• Phonetics and Speech Sciences
• /
• v.10 no.4
• /
• pp.111-119
• /
• 2018

This paper proposes a method to improve speech recognition performance by extracting various pronunciations of the pseudo-morpheme unit from an eojeol unit corpus and generating a new recognition unit considering pronunciation variations. In the proposed method, we first align the pronunciation of the eojeol units and the pseudo-morpheme units, and then expand the pronunciation dictionary by extracting the new pronunciations of the pseudo-morpheme units at the pronunciation of the eojeol units. Then, we propose a new recognition unit that relies on pronunciation by tagging the obtained phoneme symbols according to the pseudo-morpheme units. The proposed units and their extended pronunciations are incorporated into the lexicon and language model of the speech recognizer. Experiments for performance evaluation are performed using the Korean speech recognizer with a trigram language model obtained by a 100 million pseudo-morpheme corpus and an acoustic model trained by a multi-genre broadcast speech data of 445 hours. The proposed method is shown to reduce the word error rate relatively by 13.8% in the news-genre evaluation data and by 4.5% in the total evaluation data.

• KIPS Transactions on Software and Data Engineering
• /
• v.1 no.2
• /
• pp.91-102
• /
• 2012

Computer aided language learning has become popular. But the level of automation of constructing a Korean learning assistant system is not so high because a practical language learning system needs large scale knowledge resources, which is very hard to acquire. In this paper, we propose a Korean learning assistant system that utilizes easily obtainable knowledge resources like a corpus, web documents and a lexicon. Our system has three modules - problem solving, pronunciation marker and writing assistant. Automatic problem generator uses a corpus and a lexicon to make problems with one correct answer and three distracters, then verifies their suitability by utilizing frequency information from web documents. We analyze pronunciation rules for a pronunciation marker and recommend appropriate words and sentences in real-time by using data extracted from a corpus. In experiment, we evaluate 400 automatically generated problems, which show 89.9% problem suitability and 64.9% example suitability.

• ETRI Journal
• /
• v.35 no.1
• /
• pp.89-99
• /
• 2013

The problem of the sentence-based pronunciation evaluation task is defined in the context of subjective criteria. Three subjective criteria (that is, the minimum subjective word score, the mean subjective word score, and first impression) are proposed and modeled with the combination of word-based assessment. Then, the subjective criteria are approximated with objective sentence pronunciation scores obtained with the combination of word-based metrics. No a priori studies of common mistakes are required, and class-based language models are used to incorporate incorrect and correct pronunciations. Incorrect pronunciations are automatically incorporated by making use of a competitive lexicon and the phonetic rules of students' mother and target languages. This procedure is applicable to any second language learning context, and subjective-objective sentence score correlations greater than or equal to 0.5 can be achieved when the proposed sentence-based pronunciation criteria are approximated with combinations of word-based scores. Finally, the subjective-objective sentence score correlations reported here are very comparable with those published elsewhere resulting from methods that require a priori studies of pronunciation errors.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.6
• /
• pp.434-442
• /
• 2003

Software internationalization, the process of making software easier to localize for specific languages, has deep implications when applied to speech technology, where the goal of the task lies in the very essence of the particular language. A greatdeal of work and fine-tuning has gone into language processing software based on ASCII or a single language, say English, thus making a port to different languages difficult. The inherent identity of a language manifests itself in its lexicon, where its character set, phoneme set, pronunciation rules are revealed. We propose a decomposition of the lexicon building process, into four discrete and sequential steps. For preprocessing to build a lexical model, we translate from specific language code to unicode. (step 1) Transliterating code points from Unicode. (step 2) Phonetically standardizing rules. (step 3) Implementing grapheme to phoneme rules. (step 4) Implementing phonological processes.

• MALSORI
• /
• no.57
• /
• pp.123-137
• /
• 2006

This paper aims to propose a method of detecting irregular pronunciations for Korean POI data adopting the notion of the Prosodic Word based on the Prosodic Phonology (Selkirk 1984, Nespor and Vogel 1986) and Intonational Phonology (Jun 1996). In order to show the performance of the proposed method, the detection experiment was conducted on the 250,000 POI data. When all the data were trained, 99.99% of the exceptional prosodic words were detected, which shows the stability of the system. The results show that similar ratio of exceptional prosodic words (22.4% on average) were detected on each stage where a certain amount of the training data were added. Being intended to be an example of an interdisciplinary study of linguistics and computer science, this study will, on the one hand, provide an understanding of Korean language from the phonological point of view, and, on the other hand, enable a systematic development of a multiple pronunciation lexicon for Korean TTS or ASR systems of high performance.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.2
• /
• pp.31-35
• /
• 1997

In this paper, we propose a variable vocabulary word recognizer that is able to recognize new words not exist in training data. For the variable vocabulary word recognizer, we must have an on-line lexicon generator to transform new candidate words to the corresponding pronunciation sequences of phones without any large lexicon table. And, we also must make outputs. In order to model the phones and allophones reliably, we define Korean allophones by triphone clustering based on phonetic knowledge of preceding and succeeding phones of each phone. Using the clustering method, we generated 1,548 allophones with POW (Phonetically Optimized Words) 3,848 word DB. We evaluated the proposed word recognizer with POW 3,848 DB, PBW (Phonetically Balanced Words) 445 DB, and 244 word DB in hotel reservation task. Experimental results showed word recognition accuracy of 79.6% for the POW DB corresponding to vocabulary-dependent case, 79.4% in case of 445 word lexicon and 88.9% in case of 100 word lexicon for the PBW DB, and 71.4% for the hotel reservation DB corresponding to vocabulary-independent case.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.5
• /
• pp.530-536
• /
• 2021

Recently, due to the development of deep learning, end-to-end speech recognition, which directly maps graphemes to speech signals, shows good performance. Especially, among the end-to-end models, conformer shows the best performance. However end-to-end models only focuses on the probability of which grapheme will appear at the time. The decoding process uses a greedy search or beam search. This decoding method is easily affected by the final probability output by the model. In addition, the end-to-end models cannot use external pronunciation and language information due to structual problem. Therefore, in this paper conformer with lexicon transducer is proposed. We compare phoneme-based model with lexicon transducer and grapheme-based model with beam search. Test set is consist of words that do not appear in training data. The grapheme-based conformer with beam search shows 3.8 % of CER. The phoneme-based conformer with lexicon transducer shows 3.4 % of CER.

• Proceedings of the KSPS conference
• /
• 2006.11a
• /
• pp.80-83
• /
• 2006

The purpose of this paper is to build a pronunciation lexicon with estimated likelihoods of the phonetic rules based on the phonetic realizations and therefore to improve the performance of CSR using the dictionary. In the baseline system, the phonetic rules and their application probabilities are defined with the knowledge of Korean phonology and experimental tuning. The advantage of this approach is to implement the phonetic rules easily and to get stable results on general domains. However, a possible drawback of this method is that it is hard to reflect characteristics of the phonetic realizations on a specific domain. In order to make the system reflect phonetic realizations, the likelihood of phonetic rules is reestimated based on the statistics of the realized phonemes using a forced-alignment method. In our experiment, we generates new lexica which include pronunciation variants created by reestimated phonetic rules and its performance is tested with 12 Gaussian mixture HMMs and back-off bigrams. The proposed method reduced the WER by 0.42%.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.1E
• /
• pp.3-11
• /
• 2002

Broadcast news transcription is one of the hardest tasks in speech recognition because broadcast speech signals have much variability in speech quality, channel and background conditions. We developed a Korean broadcast news speech recognizer. We used a morpheme-based dictionary and a language model to reduce the out-of·vocabulary (OOV) rate. We concatenated the original morpheme pairs of short length or high frequency in order to reduce insertion and deletion errors due to short morphemes. We used a lexicon with multiple pronunciations to reflect inter-morpheme pronunciation variations without severe modification of the search tree. By using the merged morpheme as recognition units, we achieved the OOV rate of 1.7% comparable to European languages with 64k vocabulary. We implemented a hidden Markov model-based recognizer with vocal tract length normalization and online speaker adaptation by maximum likelihood linear regression. Experimental results showed that the recognizer yielded 21.8% morpheme error rate for anchor speech and 31.6% for mostly noisy reporter speech.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2005.05a
• /
• pp.233-238
• /
• 2005

본 논문에서는 학습을 이용한 발음 변이 모델링을 통해 특정 영역에 최적화된 발음 사전 자동 생성의 방법을 제시하였다. 학습 방법을 이용한 발음 변이 모델링의 오류를 최소화 하기 위하여 본 논문에서는 발음 변이 규칙의 적응 기법을 도입하였다. 발음 변이 규칙의 적응은 대용량 음성 말뭉치에서 발음 변이 규칙을 유도한 후, 상대적으로 작은 용량의 음성 말뭉치에서 유도한 규칙과의 결합을 통해 이루어 진다. 본 논문에서 사용된 발음 사전은 해당 형태소의 앞 뒤 음소 문맥의 음운 현상을 반영한 발음 사전이며, 학습 방법으로 얻어진 발음 변이 규칙을 대용량 문자 말뭉치에 적용하여 해당 형태소의 발음을 자동 생성하였다. 발음 사전의 평균 발음의 수는 적용된 발음 변이 규칙의 확률 값들의 한계 값 조정에 의해 이루어졌다. 기존의 지식 기반의 발음 사전과 비교 할 때, 본 방법론으로 작성된 발음 사전을 이용한 대화체 음성 인식 실험에서 0.8%의 단어 오류율(WER)이 감소하였다. 또한 사전에 포함된 형태소의 평균 발음 변이 수에서도 기존의 방법론에서 보다 5.6% 적은 수에서 최상의 성능을 보였다.