A BAYESIAN APPROACH TO INTER-TASK FUSION FOR SPEAKER RECOGNITION
| Type of publication: | Conference paper |
| Citation: | Madikeri_ICASSP2019_2019 |
| Publication status: | Published |
| Booktitle: | In Proceedings of ICASSP 2019 |
| Year: | 2019 |
| Month: | May |
| Pages: | 5786-5790 |
| Location: | Brighton, ENGLAND |
| ISSN: | 1520-6149 |
| ISBN: | 978-1-4799-8131-1 |
| Crossref: | Madikeri_Idiap-RR-07-2020: |
| Abstract: | In i-vector based speaker recognition systems, back-end classifiers are trained to factor out nuisance information and retain only the speaker identity. As a result, variabilities arising due to gender, language and accent ( among many others) are suppressed. Inter-task fusion, in which such metadata information obtained from automatic systems is used, has been shown to improve speaker recognition performance. In this paper, we explore a Bayesian approach towards inter-task fusion. Speaker similarity score for a test recording is obtained by marginalizing the posterior probability of a speaker. Gender and language probabilities for the test audio are combined with speaker posteriors to obtain a final speaker score. The proposed approach is demonstrated for speaker verification and speaker identification tasks on the NIST SRE 2008 dataset. Relative improvements of up to 10% and 8% are obtained when fusing gender and language information, respectively. |
| Keywords: | bayesian fusion, inter-task fusion, speaker recognition |
| Projects |
SIIP |
| Authors | |
| Added by: | [UNK] |
| Total mark: | 0 |
|
Attachments
|
|
|
|
|
Notes
|
|
|
|
|
