12th Annual Conference of the International Speech Communication Association

Florence, Italy
August 27-31. 2011

Acoustic-Linguistic Recognition of Interest in Speech with Bottleneck-BLSTM Nets

Martin Wöllmer, Felix Weninger, Florian Eyben, Björn Schuller

Technische Universität München, Germany

This paper proposes a novel technique for speech-based interest recognition in natural conversations. We introduce a fully automatic system that exploits the principle of bidirectional Long Short-Term Memory (BLSTM) as well as the structure of so-called bottleneck networks. BLSTM nets are able to model a self-learned amount of context information, which was shown to be beneficial for affect recognition applications, while bottleneck networks allow for efficient feature compression within neural networks. In addition to acoustic features, our technique considers linguistic information obtained from a multi-stream BLSTM-HMM speech recognizer. Evaluations on the TUM AVIC corpus reveal that the bottleneck-BLSTM method prevails over all approaches that have been proposed for the Interspeech 2010 Paralinguistic Challenge task.

Full Paper

Bibliographic reference.  Wöllmer, Martin / Weninger, Felix / Eyben, Florian / Schuller, Björn (2011): "Acoustic-linguistic recognition of interest in speech with bottleneck-BLSTM nets", In INTERSPEECH-2011, 77-80.