• The Journal of the Acoustical Society of Korea
• /
• v.23 no.2
• /
• pp.173-179
• /
• 2004

Due to the insufficiency of training data in large vocabulary continuous speech recognition, similar context dependent phones can be clustered by decision trees to share the data. When the decision trees are built and used to predict unseen triphones, a phonetic question set is required. The phonetic question set, which contains categories of the phones with similar co-articulation effects, is usually generated by phonetic or linguistic experts. This knowledge-based approach for generating phonetic question set, however, may reduce the homogeneity of the clusters. Moreover, the experts must adjust the question sets whenever the language or the PLU (phone-like unit) of a recognition system is changed. Therefore, we propose a data-driven method to automatically generate phonetic question set. Since the proposed method generates the phone categories using speech data distribution, it is not dependent on the language or the PLU, and may enhance the homogeneity of the clusters. In large vocabulary speech recognition experiments, the proposed algorithm has been found to reduce the error rate by 14.3%.

• Phonetics and Speech Sciences
• /
• v.1 no.1
• /
• pp.41-46
• /
• 2009

Duration of the preceding vowel is known to vary as a function of the (phonological or phonetic) voicing feature of the following consonant. This study raises a question against this general belief. A spectrographic experiment using 14 Korean obstruents (three sets of stops: /p, p', $p^h$/, /t, t', $t^h$/, /k, k', $k^h$/; one set of affricates: /c, c', $c^h$/; one set of fricatives: /s, s'/) reveals that (1) phonetic voicing in the intervocalic lax consonants /p, t, k, c, s/ has nothing to do with the duration of the preceding vowel; (2) vowel length is significantly shorter before tense consonants than before their lax cognates while tense consonants are significantly longer than their lax cognates. Importantly, Korean obstruents are all phonologically voiceless. Therefore, the voicing feature is rejected as the cause of preconsonantal vowel shortening in Korean both phonetically and phonologically. It is suggested that the temporal phenomenon is basically a kind of physiologically-motivated coarticulation though it is restricted by the phonology of a given language. To meet this assumption, the feature voicing should be replaced with the feature tenseness as the cause, which will enable us to explain the temporal phenomenon on the same basis irrespective of language.

• Lingua Humanitatis
• /
• v.5
• /
• pp.221-264
• /
• 2003

This paper takes issue with the idea of language as a 'serial-time structure' as opposed to the 'real-time event' of speech, an idea entrenched in Chomskyan model of linguistic theory. The discussion centers around the leitmotif question: Is language constructed entirely from a finite set of apriori discrete symbol types, as the 'competence vs performance' dichotomy implies\ulcorner A set of linguistic patterns examined in this study, largely with regard to phonological considerations, points to the evidence to the contrary. That is, while the patterns may be said to be linguistically distinct, they are not discretely, different, i.e. not different enough to be reliably differentiated. It is demonstrated that much of current research in phonology, including the most recent Optimality Theory, is misdirected in that it falsely presupposes a discrete universal phonetic inventory. The main thrust of the present study is that there is no sharp boundary between 'competence' defined as the formal, symbolic, discrete time domain of language and human cognition on the one hand and 'performance' as the continuous, fuzzy, real-time domain of human physiology on the other.

• Proceedings of the KSPS conference
• /
• 2003.10a
• /
• pp.79-82
• /
• 2003

The purpose of this study is to examine the regions of the cerebrum that handles the lexical and idiomatic ambiguity. The stimuli sets consist of two parts, and each part has 20 sets of sentences. For each part, 10 sets are experimental conditions and the other 10 sets are control conditions. Each set has two sentences, the 'context' and 'target' sentences, and a sentence-verification question for guaranteeing patients' concentration to the task. The results based on 15 patients showed that significant activation is present in the right frontal lobe of the cerebral cortex for both kinds of ambiguity. It means that right hemisphere participates in the resolution of ambiguity, and there are no regions specified for lexical ambiguity or idiomatic ambiguity alone.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.2
• /
• pp.199-210
• /
• 2002

In this paper, we carried out the study on speech recognition using the KM-Net topology design algorithm based on decision tree state-clustering to improve the performance of acoustic models in speech recognition. The Korean has many allophonic and grammatical rules compared to other languages, so we investigate the allophonic variations, which defined the Korean phonetics, and construct the phoneme question set for phonetic decision tree. The basic idea of the HM-Net topology design algorithm is that it has the basic structure of SSS (Successive State Splitting) algorithm and split again the states of the context-dependent acoustic models pre-constructed. That is, it have generated. the phonetic decision tree using the phoneme question sets each the state of models, and have iteratively trained the state sequence of the context-dependent acoustic models using the PDT-SSS (Phonetic Decision Tree-based SSS) algorithm. To verify the effectiveness of the above algorithm we carried out the speech recognition experiments for 452 words of center for Korean language Engineering (KLE452) and 200 sentences of air flight reservation task (YNU200). Experimental results show that the recognition accuracy has progressively improved according to the number of states variations after perform the splitting of states in the phoneme, word and continuous speech recognition experiments respectively. Through the experiments, we have got the average 71.5%, 99.2% of the phoneme, word recognition accuracy when the state number is 2,000, respectively and the average 91.6% of the continuous speech recognition accuracy when the state number is 800. Also we haute carried out the word recognition experiments using the HTK (HMM Too1kit) which is performed the state tying, compared to share the parameters of the HM-Net topology design algorithm. In word recognition experiments, the HM-Net topology design algorithm has an average of 4.0% higher recognition accuracy than the context-dependent acoustic models generated by the HTK implying the effectiveness of it.