未来の音声コーパス構築において,複雑な信号処理,統計的分類やエキスパート知識を組み合わせた自動的手法によって,多段階の言語的情報が付与されることが期待される。ラベル付与ツールはさまざまな言語や発話様式を幅広く適応し,または世界中の音声研究者や技術者に容易にアクセス可能なことが重要である。本論文は,調音に基づいた特徴,音素区切り,音節,単語,韻律パターンに適切なフレーズ単位など,幅広く表現した言語的要素を抽出可能な自動ラベル付与ツールの開発に焦点を当てた世界中の今までの研究を記述する。最終的なゴールは,いかなる言語においても,発話様式,声質や言語的文脈などを幅広く精密で自動的に特徴づけることである。

抄録全体を表示

In an attempt to construct a large-scale database of spontaneous speech, the authors planned to give segmental and prosodic labels to spontaneous Japanese speech. This paper reports the method of this labeling and its performance. First, the performance of automatic segmental labeling by Hidden Markov Model was verified. Sample speech of about four hour long was automatically phoneme labeled and compared to the results of hand-labeling. It turned out that average of label boundary difference with hand labeled data was 14.3[ms]. Second, the performance of prosodic labeling by newly proposed labeling scheme named X-JToBI (eXtended J_ToBI) was verified. The analysis of labeled data showed that newly added inventories appeared in the data of spontaneous speech and rate of inter-labeler agreement increased in nearly all types of labels.

抄録全体を表示

音声コミュニケーションに於いて,声質は,発話者の感情や個人情報など音声の非言語情報の伝達に重要な役割を果たしている。特に,歌声に於いては,声質は多様な歌唱法を規定する重要な特徴の一つとなっている。本稿では,歌声に於ける声質の生成機構について,生理学的観察,物理モデル,信号モデルの視点から論じる。先ず,声帯振動様式によって特徴付けられる声区について概観し,次に,アジアの歌唱法に特徴的に見られる喉詰型発声に関して,声帯振動,喉頭上部の狭窄,声門上構造物の振動の,声質への影響について考察を行なう。

抄録全体を表示

本論文では,最初に,音声コミュニケーションの声質制御について,音声学および音声工学におけるこれまでの研究を紹介する。さまざまな声質を表現する声帯音源波形の定量的モデルの構築を目指し,Laverが提案した声質の記述的な枠組(1980)を振り返り,発声様式の音響分析に関する研究のこれまでの発展について概観する。次に,これら従来の研究を踏まえ,声門体積速度微分波形のモデル化に関する全体論的なアプローチを提案する。幾つかの発声様式を表現するLaverの音声データ(1980)を用い,それぞれの発声の声帯音源波形の1周期について,主成分分析を応用した予備的評価を報告する。最上の4主成分は全データの分散の9割を占め,定性的にも解釈可能である。これら抽出された主成分を,分析/変換/合成の枠組みにおける制御パラメータとして用い,発話データの声質変換を試みた。

抄録全体を表示

Medical Ultrasonography (US) can be useful for analysing movements inside the mouth during speech. CT and MRI scans may provide more detailed information of the anatomical gestures, but the lightness, speed, portability, and safety of US offer advantages to the phonetician. This paper describes the techniques and data of US scans with respect to phonetics research, and describes the gestures that are associated with the phonemes of Japanese speech. In addition, we propose the use of image-processing techniques for deriving a laryngeal elevation index (LEI) from the US data, to provide a contour which can be used to reveal prosodic information.

抄録全体を表示

The aim of this paper is to investigate the occurrence of vowel devoicing in both Korean and Japanese spoken by Korean learners of Japanese. Close vowels between voiceless consonants in Korean and Japanese were analyzed acoustically. The remarkable character of Korean vowel devoicing is that the occurrence rate of devoicing differs considerably among individual subjects, irrespective of dialect. As for Japanese spoken by Korean learners, there is a high correlation between occurrence rates of devoicing in Korean and in Japanese.

抄録全体を表示

This paper proposes a method of comprehensively and structurally characterizing the segmental aspect of speech. After building phoneme HMMs from speech samples, distances between any two states of any phoneme HMMs were calculated to make a state-based distance matrix. With Ward's method of hierarchical clustering, a tree diagram of the distance matrix was generated to visualize comprehensive structure embedded in the pronunciation. Using American English (AE) and Japanese English (JE) speech samples, two kinds of trees were drawn. Comparison between the two trees clearly showed the well-known Japanese habits in speaking English. Using the distance matrix of JE, the compatibility between phonetic structure in the pronunciation of JE and lexical structure of the entire vocabulary of AE was estimated based upon Cohort model of isolated word perception.

抄録全体を表示

This paper is concerned with the tone of loanwords in Kagoshima dialect. Kagoshima dialect has a two-pattern tone system. Type-A is a falling type tone, and Type-B is a rising type tone. Most of loanwords are pronounced Type-A in Kagoshima, but for the young people, the number of Type-B loanwords has been increasing. This suggests that the young people in Kagoshima are affected by the Tokyo accent which contains many level tone loanwords.

抄録全体を表示

Metalinguistic Awareness of syllabic morae is the ability to recognize a syllabic mora (a nasal coda, a geminate stop consonant, and a long vowel in this study) as one mora. It has been said to play a significant role for the acquisition of Japanese Kana letters, especially for syllabic morae. However the process of the acquisition is yet to be made clear. The goal of this study is to therefore consider how such metalinguistic awareness of syllabic morae is developed in young Japanese children, through the comparison between children living in Yokohama where morae are counted and Fukaura where morae are not. Two small studies were conducted to examine whether they use mora or syllable in segmenting words with special morae. The results show that the children living in Fukaura recognize syllabic morae less than those in Yokohama, especially for long vowels and nonsense words. It suggests that input is significant for young children to develop their metalinguistic awareness of syllabic morae.

抄録全体を表示