Spoken document retrieval for the languages of Hong Kong

The advent of the information age has brought massive digital libraries of multimedia and multilingual content. This creates a high demand for multimedia and multilingual indexing and retrieval technologies, e.g. those applicable to audio archives. The paper reports on the development of spoken document retrieval (SDR) technologies, where speech recognition is combined with information retrieval for searching audio information in Cantonese, Mandarin and English: the languages of Hong Kong. We tackle the SDR problem in a monolingual setting, where the textual queries and the audio documents are expressed in the same language. The authors describe their system, the interface design for audio information visualization, as well as their experiments on spoken document retrieval.