Multimodal embedding fusion for robust speaker role recognition in video broadcast

Abstract : Person role recognition in video broadcasts consists in classifying people into roles such as anchor, journalist, guest, etc. Existing approaches mostly consider one modality, either audio (speaker role recognition) or image (shot role recognition), firstly because of the non-synchrony between both modalities, and secondly because of the lack of a video corpus annotated in both modalities. Deep Neural Networks (DNN) approaches offer the ability to learn simultaneously feature representations (embeddings) and classification functions. This paper presents a multimodal fusion of audio, text and image embeddings spaces for speaker role recognition in asynchronous data. Monomodal embeddings are trained on exogenous data and fine-tuned using a DNN on 70 hours of French Broadcasts corpus for the target task. Experiments on the REPERE corpus show the benefit of the embeddings level fusion compared to the monomodal embeddings systems and to the standard late fusion method.
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01475413
Contributor : Sebastien Delecraz <>
Submitted on : Thursday, February 23, 2017 - 4:45:43 PM
Last modification on : Monday, March 4, 2019 - 2:04:14 PM
Long-term archiving on : Wednesday, May 24, 2017 - 2:36:37 PM

File

favre_asru2015b.pdf
Files produced by the author(s)

Identifiers

Collections

Citation

Mickael Rouvier, Sebastien Delecraz, Benoit Favre, Meriem Bendris, Frédéric Bechet. Multimodal embedding fusion for robust speaker role recognition in video broadcast. Automatic Speech Recognition and Understanding, Dec 2015, Scottsdale, United States. pp.383 - 389, ⟨10.1109/ASRU.2015.7404820⟩. ⟨hal-01475413⟩

Share

Metrics

Record views

185

Files downloads

203