Abandoning emotion classes - towards continuous emotion recognition with modelling of long-range dependencies

By Martin Wöllmer, Florian Eyben, S. Reiter, Björn Schuller, Cate Cox, Ellen Douglas-Cowie, R. Cowie. Interspeech, 2008

317 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

View Paper: https://www.semanticscholar.org/paper/20966cbf8d1813e585131d63aa27b01c622df04e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Categorical and dimensional affect analysis in continuous input: Current trends and future directions (2013, background): “, [57,130,131]).”
Recognising realistic emotions and affect in speech: State of the art and lessons learnt from the first challenge (2011, methodology): “An incremental update of the prediction is also possible by classiﬁers suited for this task that may at the same time learn how much past context information is beneﬁcial (Wo¨llmer et al., 2008); some toolkits, e.g., openEAR, already allow for this type of processing (Eyben et al., 2009).”
From speech to letters - using a novel neural network architecture for grapheme based ASR (2009, background): “It has proven highly successful for sequence transcription tasks involving difficult, real-world data [3], [4], including phoneme recognition on the TIMIT corpus [5].”
Continuous Prediction of Spontaneous Affect from Multiple Cues and Modalities in Valence-Arousal Space (2011, methodology): “Performance comparison shows that the proposed output-associative fusion framework provides a significantly improved prediction accuracy compared to feature-level and model-level fusion via BLSTM-NNs.”
Context-Dependent Sentiment Analysis in User-Generated Videos (2017, background): “The higher performance improvement on the IEMOCAP dataset indicates the necessity of modeling long-range dependencies among the utterances as continuous emotion recognition is a multiclass sequential problem where a person does not frequently change emotions (Wöllmer et al., 2008).”
Deep neural networks for acoustic emotion recognition: Raising the benchmarks (2011, background): “, in [15] with an average length of 20 minutes per speaker of natural human-SAL conversations.”
Analysis of EEG Signals and Facial Expressions for Continuous Emotion Detection (2016, background): “VIDEO is a form of visual art that conveys emotions in itscontent.”
Recent Advances in Recurrent Neural Networks (2017, background): “In [106], an LSTM network is shown to have better performance than support vector machines (SVMs) and conditional random fields (CRFs).”
Understanding LSTM - a tutorial into Long Short-Term Memory Recurrent Neural Networks (2019, unknown): “Other cognitive learning tasks include emotion recognition from speech [78], text generation [67], handwriting generation [24], constituency parsing [71], and conversational modelling [72].”
Emotion representation, analysis and synthesis in continuous space: A survey (2011, background): “Of these, Bulut et al. propose an interesting approach of emotional speech re-synthesis that consists of synthesis, recognition, parameter selection and re-synthesis modules that appears to improve the human evaluation of emotionally synthesised speech [85].”
OpenEAR — Introducing the munich open-source emotion and affect recognition toolkit (2009, methodology): “Results for SAL are obtained on the same data-set partitions as in [15].”
Automatic speech emotion recognition using modulation spectral features (2011, background): “However, recent research efforts also show an increasing interest in dimensional representations of emotions for SER (Grimm et al., 2007a; Grimm et al., 2007b; Wollmer et al., 2008; Giannakopoulos et al., 2009).”