n South African Computer Journal - Automatic language identification : resisting complexity

Volume 2001, Issue 27
  • ISSN : 1015-7999
  • E-ISSN: 2313-7835



We introduce a simpler alternative to existing automatic language identification technology and quantitatively compare its performance to that of a state-of-the-art architecture on the recognised OGI Telephone Speech (OGI TS) multilingual corpus. The comparison is performed on ten two-language tasks spanning five languages. Although the system that we propose exhibits inferior performance (65% versus 81% on a 40s utterance), such an alternative may potentially provide adequate performance where cost and complexity of current technology is prohibitive. In addition, it may serve as an avenue of investigation for improving scalability of automatic language identification systems.

Loading full text...

Full text loading...


Article metrics loading...


This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error