Abstract for hain_icslp98

Proc. ICSLP'98, Sydney


T. Hain, P.C. Woodland


Broadcast news audio data contains a wide variety of different speakers and audio conditions (channel and background noise). This paper describes a segmentation, gender detection and audio classification scheme for such data which aims to provide a speech recogniser with a stream of reasonably-sized segments, each from a single speaker and audio type while discarding non-speech data. Each segment is labelled as either narrow or wide band and from either a female or male speaker. The segmentation system has been evaluated on the DARPA 1997 broadcast news data set and detailed segmentation accuracy results are presented. It is shown that the speech recognition accuracy for these automatically derived segments is very nearly the same as that for manually segmented data.

(ftp:) hain_icslp98.ps.gz (http:) hain_icslp98.ps.gz
PDF (automatically generated from original PostScript document - may be badly aliased on screen):
  (ftp:) hain_icslp98.pdf | (http:) hain_icslp98.pdf

If you have difficulty viewing files that end '.gz', which are gzip compressed, then you may be able to find tools to uncompress them at the gzip web site.

If you have difficulty viewing files that are in PostScript, (ending '.ps' or '.ps.gz'), then you may be able to find tools to view them at the gsview web site.

We have attempted to provide automatically generated PDF copies of documents for which only PostScript versions have previously been available. These are clearly marked in the database - due to the nature of the automatic conversion process, they are likely to be badly aliased when viewed at default resolution on screen by acroread.