3rd year projects 2002 - 03

  • JPB-UG-1: Web-based query and retrieval of audio data (Katan Patel)

Mail all

The project descriptions below are only intended as starting points. If you wish to discuss possibilities in greater detail I encourage you to email me to arrange a meeting.


JPB-UG-1: Web-based query and retrieval of audio data

Description

Many corpora of spoken language exist. Such corpora typically contain audio recordings of speech material (sentences, passages, words, smaller units). They usually contain transcriptions too, although the level of detail varies from corpus to corpus. What is needed is a uniform access procedure to allow researchers to extract just those portions which meet certain criteria. For example, the ShATR corpus [1] contains details of conversations recorded by eight microphones simultaneously. A user might issue a query such as: find all places where at least two talkers are speaking simultaneously. The aim of this project is to develop a web-based interface which allows users to formulate and refine a range of queries and to receive the requested data in a uniform format. To start, a simple corpus containing transcribed speech from a single talker will be used. Later, more demanding corpora such as ShATR should be handled.

Resources

  • You will be provided with a number of corpora.

Reading

  • Further information on the ShATR multi-speaker corpus can be found in Karlsen et al., available from PDG

Useful prior knowledge

  • Experience with web technology e.g. cgi scripts, perl etc would be useful

  • COM3130, COM3140, new course on text processing

[TOP]