This work addresses the problem of automatic speaker localization and tracking in a real lecture scenario.Evaluation criteria recently adopted under CHIL and NIST benchmarking are outlined.Two speaker localization systems are described, which are based on the use of Cross-power Spectrum Phase analysis and of Global Coherence Field. Benchmarking results were obtained on a set of 13 lectures and showed an average RMS error of about 30 cm in the speaker localization.
Speaker Localization in CHIL Lectures: Evaluation Criteria and Results
Omologo, Maurizio;Svaizer, Piergiorgio;Brutti, Alessio;Cristoforetti, Luca
2006-01-01
Abstract
This work addresses the problem of automatic speaker localization and tracking in a real lecture scenario.Evaluation criteria recently adopted under CHIL and NIST benchmarking are outlined.Two speaker localization systems are described, which are based on the use of Cross-power Spectrum Phase analysis and of Global Coherence Field. Benchmarking results were obtained on a set of 13 lectures and showed an average RMS error of about 30 cm in the speaker localization.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
