A feature selection-based speaker clustering method for paralinguistic tasks

Gosztolya, Gábor; Tóth, László

doi:10.1007/s10044-017-0612-0

Cited by 6 publications

(5 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By reviewing the health based Computational Paralinguistics Challenge (ComParE) series and Audio/Visual Emotion Challenge (AVEC) workshops we observed a steady increase in the percentage of deep learning based entries from none in 2011 through to two-thirds in ComParE-2017 However, it is clear from this review that deep learning systems are still not yet a dominant force in speech and health. While two challenges have been won using a deep learning based approach: 2015's Parkinson's condition sub-challenge [118] and 2017's cold and flu sub-challenge [140], it is debatable given the results in [126] if the winning approach in 2015 was due to the deep learning approach or the postprocessing speaker clustering method employed [118]. Furthermore, the majority of challenge participants still use very conventional feature extraction and classification paradigms.…”

Section: Resultsmentioning

confidence: 99%

“…Kaya et al [125] approach, used a novel framework based on Fisher vector encoding of extracted features and cascaded normalisation to account for variability due to differing speakers characteristics and spoken content. Recently, DRN approaches were shown to produce equivalent performance as conventional linear SVM and AdaBoost systems for the eating task [126]. However, this comparison was not the main aim of the authors.…”

Section: Eating (2015)mentioning

confidence: 98%

See 1 more Smart Citation

Speech analysis for health: Current state-of-the-art and the increasing impact of deep learning

Cummins

Baird

Schuller

2018

Methods

131

View full text Add to dashboard Cite

Due to the complex and intricate nature associated with their production, the acoustic-prosodic properties of a speech signal are modulated with a range of health related effects. There is an active and growing area of machine learning research in this speech and health domain, focusing on developing paradigms to objectively extract and measure such effects. Concurrently, deep learning is transforming intelligent signal analysis, such that machines are now reaching near human capabilities in a range of recognition and analysis tasks. Herein, we review current state-of-the-art approaches with speech-based health detection, placing a particular focus on the impact of deep learning within this domain. Based on this overview, it is evident while that deep learning based solutions be become more present in the literature, it has not had the same overall dominating effect seen in other related fields. In this regard, we suggest some possible research directions aimed at fully leveraging the advantages that deep learning can offer speech-based health detection.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Eating (2015)mentioning

confidence: 98%

Speech analysis for health: Current state-of-the-art and the increasing impact of deep learning

Cummins

Baird

Schuller

2018

Methods

131

View full text Add to dashboard Cite

show abstract

“…As can be observed, our system obtains the second best result, only from behind [5], that was the winner of the 2014 COMPARE challenge for CL level classification. [30] 63.10 Fusion of [2], MFCC+SDC supervectors and SCF supervectors + GMM-SVM [3] 63.70 Feature Selection + Speaker Clustering + SVM [31] 64.80 LSTM Unnormalized Saliency (this paper) 66.80 Fusion of 4 Speech Streams + i-Vectors + SVM [5] 68.90…”

Section: Resultsmentioning

confidence: 99%

A Saliency-Based Attention LSTM Model for Cognitive Load Classification from Speech

Gallardo-Antolín¹,

Montero²

2019

Interspeech 2019

View full text Add to dashboard Cite

Cognitive Load (CL) refers to the amount of mental demand that a given task imposes on an individual's cognitive system and it can affect his/her productivity in very high load situations. In this paper, we propose an automatic system capable of classifying the CL level of a speaker by analyzing his/her voice. Our research on this topic goes into two main directions. In the first one, we focus on the use of Long Short-Term Memory (LSTM) networks with different weighted pooling strategies for CL level classification. In the second contribution, for overcoming the need of a large amount of training data, we propose a novel attention mechanism that uses the Kalinli's auditory saliency model. Experiments show that our proposal outperforms significantly both, a baseline system based on Support Vector Machines (SVM) and a LSTM-based system with logistic regression attention model.

show abstract

“…However, for specific computational paralinguistic tasks it was shown (see e.g. [22,26,66,67]) that speaker-wise feature standardization might assist the subsequent classification steps. Therefore, in our preliminary tests we also experimented with this standardization approach for both the original and the PTFE feature sets.…”

Section: Feature Standardizationmentioning

confidence: 99%

Posterior-thresholding feature extraction for paralinguistic speech classification

Gosztolya

2019

Knowledge-Based Systems

Self Cite

View full text Add to dashboard Cite

The standard approach for handling computational paralinguistic speech tasks is to extract several thousand utterance-level features from the speech excerpts, and use machine learning methods such as Support Vector Machines and Deep Neural Networks (DNNs) for the actual classification task. In contrast, Automatic Speech Recognition handles the speech signal in small, equal-sized parts called frames. Although the speech community has developed techniques for efficient frame classification, these efforts have mostly been ignored in computational paralinguistics. In this study we propose a simple, three-step technique to utilize frame-level DNN training know-how in computational paralinguistics. We show that this method by itself provides good accuracy scores, and by combining it with the standard paralinguistic classification approach, we get close to the performance of heavyweight, state-of-the-art techniques such as Fisher vector analysis. However, our approach has the advantage that it can be easily realized by using standard speech recognition tools. To demonstrate the generic applicability of this three-step method proposed, we performed our experiments on four different corpora containing different paralinguistic tasks. Overall, we were able to achieve improvements over the baseline score in all four cases, leading to relative error reductions of up to 19%.

show abstract

A feature selection-based speaker clustering method for paralinguistic tasks

Cited by 6 publications

References 35 publications

Speech analysis for health: Current state-of-the-art and the increasing impact of deep learning

Speech analysis for health: Current state-of-the-art and the increasing impact of deep learning

A Saliency-Based Attention LSTM Model for Cognitive Load Classification from Speech

Posterior-thresholding feature extraction for paralinguistic speech classification

Contact Info

Product

Resources

About