The HUMAINE Database

Douglas‐Cowie, Ellen; Cox, Cate; Martin, Jean‐Claude; Devillers, Laurence; Cowie, Roddy; Sneddon, Ian; McRorie, Margaret; Pélachaud, Catherine; Peters, Christopher J.; Lowry, Orla; Batliner, Anton; Hönig, Florian

doi:10.1007/978-3-642-15184-2_14

Cited by 37 publications

(25 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These are shown in Fig. 2 and are detailed in [7]. Each character has a different 'personality': 'Prudence' is matter-of-fact, 'Spike' is aggressive, 'Obadiah' is pessimistic, and 'Poppy' is cheerful.…”

Section: The Challenge Of Continuous Emotion Recognitionmentioning

confidence: 99%

“…More details on the configurations of the specific networks used for evaluations within this work can be found in Sect. 7.…”

Section: Lstm-rnnmentioning

confidence: 99%

“…Such an annotation tool is FEELtrace, which was introduced in [6]. The first, and so far only, naturalistic database annotated using the FEELtrace system, which is made available to researchers is the Belfast naturalistic sub-set of the HUMAINE database [7]. This sub-set is also known as the Belfast Sensitive Artificial Listener (SAL) data and will be used for all experiments in this article.…”

Section: Databasementioning

confidence: 99%

See 2 more Smart Citations

On-line emotion recognition in a 3-D activation-valence-time continuum using acoustic and linguistic cues

Eyben

Wöllmer

Graves

et al. 2009

J Multimodal User Interfaces

Self Cite

124

View full text Add to dashboard Cite

For many applications of emotion recognition, such as virtual agents, the system must select responses while the user is speaking. This requires reliable on-line recognition of the user's affect. However most emotion recognition systems are based on turnwise processing. We present a novel approach to on-line emotion recognition from speech using Long Short-Term Memory Recurrent Neural Networks. Emotion is recognised frame-wise in a two-dimensional valence-activation continuum. In contrast to current state-of-the-art approaches, recognition is performed on low-level signal frames, similar to those used for speech recognition. No statistical functionals are applied to low-level feature contours. Framing at a higher level is therefore unnecessary and regression outputs can be produced in real-time for every low-level input frame. We also investigate the benefits of including linguistic features on the signal frame level obtained by a keyword spotter.

show abstract

Section: The Challenge Of Continuous Emotion Recognitionmentioning

confidence: 99%

“…More details on the configurations of the specific networks used for evaluations within this work can be found in Sect. 7.…”

Section: Lstm-rnnmentioning

confidence: 99%

Section: Databasementioning

confidence: 99%

See 1 more Smart Citation

On-line emotion recognition in a 3-D activation-valence-time continuum using acoustic and linguistic cues

Eyben

Wöllmer

Graves

et al. 2009

J Multimodal User Interfaces

Self Cite

124

View full text Add to dashboard Cite

show abstract

“…We used the Sensitive Artificial Listener (SAL) Database [3] for this work. It contains naturalistic audio-visual conversational data taking place between a participant and a human operated avatar.…”

Section: Data Setmentioning

confidence: 99%

A multi-layer hybrid framework for dimensional emotion classification

Nicolaou

Güneş

Pantić

2011

Proceedings of the 19th ACM International Conference on Multimedia

View full text Add to dashboard Cite

This paper investigates dimensional emotion prediction and classification from naturalistic facial expressions. Similarly to many pattern recognition problems, dimensional emotion classification requires generating multi-dimensional outputs. To date, classification for valence and arousal dimensions has been done separately, assuming that they are independent. However, various psychological findings suggest that these dimensions are correlated. We therefore propose a novel, multi-layer hybrid framework for emotion classification that is able to model inter-dimensional correlations. Firstly, we derive a novel geometric feature set based on the (a)symmetric spatio-temporal characteristics of facial expressions. Subsequently, we use the proposed feature set to train a multi-layer hybrid framework composed of a temporal regression layer for predicting emotion dimensions, a graphical model layer for modeling valence-arousal correlations, and a final classification and fusion layer exploiting informative statistics extracted from the lower layers. This framework (i) introduces the Auto-Regressive Coupled HMM (ACHMM), a graphical model specifically tailored to accommodate not only inter-dimensional correlations but also to exploit the internal dynamics of the actual observations, and (ii) replaces the commonly used Maximum Likelihood principle with a more robust final classification and fusion layer. Subject-independent experimental validation, performed on a naturalistic set of facial expressions, demonstrates the effectiveness of the derived feature set, and the robustness and flexibility of the proposed framework.

show abstract

“…To illustrate these issues we can mention that although it is common to consider the 'big six' emotions of joy, sadness, fear, disgust, anger, and surprise, along with neutral, Douglas-Cowie, Cox et al [1], proposed a far greater list of 48 emotion categories; in the INTERSPEECH Challenges from 2009 to 2012, the number of proposed features has increased from 384 to 6125 [2,3] without a final set being decided upon; finally, as Scherer [4] reports, a wide range of classifiers, such as linear discriminant classifiers, k-nearest neighbor (KNN), Gaussian mixture model, support vector machines, decision tree algorithms (DT) and hidden Markov models have all been examined, and no definitive classifier has been chosen.…”

Section: Introductionmentioning

confidence: 99%

Deep Learning for Emotional Speech Recognition

Sánchez-Gutiérrez

Albornoz

Licona

et al. 2014

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Emotional speech recognition is a multidisciplinary research area that has received increasing attention over the last few years. The present paper considers the application of restricted Boltzmann machines (RBM) and deep belief networks (DBN) to the difficult task of automatic Spanish emotional speech recognition. The principal motivation lies in the success reported in a growing body of work employing these techniques as alternatives to traditional methods in speech processing and speech recognition. Here a well-known Spanish emotional speech database is used in order to extensively experiment with, and compare, different combinations of parameters and classifiers. It is found that with a suitable choice of parameters, RBM and DBN can achieve comparable results to other classifiers.

show abstract

The HUMAINE Database

Cited by 37 publications

References 8 publications

On-line emotion recognition in a 3-D activation-valence-time continuum using acoustic and linguistic cues

On-line emotion recognition in a 3-D activation-valence-time continuum using acoustic and linguistic cues

A multi-layer hybrid framework for dimensional emotion classification

Deep Learning for Emotional Speech Recognition

Contact Info

Product

Resources

About