• Proceedings of the KSPS conference
• /
• 2003.05a
• /
• pp.115-118
• /
• 2003

Channel distortion and coarticulation effect in the connected digit telephone speech make it difficult to recognize, and degrade recognition performance in the telephone environment. In this paper, as a basic research to improve the recognition performance of Korean connected digit telephone, error patterns are investigated and analyzed. Telephone digit speech database released by SITEC with HTK system is used for recognition experiments. Both DWFBA and MRTCN methods are used for feature extraction and channel compensation, respectively. Experimental results are discussed with our findings.

• MALSORI
• /
• no.35_36
• /
• pp.189-209
• /
• 1998

This study discusses Korean speech database that has been designed and constructed for common use, especially focusing on designing a list of words or sentences that covers various phonological environments. As the results, PBW(Phonetically Balanced words) and PBS(Phonetically Balanced Sentences) was selected from balanced text corpus using maximum entropy method. And, implemented workbench for spoken language data acquisition is presented in this paper. The workbench consists of grapheme to phoneme converter, utterance list selection module, speech data editing module, multi-layer labelling module, and phoneme context search module.

• MALSORI
• /
• no.54
• /
• pp.63-72
• /
• 2005

The performance of speech recognition system is degraded by the mismatch between training and test environments. Many studies have been presented to compensate for noise components in the cepstral domain. Recently, higher order cepstral moment normalization method has been introduced to improve recognition accuracy. In this paper, we present real-time high order moment normalization method with post-processing smoothing filter to reduce the parameter estimation error in higher order moment computation. In experiments using Aurora2 database, we obtained error rate reduction of 44.7% with proposed algorithm in comparison with baseline system.

• Proceedings of the KSPS conference
• /
• 2005.11a
• /
• pp.17-20
• /
• 2005

The purpose of this project is to develop a device that can automatically measure pronunciation of English speech produced by Korean learners of English. Pronunciation proficiency will be measured largely in two areas; suprasegmental and segmental areas. In suprasegmental area, intonation and word stress will be traced and compared with those of native speakers by way of statistical methods using tilt parameters. Durations of phones are also examined to measure speakers' naturalness of their pronunciations. In doing so, statistical duration modelling from a large speech database using CART will be considered. For segmental measurement of pronunciation, acoustic probability of a phone, which is a byproduct when doing the forced alignment, will be a basis of scoring pronunciation accuracy of a phone. The final score will be a feedback to the learners to improve their pronunciation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.19 no.5
• /
• pp.1243-1250
• /
• 2015

This paper presents an effective recognition method of VPI patient's speech for a VPI speech reconstruction system. Speaker adaptation technique is employed to improve VPI speech recognition. This paper proposes to use simulated speech for generating an initial model for speaker adaptation, in order to effectively utilize the small size of VPI speech for model adaptation. We obtain 83.60% in average word accuracy by applying MLLR for speaker adaptation. The proposed speaker adaptation method using simulated speech model brings 6.38% improvement in average accuracy. The experimental results demonstrate that the proposed speaker adaptation method is highly effective for developing recognition system of VPI speech which is not suitable for constructing large-size speech database.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.18 no.7
• /
• pp.1740-1748
• /
• 2014

This paper focuses on listening test and acoustic analysis of patients' speech with velopharyngeal insufficiency (VPI) and normal speakers' simulation speech. In this research, a set consisting of 50-words, vowels and single syllables is determined for speech database construction. A web-based listening evaluation system is developed for a convenient/automated evaluation procedure. The analysis results show the trend of incorrect recognition for VPI speech and the one for simulation speech are similar. Such similarity is also confirmed by comparing the formant locations of vowel and spectrum of consonant sounds. These results show that the simulation method for VPI speech is effective at generating the speech signals similar to actual VPI patient's speech. It is expected that the simulation speech data can be effectively employed for our future work such as acoustic model adaptation.

• The Journal of the Acoustical Society of Korea
• /
• v.18 no.3
• /
• pp.17-26
• /
• 1999

We present the first speech coarticulation database of Korean, English and Konglish/sup 3)/ named "SORIDA"/sup 4)/, which is designed to cover the maximum number of representations of coarticulation in these languages [1]. SORIDA features a compact database which is designed to contain a maximum number of triphones in a minimum number of prompts. SORIDA contains all consonantal triphones and vowel allophones in 682 Korean prompts of word length and in 717 English prompt words, spoken five times by speakers of balanced genders, dialects and ages. Korean prompts are synthesized lexicons which maximize their coarticulation variation disregarding any stress phenomena, while English prompts are natural words that fully reflect their stress effects with respect to the coarticulation variation. The prompts are designed differently because English phonology has stress while Korean does not. An intermediate language, Konglish has also been modeled by two Korean speakers reading 717 English prompt words. Recording was done in a controlled laboratory environment with an AKG Model C-100 microphone and a Fostex D-5 digital-audio-tape (DAT) recorder. The total recording time lasted four hours. SORIDA CD-ROM is available in one disk of 22.05 kHz sampling rate with a 16 bit sample size. SORIDA digital audio-tapes are available in four 124-minute-tapes of 48 kHz sampling rate. SORIDA′s list of phonetically-rich-words is also available in English and Korean.

• MALSORI
• /
• no.44
• /
• pp.73-82
• /
• 2002

Channel distortion degrades the performance of speech recognizer in telephone environment. It mainly results from the bandwidth limitation and variation of transmission channel. Variation of channel characteristics is usually represented as baseline shift in the cepstrum domain. Thus undesirable effect of the channel variation can be removed by subtracting the mean from the cepstrum. In this paper, to improve the recognition performance of Korea connected digit telephone speech, channel compensation methods such as CMN (Cepstral Mean Normalization), RTCN (Real Time Cepatral Normalization), MCMN (Modified CMN) and MRTCN (Modified RTCN) are applied to the static MFCC. Both MCMN and MRTCN are obtained from the CMN and RTCN, respectively, using variance normalization in the cepstrum domain. Using HTK v3.1 system, recognition experiments are performed for Korean connected digit telephone speech database released by SITEC (Speech Information Technology & Industry Promotion Center). Experiments have shown that MRTCN gives the best result with recognition rate of 90.11% for connected digit. This corresponds to the performance improvement over MFCC alone by 1.72%, i.e, error reduction rate of 14.82%.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.4
• /
• pp.315-322
• /
• 2003

We propose a channel-attentive Mel frequency cepstral coefficient (CAMFCC) extraction method to improve the recognition performance of speech that is partially corrupted in the frequency domain. This method introduces weighting terms both at the filter bank analysis step and at the output probability calculation of decoding step. The weights are obtained for each frequency channel of filter bank such that the more reliable channel is emphasized by a higher weight value. Experimental results on TIDIGITS database corrupted by various frequency-selective noises indicated that the proposed CAMFCC method utilizes the uncorrupted speech information well, improving the recognition performance by 11.2% on average in comparison to a multi-band speech recognition system.

• ETRI Journal
• /
• v.44 no.3
• /
• pp.476-490
• /
• 2022

With the recent advances in technology, automatic speech recognition (ASR) has been widely used in real-world applications. The efficiency of converting large amounts of speech into text accurately with limited resources has become more vital than ever. In this study, we propose a method to rapidly recognize a large speech database via a transformer-based end-to-end model. Transformers have improved the state-of-the-art performance in many fields. However, they are not easy to use for long sequences. In this study, various techniques to accelerate the recognition of real-world speeches are proposed and tested, including decoding via multiple-utterance-batched beam search, detecting end of speech based on a connectionist temporal classification (CTC), restricting the CTC-prefix score, and splitting long speeches into short segments. Experiments are conducted with the Librispeech dataset and the real-world Korean ASR tasks to verify the proposed methods. From the experiments, the proposed system can convert 8 h of speeches spoken at real-world meetings into text in less than 3 min with a 10.73% character error rate, which is 27.1% relatively lower than that of conventional systems.