 보상하기 위하여 프리 엠퍼시스 필터를 통해 보상하고 있다. 프리 엠퍼시스 필터를 간단한 수식으로 표현하면 y(n)=s(n)-As(n-1)와 같이 차분 방정식으로 나타낼 수 있다. 여기서 A값은 보통 0.9에서 1사이의 값을 사용한다. 본 논문에서는 QMF 필터를 이용하여 입력신호를 고주파와 저주파의 2개의 대역으로 분할하여 각 밴드에 프리 엠퍼시스 필터를 적용하여 약화되어진 특성을 정확히 보상하는 방법을 제안한다.

• Journal of the Institute of Electronics Engineers of Korea TE
• /
• v.39 no.3
• /
• pp.21-27
• /
• 2002

The digital watermarking methods have been proposed as a solution for solving the illegal copying and proof of ownership problems in the context of multimedia data. But it is still difficult to have been overcame the problem of the protection of property to multimedia data, such as digital images, digital video, and digital audio. This paper describes a watermarking algorithm that embeds non-linearly audio logo watermark data which is converted from audio signal of the ownership in the components of pixel intensities in an original image and that insists of ownership by hearing the audio signal transformed from the extracted audio logo through the speaker. Experimental results show that our algorithm using audio logo proposed in this paper is robust against attacks such as particularly lossy JPEG image compression. 

• Journal of Broadcast Engineering
• /
• v.22 no.5
• /
• pp.632-642
• /
• 2017

In this paper, we propose a DNN based speech detection system using acoustic characteristics and context information of media audio. The speech detection for discriminating between speech and non-speech included in the media audio is a necessary preprocessing technique for effective speech processing. However, since the media audio signal includes various types of sound sources, it has been difficult to achieve high performance with the conventional signal processing techniques. The proposed method improves the speech detection performance by separating the harmonic and percussive components of the media audio and constructing the DNN input vector reflecting the acoustic characteristics and context information of the media audio. In order to verify the performance of the proposed system, a data set for speech detection was made using more than 20 hours of drama, and an 8-hour Hollywood movie data set, which was publicly available, was further acquired and used for experiments. In the experiment, it is shown that the proposed system provides better performance than the conventional method through the cross validation for two data sets.

• Journal of Korea Multimedia Society
• /
• v.10 no.11
• /
• pp.1530-1537
• /
• 2007

In this paper, we implemented a speech emotion recognition system that can distinguish human emotional states from recorded speech captured by a single microphone and classify them into four categories: neutrality, happiness, sadness and anger. In general, a speech recorded with a microphone contains background noises due to the speaker environment and the microphone characteristic, which can result in serious system performance degradation. In order to minimize the effect of these noises and to improve the system performance, a MA(Moving Average) filter with a relatively simple structure and low computational complexity was adopted. Then a SFS(Sequential Forward Selection) feature optimization method was implemented to further improve and stabilize the system performance. For speech emotion classification, a SVM pattern classifier is used. The experimental results indicate the emotional classification performance around 65% in the computer simulation and 62% on the ARM platform.

• The Journal of the Acoustical Society of Korea
• /
• v.32 no.5
• /
• pp.438-445
• /
• 2013

In this paper, we propose an enhancement of Japanese acoustic model which is trained with Korean speech database by using several combination strategies. We describe the strategies for training more than two language combination, which are Cross-Language Transfer, Cross-Language Adaptation, and Data Pooling Approach. We simulated those strategies and found a proper method for our current Japanese database. Existing combination strategies are generally verified for under-resourced Language environments, but when the speech database is not fully under-resourced, those strategies have been confirmed inappropriate. We made tyied-list with only object-language on Data Pooling Approach training process. As the result, we found the ERR of the acoustic model to be 12.8 %.

• The Journal of the Acoustical Society of Korea
• /
• v.15 no.4
• /
• pp.83-92
• /
• 1996

This paper describes Lombard effect compensation and noise suppression so as to reduce speech recognition error in noisy environments. Lombard effect is represented by the variation of spectral envelope of energy normalized word and the variation of overall vocal intensity. The variation of spectral envelope can be compensated by linear transformation in cepstral domain. The variation of vocal intensity is canceled by spectral magnitude normalization. Spectral subtraction is use to suppress noise contamination, and band-pass filtering is used to emphasize dynamic features. To understand Lombard effect and verify the effectiveness of the proposed method, speech data are collected in simulated noisy environments. Recognition experiments were conducted with contamination by noise from automobile cabins, an exhibition hall, telephone booths in down town, crowded streets, and computer rooms. From the experiments, the effectiveness of the proposed method has been confirmed.

• Phonetics and Speech Sciences
• /
• v.11 no.3
• /
• pp.39-47
• /
• 2019

We propose a method to improve the performance of spontaneous speech recognizers by extending their phone set using speech data. In the proposed method, we first extract variable-length phoneme-level segments from broadcast speech signals, and convert them to fixed-length latent vectors using an long short-term memory (LSTM) classifier. We then cluster acoustically similar latent vectors and build a new phone set by choosing the number of clusters with the lowest Davies-Bouldin index. We also update the lexicon of the speech recognizer by choosing the pronunciation sequence of each word with the highest conditional probability. In order to analyze the acoustic characteristics of the new phone set, we visualize its spectral patterns and segment duration. Through speech recognition experiments using a larger training data set than our own previous work, we confirm that the new phone set yields better performance than the conventional phoneme-based and grapheme-based units in both spontaneous speech recognition and read speech recognition.