Call-for-Papers of Special Sessions

Special Session on Rich Information Annotation and Spoken Language Processing

Special Session on Robust Techniques for Organizing and Retrieving Spoken Documents

Special Session on Speaker Recognition

Special Panel Session on Multilingual Corpus Development 

Please submit your papers to the special session chairs directly.

1. Special Session on Rich Information Annotation and Spoken Language Processing

Chair: Dr Jianhua Tao, CAS, Beijing.

In the past several years, the research of Rich Information, such as paralinguistic information, etc., has become an interdisciplinary field, which covers speech synthesis, automatic speech recognition, speaker identification, spoken document retrieval, and natural language processing. The main goals of the research in the feild are the production, recognition or retrieval of spontaneous speech. As the different goals interact with each other, the synergistic use of disparate forms of analysis is critical. Focused on the fundamental research in human communication, Rich Information of speech is a great help for this research.

The purpose of this special session is to present the recent advances in the areas of Rich Information processing for speech, audio and spoken dialogue. We invite papers on, but not limited to, the following topics:

  • Rich Transcription Annotation on speech
  • Incorporation of prosodic and supralexical information
  • Speech Recognition based on Rich Information processing
  • Expressive Speech Synthesis
  • Natural Language Processing for Rich Transcription
  • Paradigms for Data Collection
  • Usability and human factors studies
  • Performance analysis and evaluation
  • Tools and solutions for rich transcription

2. Special Session on Robust Techniques for Organizing and Retrieving Spoken Documents

Chair: Dr Hsin-Min Wang, Academia Sinica, Taipei.

Multimedia data containing speech are considered “spoken documents”. As the cost of storage deceases and the bandwidth of communication increases, there has been a rapid growth of multimedia information on the Internet or in online archives. There is increasing interest in efficiently and effectively processing speech content for multimedia indexing and search. Therefore, organizing and retrieving spoken documents is becoming a research area of prime importance. This special session aims to solicit papers covering all aspects of the technologies and systems in the area. The topics cover, but are not limited to, the following:

  • Spoken document segmentation, clustering and transcription
  • Spoken document indexing and retrieval
  • Spoken document summarization and information extraction
  • Spoken document understanding, organization and visualization
  • Crosslingual and multilingual spoken document retrieval
  • Spoken dialogue systems for accessing spoken archives
  • Spoken document corpora
  • Performance evaluation metrics

3. Special Session on Speaker Recognition

Chair: Dr Thomas Fang Zheng, Tsinghua Univ., Beijing. Email:

Speaker recognition (or voiceprint recognition, VPR) is one of the most important branches in speech processing. Its applications become wider and wider in various fields, such as public security, anti-terrorism, justice, telephony banking, personal services, and so on. However, there are still many fundamental and theoretical problems to solve, such as issues of background noises, cross-channel, multiple-speakers, and short speech segment for training and testing.

The purpose of this special session is to invite researchers in this field to present their state-of-art technical achievements. Papers are invited to cover, but not limited to, the following topics:

  • Text-dependent and text-independent speaker identification
  • Text-dependent and text-independent speaker verification
  • Speaker detection
  • Speaker segmentation
  • Speaker tracking
  • Speaker recognition systems and application
  • Resource creation for speaker recognition

This special session also provides a platform for developers in this field to evaluate their speaker recognition systems using the same database provided by this special session. Evaluation of speaker recognition systems will cover the following tasks:

  • Text-independent speaker identification
  • Text-dependent and text-independent speaker verification
  • Text-independent cross-channel speaker identification
  • Text-dependent and text-independent cross-channel speaker verification

Final details on these tasks (including evaluation criteria) will be made available in due course. The development and testing data will be provided by the Chinese Corpus Consortium (CCC). The data sets will be extracted from two CCC databases, which are CCC-VPR3C2005 and CCC-VPR2C2005-1000. Participants are required to submit a full paper to the conference describing their algorithms, systems and results.

Schedule for this special session:

  • Feb. 01, 2006: On-line registration open
  • May. 01, 2006: Development data made available to participants
  • May. 21, 2006 : Test data made available to participants
  • Jun. 7, 2006 : Test results due at CCC
  • Jun. 10, 2006: Results released to participants
  • Jun. 25, 2006 (revised): Papers due (using ISCSLP standard format)
  • Jul. 25, 2006: The full set of the two databases made available to the participants of this special session upon request
  • Dec. 16, 2006: Conference presentation

Note: Due to time conflicts with ICASSP 2006, the time schedule is revised upon requests from participants.

This special session is organized by the CCC ( Please address your enquiries to Dr. Thomas Fang Zheng ( if any.

Download Speaker Recognition Evaluation Registration Form

4. Special Panel Session on Multilingual Corpus Development 

Chair: Dr. Chiu-yu Tseng, Academia Sinca, Taipei. Email:

The fast growing cross-linguistic technology development has made the call for multilingual corpus development more urgent than ever. Without multilingual corpora in text and speech forms, the calls from communities such as telecommunication, Internet users, dictionary compliers, text processing, etc. remain unanswered. In this special panel session, we would like to bring together those who have already pioneered in multilingual corpus development with concrete contributions to share with us their valuable experiences to a largely Chinese audience. We note also, with regret, that such efforts have been on languages that are more familiar to our community including Chinese, Japanese, English and Korean. Therefore, we would also like to invite those who have worked on multilingual corpus development from less familiar languages in our area such as Thai, Vietnamese, Malaysian or languages of India. Furthermore, those who are interested to use these resources or discuss establishing common benchmarks or platforms or facilitating possible collaborations across different geographical and linguistic communities are also encouraged to join the session.

The special session will be a panel discussion that aims at brainstorming and idea exchanging. Full papers are encouraged since the conference proceedings will be published by Springer, but not required. We invite presentations that cover but not limit to the following topics:

  • Multilingual dictionaries
  • Multilingual text collection
  • Multilingual speech corpus development
  • Cross-linguistic annotation design
  • Corpus development for less well known languages

Please send full papers or abstracts with presentation outlines to the chair.

©2005-2006 Chinese and Oriental Languages Information Processing Society, Singapore | Last updated on December 21, 2006 .