Abstract | ||
---|---|---|
This paper presents an automatic speaker recognition system for in- telligence applications. The system has to provide functionalities for a speaker skimming application in which databases of recorded conversations belonging to an ongoing investigation can be anno- tated and quickly browsed by an operator. The paper discusses the criticalities introduced by the characteristics of the audio sig- nals under consideration - in particular background noise and chan- nel/coding distortions - as well as the requirements and functional- ities of the system under development. It is shown that the perfor- mance of state-of-the-art approaches degrades significant ly in pres- ence of moderately high background noise. Finally, a novel speaker recognizer based on phonetic features and an ensemble classifier is presented. Results show that the proposed approach improves performance on clean audio, and suggest that it can be employed towards improved real-world robustness. |
Year | Venue | Keywords |
---|---|---|
2009 | EUSIPCO | audio signal processing,pattern classification,signal denoising,speaker recognition,speech processing,audio cleaning,audio signal characteristics,automatic speaker recognition system,background noise,channel distortion,classifier ensembling,coding distortion,intelligence applications,phonetic features,recorded conversation database,speaker skimming application |
Field | DocType | ISBN |
Audio signal,Background noise,Computer science,Feature extraction,Coding (social sciences),Speech recognition,Robustness (computer science),Speaker recognition,Speaker diarisation,Classifier (linguistics) | Conference | 978-161-7388-76-7 |
Citations | PageRank | References |
1 | 0.35 | 10 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Enrico Marchetto | 1 | 1 | 1.03 |
Federico Avanzini | 2 | 208 | 32.45 |
Federico Flego | 3 | 55 | 6.19 |