劉怡芬

助理教授

啟發我自身對跨語言與計算領域的學習契機,主要是在大學畢業之初,所參與的一項計畫:現代漢語自然口語語料庫(TMC)的收集與建置。該計畫的規模與架構均與當時其他歐(Dutch Corpus)、美(Switchboard)、日(CSJ)大型語音工程的計畫相當,包含不同語境設定的主題對話,以及語音工程任務所需的標記。之後,於碩士、博士兩個不同階段歷程,接受語言本質理論的知識與工程演算模型的實作訓練。立於這樣的基石上,著手語音工程相關的研究工作,並著眼於基礎語音訊號處理、統計模型的應用與改良。將以語言知識規則為基礎所建構的發音變異模型,應用在提升中文連續口語語音辨識效能上,其研究成果獲得語音工程領域重要期刊(The Journal of the Acoustical Society of America)的接受。爾後,延續對自然語言處理與語意理解這方面的興趣,任職於雲丰語言科技期間,將所學的知識與技術應用在中文自然口語能力評估系統的核心運算開發上。目前研究領域包含:運用語音工程技術於醫療的構音異常檢測/評估系統、華語學習者的聲調模型,以及如何運用語音訊號幫助識別兩個語意事件之間的言談關係。


學歷

  • 清華大學 資訊系統與應用學系 博士
  • 交通大學 語言與文化系 碩士
  • 實踐大學 應用外文系 學士

專長

  • 機率論
    Introduction to Probability
  • 機器學習
    Machine Learning
  • 語音辨識
    Speech Recognition
  • 自然口語處理與辨識
    Spoken Language Processing and Recognition
  • 計算語言學
    Computational Linguistics

校內經歷

  • 資訊工程學系 助理教授

校外經歷

  • 雲丰科技 語言技術研發 研發工程師
  • 中央研究院 語言所 研究助理
  • 中央研究院 語言所 研究助理

論文及參與計畫

  1. Yi-Fen Liu, Shu-Chuan Tseng and Jyh-Shing Roger Jang, "Deriving disyllabic word variants from a Chinese conversational speech corpus," Journal of the Acoustical Society of America, 140(1): 308-321, 2016-07. (SCIE)
  1. Shu-Chuan Tseng, Yi-Fen Liu and Xiang-Li Lu,"Model-assisted Lexical Tone Evaluation of three-year-old Chinese-speaking Children by also Considering Segment Production ," 2023 INTERSPEECH, 3909-3913, 2023-08. Convention Centre .
  2. Yi-Fen Liu and Xiang-Li Lu,"J-ToneNet: A Transformer-based Encoding Network for Improving Tone Classification in Continuous Speech via F0 Sequences ," 2023 INTERSPEECH, 2138-2142, 2023-08. Convention Centre .
  3. Shu-Chuan Tseng and Yi-Fen Liu,"Segment and Tone Production in Continuous Speech of Hearing and Hearing-impaired Children ," 2021 INTERSPEECH, 606-610, 2021-08. Hotel International Brno .
  4. Liu, Yi-Fen and Shu-Chuan Tseng,"Use of words in story-telling data of Chinese speaking hearing and hearing-impaired children ," The workshop on Disfluencies in Spontaneous Speech with an additional Special Day on (Dis)Fluency in Children’s Speech (DISS 2019), 91-92, 2019-09. ELTE Eotvos Lorand University .
  5. Tseng, Shu-Chuan and Yi-Fen Liu,"Establishment of normative phonological development data for Mandarin-speaking children by applying computational linguistics techniques ," The annual meeting of the Speech-Language-Hearing Association of Taiwan, 6, 2017-12. 高雄醫學大學 .
  6. Tseng, Shu-Chuan, Yi-Fen Liu, and Ye-Sheng Victor Lin,"A Chinese spoken language assessment system that utilizes computational linguistics techniques ," The 10th Asia Pacific Conference on Speech, Language and Hearing, 216, 2017-09. International University of Health and Welfare, Narita .
  7. Liu, Yi-Fen, Shu-Chuan Tseng, and Jyh-Shing Roger Jang,"Phone boundary annotation in conversational speech ," The International Conference on Language Resources and Evaluation, 848-853, 2014-05. The Harpa Conference Centre .
  8. Liu, Yi-Fen, Shu-Chuan Tseng, Jyh-Shing Roger Jang, and Cheng-Hsien Alvin Chen,"Coping Imbalanced Prosodic Unit Boundary Detection with Linguistically-motivated Prosodic Features ," INTERSPEECH, 1417-1420, 2010-09. International Conference Hall Makuhari Messe International Convention Complex .
  9. Liu, Yi-Fen, Shu-Chuan Tseng, Ya-Fang He, Yun-Ru Huang and Tze-Lun Lee,"A preliminary study for spontaneous speech understanding and processing: prosodic units in Mandarin Chinese ," The International Symposium on Linguistic Patterns in Spontaneous Speech (LPSS 2006), 97-112, 2006-11. Academia Sinica .
  1. 以Transformers關注基頻(F0)並交叉參照音節節奏韻律表徵以強化聲調調型表徵之學習/2024-08~2025-07 /NSTC113-2221-E-035-060- /主持人
  2. 保衛聲優專業:防範與檢測 Deepfake 音訊之研究/2024-07~2025-02 /113-2813-C-035-034-E /主持人
  3. 基於 Transformers 以及BERT輸入架構建立能學習聲調調型和口說韻律節奏的中文音節聲調辨認模型/2023-08~2024-07 /NSTC112-2221-E-035-067- /主持人
  4. 整合文字情緒分析與語音合成相關技術以提高故事朗讀系統的表達自然度/2023-07~2024-02 /112-2813-C-035-019-E /主持人
  5. LINE Bot 結合語音技術應用於慢性用藥輔助系統之開發與設計—以高血壓患者為例/2023-07~2024-02 /112-2813-C-035-020-E /主持人
  6. 以卷積自動編碼器和群聚分析為本的字詞聲調變化以及語句口說韻律節奏配置之研究(2/2)/2022-04~2023-03 /MOST110-2222-E-035-005-MY2 /主持人
  7. 以卷積自動編碼器和群聚分析為本的字詞聲調變化以及語句口說韻律節奏配置之研究(1/2)/2021-04~2022-03 /MOST110-2222-E-035-005-MY2 /主持人
  8. 應用文字分析技術於設計聊天輔助機器人以降低社交焦慮/2019-07~2020-02 /108-2813-C-035-006-E /主持人
  1. 「農業知識文章語料關係提取」計畫 /2021-04~2021-12 /共同主持人
  2. (國合)Design and Development of NX SAAS Application Cloud Services NX SAAS 應用雲服務的設計與開發III /2020-04~2021-03 /協同主持人
  1. Yi-Fen Liu and Shu-Chuan Tseng/Word use and word-level reduction in storytelling speech of Chinese-speaking hearing and hard of hearing children/JUDIT BÓNA /Akadémiai Kiadó /2021-07-01/ DOI: 10.1556/9789634547099
  2. Liu, Yi-Fen, and Shu-Chuan Tseng/Linguistic Patterns Detected through a Prosodic Segmentation in Spontaneous Taiwan Mandarin Speech/Linguistic Patterns in Spontaneous Speech /Institute of Linguistics, Academia Sinica /2009-01-01/ 147-166
最後更新時間:2024-10-5, 9:04 a.m. 下次更新時間:2024-10-6, 9 a.m.