Abstract | ||
---|---|---|
In large-scale deployments of speaker recognition systems the potential for legacy problems increases as the evolving technology may require conflguration changes in the system thus invalidating already existing user voice accounts. Un- less the entire database of original speech waveform were stored, users need to reenroll to keep their accounts func- tional, which, however, may be expensive and commercially not acceptable. We deflne model migration as a conversion of obsolete models to new-conflguration models without ad- ditional data and waveform requirements and investigate ways to achieve such a migration with minimum loss of sys- tem accuracy. As a proof-of-concept, an algorithm for sta- tistical migration in the Maximum A-Posteriori framework is studied and evaluated experimentally using the NIST SRE-03 dataset. The migration step is discussed in a wider conceptual framework of Conversational Biometrics. |
Year | Venue | Keywords |
---|---|---|
2004 | INTERSPEECH | statistical model,proof of concept,speaker recognition,conceptual framework |
Field | DocType | Citations |
Pattern recognition,Computer science,Speech recognition,Speaker recognition,Natural language processing,Artificial intelligence,Statistical model,Speaker diarisation | Conference | 2 |
PageRank | References | Authors |
0.44 | 5 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Jiri Navratil | 1 | 314 | 31.36 |
Ganesh N. Ramaswamy | 2 | 213 | 25.72 |
Ran D. Zilca | 3 | 59 | 6.87 |