The INTERSPEECH 2019 Computational Paralinguistics Challenge: Styrian Dialects, Continuous Sleepiness, Baby Sounds &amp; Orca Activity

Schuller, Björn; Batliner, Anton; Bergler, Christian; Pokorny, Florian B.; Krajewski, Jarek; Cychosz, Margaret; Vollmann, Ralf; Roelen, Sonja-Dana; Schnieder, Sebastian; Bergelson, Elika; Cristià, Alejandrina; Seidl, Amanda; Warlaumont, Anne S.; Yankowitz, Lisa; Nöth, Elmar; Amiriparian, Shahin; Hantke, Simone; Schmitt, Maximilian

doi:10.21437/interspeech.2019-1122

Cited by 74 publications

(119 citation statements)

References 18 publications

Supporting

Mentioning

115

Contrasting

Unclassified

Order By: Relevance

“…The so far most comprehensive standard feature set for openSMILE is the Com-ParE set. It is widely known, as it represented the official baseline feature set of the 2013-2019 Computational Paralinguistics ChallengEs (e.g., Schuller et al 2013Schuller et al , 2018Schuller et al , 2019 carried out in connection with the Annual Conferences of the International Speech Communication Association (INTERSPEECH conferences). The ComParE set comprises 6373 acoustic supra-segmental features, so-called higher-level descriptors (HLDs).…”

Section: Feature-based Representationmentioning

confidence: 99%

“…It allows for learning data representations from audio time series by means of a recurrent sequence-to-sequence autoencoder approach (Freitag et al 2017). Complementing the brute-force feature extraction tools openSMILE and openXBOW, in 2018 AUDEEP was elected as open-source representation learning tool kit for official baseline evaluation within the ongoing series of INTERSPEECH ComParE Challenges (Schuller et al 2018(Schuller et al , 2019. Thus, in that year, AUDEEP was initially applied to learn representations from preverbal data.…”

Section: Representation Learningmentioning

confidence: 99%

“…The discipline of intelligent audio analysis (Schuller 2013), which deals with the combination of advanced audio signal processing and machine learning technology, provided the methodological tools for the automatic retrieval of preverbal sound-related (meta-)information. Specific tasks are, e.g., the automatic differentiation of preverbal vocalization types (e.g., Schuller et al 2019), the automatic detection of infant distress (e.g., Chang and Li 2016;Chang et al 2015;Lavner et al 2016;Rodriguez and Caluya 2017;Schuller et al 2018), or the automatic recognition of the medical condition of the infant who produces the vocalization (e.g., Orlandi et al 2012;Pokorny et al 2016aPokorny et al , 2017. However, different learning tasks require different learning strategies.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Efficient Collection and Representation of Preverbal Data in Typical and Atypical Development

et al. 2020

Self Cite

View full text Add to dashboard Cite

Human preverbal development refers to the period of steadily increasing vocal capacities until the emergence of a child's first meaningful words. Over the last decades, research has intensively focused on preverbal behavior in typical development. Preverbal vocal patterns have been phonetically classified and acoustically characterized. More recently, specific preverbal phenomena were discussed to play a role as early indicators of atypical development. Recent advancements in audio signal processing and machine learning have allowed for novel approaches in preverbal behavior analysis including automatic vocalization-based differentiation of typically and atypically developing individuals. In this paper, we give a methodological overview of current strategies for collecting and acoustically representing preverbal data for intelligent audio analysis paradigms. Efficiency in the context of data collection and data representation is discussed. Following current research trends, we set a special focus on challenges that arise when dealing with preverbal data of individuals with late detected developmental disorders, such as autism spectrum disorder or Rett syndrome.

show abstract

Section: Feature-based Representationmentioning

confidence: 99%

Section: Representation Learningmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Efficient Collection and Representation of Preverbal Data in Typical and Atypical Development

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…More detailed statistics about the dataset and the label distribution are provided in Table 1 and Figure 1, respectively. The task of the challenge is to build a regression model that is able to predict the KSS rating for an audio recording [11].…”

Section: Sleep Corpusmentioning

confidence: 99%

Deep Neural Baselines for Computational Paralinguistics

et al. 2019

View full text Add to dashboard Cite

Detecting sleepiness from spoken language is an ambitious task, which is addressed by the Interspeech 2019 Computational Paralinguistics Challenge (ComParE). We propose an end-to-end deep learning approach to detect and classify patterns reflecting sleepiness in the human voice. Our approach is based solely on a moderately complex deep neural network architecture. It may be applied directly on the audio data without requiring any specific feature engineering, thus remaining transferable to other audio classification tasks. Nevertheless, our approach performs similar to state-of-the-art machine learning models.

show abstract

“…In the Orca Activity Challenge [1], underwater audio has to be classified as Orca or Non-Orca sound. Orcas (killer whales) are marine mammals that live in groups in every ocean in the world.…”

Section: Introductionmentioning

confidence: 99%

Spatial, Temporal and Spectral Multiresolution Analysis for the INTERSPEECH 2019 ComParE Challenge

Caraty¹,

Montacié²

2019

Interspeech 2019

View full text Add to dashboard Cite

The INTERSPEECH 2019 Orca Activity Challenge consists in the detection of the Orca sounds from underwater audio signal. Orca can produce a wide variety of sounds categorized in clicks, whistles and pulsed calls. Clicks are useful for echolocation, whistles and pulsed calls are used as social signals. Experiments were conducted on DeepAL Fieldwork Data (DLFD). Underwater sounds were recorded in northern British Columbia by a hydrophones array. Recordings were labeled by marine biologists in Orca sounds or Noise. We have investigated multiresolution analysis according to the three main relevant acoustic levels: spatial, temporal and spectral. For this purpose, we studied the beamforming array analysis, the multitemporal resolution and the multilevel wavelet decomposition. For the spatial level, a beamforming algorithm was used for denoising the underwater audio signal. For the temporal level, two sets of multitemporal three-level features were extracted using pyramidal representation. For the spectral level, in order to detect transient sound, wavelet analysis was computed using various wavelet families. At last, an Orca Activity detector was designed combining ComParE set with multitemporal and multilevel wavelet features. Experiments on the Test set have shown a significant improvement of 0.051, compared to the baseline performance of the Challenge (0.866).

show abstract

The INTERSPEECH 2019 Computational Paralinguistics Challenge: Styrian Dialects, Continuous Sleepiness, Baby Sounds & Orca Activity

Cited by 74 publications

References 18 publications

Efficient Collection and Representation of Preverbal Data in Typical and Atypical Development

Efficient Collection and Representation of Preverbal Data in Typical and Atypical Development

Deep Neural Baselines for Computational Paralinguistics

Spatial, Temporal and Spectral Multiresolution Analysis for the INTERSPEECH 2019 ComParE Challenge

Contact Info

Product

Resources

About