Next: Contents
Up: Corpus Specification
Previous: Speaker Profiles
Contents
Number of Speakers
The number of speakers is one of the most important characteristics of a spoken
language corpus. Speech corpora can be roughly divided into
the following three classes ([2], pp. 107 - 109):
- Speech corpora with 1 to 5 speakers are often used in the development of speech synthesis systems
or for basic research e.g. where invasive measurements must be made.
- Speech corpora with about 5 to 50 speakers are often used in experimental factorial research.
In general, the number of speakers and the
number of repetitions of the speech phenomena that are investigated should be
large enough for a meaningful statistical processing if factorial experimental
designs are planned.
- Speech corpora with more than 50 speakers are necessary to adequately train and test
speech recognition or speaker verification systems.
Note that a small number of speakers does not necessarily mean a small
corpus!
BITS Projekt-Account
2004-06-01