SSCS 2010 - ACM Multimedia 2010 Workshop on Searching Spontaneous Conversational Speech (SSCS 2010)

The SSCS 2010 workshop is devoted to presentation and discussion of recent research results concerning advances and innovation in the area of spoken content retrieval and the area of multimedia search that makes use of automatic speech recognition technology.

Spoken audio is a valuable source of semantic information, and speech analysis techniques, such as speech recognition, hold high potential to improve information retrieval and multimedia search. Nonetheless, speech technology remains underexploited by multimedia systems, in particular, by those providing access to multimedia content containing spoken audio. Early success in the area of broadcast news retrieval has yet to be extended to application scenarios in which the spoken audio is unscripted, unplanned and highly variable with respect to speaker and style characteristics. The SSCS 2010 workshop is concerned with a wide variety of challenging spoken audio domains, including: lectures, meetings, interviews, debates, conversational broadcast (e.g., talkshows), podcasts, call center recordings, cultural heritage archives, social video on the Web and spoken natural language queries. As speech steadily moves closer to rivaling text as a medium for access and storage of information, the need for technologies that can effectively make use of spontaneous conversational speech to support search becomes more pressing.

In order to move the use of speech and spoken content in retrieval applications and multimedia systems beyond the current state of the art, sustained collaboration of researchers in the areas of speech recognition, audio processing, multimedia analysis and information retrieval is necessary. Motivated by the aim of providing a forum where these disciplines can engage in productive interaction and exchange, Searching Spontaneous Conversational Speech (SSCS) workshops have been held in conjunction with ACM SIGIR (SSCS 2007 and SSCS 2008) and ACM Multimedia (SSCS 2009). The SSCS workshop series continues at ACM Multimedia 2010 with a focus on research that strives to move retrieval systems beyond conventional queries and beyond the indexing techniques used in traditional mono-modal settings or text-based applications. We welcome contributions on a range of trans-disciplinary research issues related to these research challenges, including:

? Information Retrieval techniques in the speech domain
(e.g., applied to speech recognition lattices)
? Multimodal search techniques exploiting speech transcripts
(audio/speech/video fusion techniques including re-ranking)
? Search effectiveness
(e.g., evidence combination, query/document expansion)
? Exploitation of audio analysis
(e.g., speaker’s emotional state, speaker characteristics, speaking style)
? Integration of higher level semantics,
including topic segmentation and cross-modal concept detection
? Spoken natural language queries, Spoken Web
? Large-scale speech indexing approaches
(e.g., collection size, search speed)
? Multilingual settings (e.g., multilingual collections, cross-language access)
? Advanced interfaces for results display
and playback of multimedia with a speech track
? Exploiting user contributed information,
including tags, rating and user community structure
? Affordable, light-weight solutions for small collections, i.e., for the long tail

Contributions for oral presentations (short papers of 4 pages or long papers of 6 pages) and demonstration papers (4 pages) will be accepted. Please format your papers according to the submission guidelines of ACM Multimedia 2010 and upload them to the SSCS 2010 submission system. Submission deadline is June 10, 2010.