Person name recognition in ASR outputs using continuous context models

Benjamin Bigot (1) , Grégory Senay (1) , Georges Linarès (1) , Corinne Fredouille (1) , Richard Dufour (1)

1 LIA - Laboratoire Informatique d'Avignon

The detection and characterization, in audiovisual documents, of speech utterances where person names are pronounced, is an important cue for spoken content analysis. This paper tackles the problematic of retrieving spoken person names in the 1-Best ASR outputs of broadcast TV shows. Our assumption is that a person name is a latent variable produced by the lexical context it appears in. Thereby, a spoken name could be derived from ASR outputs even if it has not been proposed by the speech recognition system. A new context modelling is proposed in order to capture lexical and structural information surrounding a spoken name. The fundamental hypothesis of this study has been validated on broadcast TV documents available in the context of the REPERE challenge.

Mots clés: Index Terms— spoken document retrieval; spoken name detection; lexical context representation; Index Terms— spoken document retrieval; Index Terms— spoken document retrieval; Index Terms— spoken document retrieval

Domaines: Informatique [cs]

Benjamin Bigot, Grégory Senay, Georges Linarès, Corinne Fredouille, Richard Dufour. Person name recognition in ASR outputs using continuous context models. IEEE International Conference on Acoustics, Speech and Signal Processing , May 2013, Vancouver, Canada. ⟨10.1109/ICASSP.2013.6639318⟩. ⟨hal-01314411⟩