Find research datasets worth reusing

opencc-by-nc-4.0Jan 2022View details →

EmoWOZ: A Large-Scale Corpus and Labelling Scheme for Emotion Recognition in Task-Oriented Dialogue Systems

This is the dataset created for the paper, "EmoWOZ: A Large-Scale Corpus and Labelling Scheme for Emotion Recognition in Task-Oriented Dialogue Systems" (https://arxiv.org/abs/2109.04919). EmoWOZ is based on MultiWOZ, a multi-domain task-oriented dialogue dataset (https://github.com/budzianowski/multiwoz). It contains more than 11K task-oriented dialogues with more than 83K emotion annotations of user utterances. In addition to Wizard-of-Oz dialogues from MultiWOZ, we collect human-machine dialogues within the same set of domains to sufficiently cover the space of various emotions that can happen during the lifetime of a data-driven dialogue system. There are 7 emotion labels, which are adapted from the OCC emotion models. For data format and label definition, please refer to README.md.

opencc-by-4.0Aug 2022View details →

Hi, KIA: A Speech Emotion Recognition Dataset for Wake-Up Words

Hi,KIA dataset is a shared short Wakeup Word database focusing on perceived emotion in speech The dataset contains 488 Wakeup Word speech.  For more detailed information about the dataset, please refer to our paper: Hi, KIA: A Speech Emotion Recognition Dataset for Wake-Up Words File Description <ul> <li>wav/: wav files. <ul> <li>Filename f`{gender}_{pid}_{scene}_{trial}_{emotion}.wav` The first letter was used to express emotion.  </li> </ul> </li> <li>annotation/: Information related to annotation and human validation of the entire speech</li> <li> split: 8fold data split with {train, valid, test}.csv  </li> <li> handcraft: Features used for data EDA and baseline performance </li> <li> best_weights: wav2vec2.0 context network finetuning weights for re-implementation. Due to file size, we attach only fold M1, F5 </li> </ul>   Reference <ul> </ul> Hi, KIA: A Speech Emotion Recognition Dataset for Wake-Up Words [[ArXiv](https://arxiv.org/abs/2211.03371)] ``` @inproceedings{kim2022hi,   title={Hi, KIA: A Speech Emotion Recognition Dataset for Wake-Up Words},   author={Taesu Kim, SeungHeon Doh, Gyunpyo Lee, Hyung seok Jun, Juhan Nam, Hyeon-Jeong Suk},   booktitle={Proceedings of the 14th Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA)},   year={2022} } ```

Figure 9. Classification accuracy regardless the ethnic group (Total accuracy 75%)-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

Our experiments show that, the impact of ethnic group on the accuracy of emotions recognition is a positive where the accuracy of emotion recognition considering ethnic group is 83.3% as shown in Figure 8, and we got 75% of accuracy regardless ethnic group as shown in Figure 9.

Figure 8. Classification accuracy of emotions considering the ethnic group-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

To study the accuracy of emotion recognition for our approach regardless the ethnic group we used 108 images for the training representing six emotions of six persons. For testing, we used 36 images representing six emotions of six persons. On the other hand, to study the accuracy of emotion recognition for our approach considering the ethnic group we used 36 images for training for each ethnic group representing six emotions of six persons, and test the classifier by using 12 images representing six emotions of six persons.

Figure 6. Interface of FFE program-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

Points, Features Extraction and save all input information for the classifier (Features, Ethnic group, Gender and emotion). Figure 6 shows the interface of FFE program.

Figure 7. Samples of MSFDE dataset-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

We performed groups of experiments to study the impact of ethnic group (race) in the accuracy of emotion recognition with three kinds of ethnic groups (Asian, Caucasian as African). So we have three experiments, each experiment has a neural network as a classifier, and each neural network has three layers where there are 16 neurons in the hidden layer except Asian network has 17 neurons (the best result with 17 neurons for Asians).

Figure 3. 46 points are selected on face elements to describe the emotions.-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

The number of points and the position of points are not standardized, but it is depending on the features that will be extracted, and used for the classifier. Many researches use various number of points and positions based on their view about the feature to be considered [13] [18] [19]. Figure 3 shows the points we used.

Figure 1. A proposed approach-of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

Our proposed approach uses the face expression to detect the emotions through five steps that shows in Figure 1.

Figure. 2. Examples of Angry from different races. (A,B) African. (C,D) Asian. (E,F) Caucasian.-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

We chose 46 points which are distributed over human face image and use these points for features extraction. The choice of these points is to determine the shape of each element of the face (eyes, eyebrows and mouth), because the shape of these elements is changeable for each emotion, but these changes are different for each race as shown in Figure 2.

Figure 4. Distance between eyebrow and eye.-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

Based on what we stated above, we need to extract 28 features, which describe the distances between certain points explained in the previous stage, these features are classified into six groups, and each group describes the features of one face element. All features are a vertical distances between two points. Group one contains seven features for mouth, groups two and three contains 14 features for eyes, groups four and five contain six features for eyebrows, and the last group has one feature only which is the distance between the beginning of the eyebrow and the beginning of the eye in same side, this is significant (from point 23 to 15) because it is used to measure the distance of eyebrow from the eye. This feature is shown in Figure 4 by a line.

Figure 5. ANN Structure 4.-Impact of Ethnic Group on Human Emotion Recognition Using Backpropagation Neural Network

For classification purpose of the emotions, we use ANN of supervised learning based on backpropagation algorithm. Backpropagation neural network architecture is used with its standards learning function with 28 inputs representing the extracted features and 6 outputs representing 6 emotions, happy, sad, angry, fear, shame and disgust. the emotions. We have also a hidden layer with 16 nodes selected after various trails to obtain the best results. The used ANN is depicted in Figure 5.

opencc-by-4.0Aug 2015View details →

Figure 2. Emotion recognition using EEG-Classification of Human Emotion from Deap EEG Signal Using Hybrid Improved Neural Networks with Cuckoo Search

This section describes that collection of EEG signals for different emotion recognition experiments. The electroencephalography signals of 32 participants were recorded during one minute videos. Based on that participants are rated in terms of valence and arousal, like/dislike, familiarities and dominance. Emotion related ratings are given based on the online self assessment which is 120 one minute extracted music videos, which are rated by 14-16 volunteers based on arousal and valence. The EEG signals were recorded using 64 electrodes, first 62 electrodes are active electrode, one for reference and remaining one is ground electrode. All the electrodes are placed on the scalp which is made up of the Ag/Ag-Cl. Figure 2 shown in below which is represent the basic EEG signal recording methods and emotion analysis process.

opencc-by-4.0Dec 2018View details →

Emotion recognition: social appraisal and offset task, 8-12 yo.

Data from social appraisal task and offset task in 57 8-12 yo. school children, with pedagogy school type information.

opencc-by-4.0Jul 2021View details →

EMOPIA: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation

EMOPIA (pronounced ‘yee-mò-pi-uh’) dataset is a shared multi-modal (audio and MIDI) database focusing on perceived emotion in pop piano music, to facilitate research on various tasks related to music emotion. The dataset contains 1,087 music clips from 387 songs and clip-level emotion labels annotated by four dedicated annotators.  For more detailed information about the dataset, please refer to our paper: <a href="https://arxiv.org/abs/2108.01374">EMOPIA: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation</a>.  File Description <ul> <li>midis/: midi clips transcribed using GiantMIDI. <ul> <li>Filename `Q1_xxxxxxx_2.mp3`: Q1 means this clip belongs to Q1 on the V-A space; xxxxxxx is the song ID on YouTube, and the `2` means this clip is the 2nd clip taken from the full song.</li> </ul> </li> <li>metadata/: metadata from YouTube. (Got when crawling)</li> <li> songs_lists/: YouTube URLs of songs. </li> <li> tagging_lists/: raw tagging result for each sample. </li> <li> label.csv: metadata that records filename, 4Q label, and annotator. </li> <li> metadata_by_song.csv: list all the clips by the song. Can be used to create the train/val/test splits to avoid the same song appear in both train and test. </li> <li> scripts/prepare_split.ipynb: the script to create train/val/test splits and save them to csv files. </li> </ul> ------ 2.2 Update <ul> <li>Add tagging files in tagging_lists/ that are missing in the previous version.</li> <li>Add timestamps.json for easier usage. It records all the timestamps in dict format. You can see scripts/load_timestamp.ipynb for the format example.</li> <li>Add scripts/timestamp2clip.py: After the raw audio are crawled and put in audios/raw, you can use this script to get audio clips. The script will read timestamps.json and use the timestamp to extract clips. The clips will be saved to audios/seg folder.</li> <li>remove 7 midi files that were added by mistake, and also corrected the number in metadata_by_song.csv.</li> </ul>   2.1 Update Add one file and one folder: <ul> <li>key_mode_tempo.csv: key, mode, and tempo information extracted from files.</li> <li>CP_events/:  CP events used in our paper. Extracted using this <a href="https://github.com/YatingMusic/compound-word-transformer/blob/main/dataset/representations/uncond/cp/corpus2events.py">script</a>, and add the emotion event to the front.</li> </ul> Modify one folder: <ul> <li>The REMI_events/ files in version 2.0 contain some information that is not related to the paper, so remove it.</li> </ul>   2.0 Update Add two new folders: <ul> <li>corpus/:  processed data that following <a href="https://github.com/YatingMusic/compound-word-transformer/blob/main/dataset/Dataset.md">the preprocessing flow</a>. (Please notice that although we have <code>1078</code> clips in our dataset, we lost some clips during steps 1~4 of the flow, so the final number of clips in this <code>corpus</code> is <code>1052</code>, and that's the number we used for training the generative model.)</li> <li>REMI_events/: REMI event for each midi file. They are generated using this <a href="https://github.com/YatingMusic/compound-word-transformer/blob/main/dataset/representations/uncond/remi/corpus2events.py">script</a>.</li> </ul> --------        Cite this dataset <pre><code>@inproceedings{{EMOPIA}, author = {Hung, Hsiao-Tzu and Ching, Joann and Doh, Seungheon and Kim, Nabin and Nam, Juhan and Yang, Yi-Hsuan}, title = {{MOPIA}: A Multi-Modal Pop Piano Dataset For Emotion Recognition and Emotion-based Music Generation}, booktitle = {Proc. Int. Society for Music Information Retrieval Conf.}, year = {2021} }</code></pre>

opencc-by-4.0Apr 2023View details →

Appendix: Informative Speech Features based on Emotion Classes and Gender in Explainable Speech Emotion Recognition

Appendix tables for the paper "Informative Speech Features based on Emotion Classes and Gender in Explainable Speech Emotion Recognition". Feature informativeness information was gathered using SHAP values. TABLE VIII: Table of Statistics and Individual t-Test Results Between Each Emotion vs Neutral TABLE IX: (continue)Table of Statistics and Individual t-Test Results Between Each Emotion vs Neutral TABLE X: Table of Statistics and Individual t-Test Results Between Genders TABLE XI: (continue) Table of Statistics and Individual t-Test Results Between Genders TABLE XII: Table of 5 the most informative feature for each model, according to SHAP values TABLE XIII: (continue)Table of 5 the most informative feature for each model, according to SHAP values

opencc-by-4.0Jul 2023View details →

GamER: Gameplay Physiological Signal Dataset for Emotion Recognition

A dataset of physiological signals (EEG, ECG, EDA, EOG and Respiration) collected from 48 subjects while they were playing simple games designed to elicit emotional responses. Data were collected using the Biosignals PLUX Researcher kit, which allows up to 10 hours of signal recordings at up to 3kHz sampling rate and 16-bit resolution per channel, while recording data from up to eight sensors simultaneously. The dataset consists of 3 files: <ol> <li>data.zip: Contains the physiological signals. They are organized in 1 folder per subject, which contains the recorded data in .h5 and .txt format as provided by the Researcher kit.</li> <li>emotion_annotation.csv: Contains the emotion annotation that was produced by a psychologist based on the subjects self-reports and by examining the video of the data collection process (the videos are not included for privacy reasons)</li> <li>video_offsets.csv: Additional information for synchronizing the video-based annotations with the signals.</li> </ol> For more information, refer to the accompanying publication. Disclaimer: All subjects have provided written consent for the publication of their anonymized physiological signals.

opencc-by-4.0Aug 2019View details →

Emotion recognition HD - Data and R code

Data and R code - Recognition of emotion from subtle and non-stereotypical dynamic facial expressions in Huntington's diseaseRecognition of emotion from subtle and non-stereotypical dynamic facial expressions in Huntington's disease

opencc-by-4.0Nov 2023View details →

Use of artificial intelligence techniques for the recognition of human emotions: a bibliometric analysis

Human emotion recognition with AI uses physiological, audiovisual, and linguistic signals. Despite its importance and great progress in emotion recognition, several challenges remain in generalization and evaluation through standards and shared data, as well as other research gaps. Therefore, the objective is to analyze the scientific production on the use of artificial intelligence techniques for the recognition of human emotions. This study uses bibliometric analysis following the guidelines of the PRISMA-2020 statement for literature reviews. Based on the results of the bibliometrics on the use of artificial intelligence techniques in for the recognition of human emotions, significant conclusions are obtained that improve the understanding of the current panorama in this field of research. A growing interest in the subject is observed during the years 2023, 2022, 2021 and 2020, which demonstrates the relevance and potential of artificial intelligence in the recognition of human emotions. A cubic polynomial growth in the number of scientific articles is observed, demonstrating a constant expansion of knowledge and support for future trends. Leading authors and journals are identified, highlighting global collaboration in China and India. The thematic evolution shows maturity and progressive specialization, with emerging concepts that promise future research and innovative applications.

opencc-by-nc-4.0Jun 2023View details →

The EmoHI Test stimuli: Measuring vocal emotion recognition in hearing-impaired populations

Before reading this file, make sure you have read the README.1.pdf file. That file also contains information about the license these materials are distributed under. Versions <ul> <li>Version 2: This is the current version. To cite this version specifically, use DOI 10.5281/zenodo.7997063. In this version we fixed some naming mistakes in the files (in 8 of the files the sentence was identified as <code>t2</code> instead of <code>s2</code>), and added two missing stimuli (<code>t5_neutral_t2_u04.wav</code> and <code>t5_sad_t1_u05.wav</code>).</li> <li>Version 1: This was the initial version. To cite that version specifically, use DOI 10.5281/zenodo.3689710.</li> </ul> The latest version of the EmoHI material can be downloaded from <a href="https://doi.org/10.5281/zenodo.3689709">https://doi.org/10.5281/zenodo.3689709</a>. Please always check that you have the latest version, and that you comply with the current license requirements. The EmoHI Test The EmoHI Test was developed to measure the accuracy at which participants can recognize vocal emotions based on pseudospeech sentences that were produced in a happy, angry sad, or neutral manner. The EmoHI Test recordings are particularly suitable for testing hearing-impaired populations due to their high sound quality. All recordings, including the ones that were used in Nagels et al. (2020, PeerJ, <a href="https://doi.org/10.7717/peerj.8773">doi: 10.7717/peerj.8773</a>), are made available here. The stimuli were recorded in an anechoic room at a sampling rate of 44.1 kHz. The microphone was placed at a distance of approximately 30 cm (12 in) from the speaker. The recordings were made by connecting a standing Røde NT1 microphone to a Presonus TubePre V2 preamplifier and a TASCAM DR-100 portable digital recorder. The gain of the recordings was adjusted for each emotion production using the preamplifier to record the stimuli at an intensity level that was approximately the same across emotions to reduce large intensity differences between the recordings of different emotions. The files are not RMS equalized. Citation When using this repository in your research, please cite the repository itself. For this version: <blockquote> Nagels L., Gaudrain E., Hendriks P., & Başkent D. (2023, June 2). The EmoHI Test stimuli: Measuring vocal emotion recognition in hearing-impaired populations. Version 2. Zenodo. <a href="https://doi.org/10.5281/zenodo.7997063">https://doi.org/10.5281/zenodo.7997063</a> </blockquote> Also cite the PeerJ article that describes the material: <blockquote> Nagels L., Gaudrain E., Vickers D., Matos Lopes M., Hendriks P., Başkent D. (2020). Development of vocal emotion recognition in school-age children: The EmoHI test for hearing-impaired populations. PeerJ 8:e8773 <a href="https://doi.org/10.7717/peerj.8773">https://doi.org/10.7717/peerj.8773</a> </blockquote> Sound file name structure The sound files are named using the following convention: <code>t[1-6]_{emotion}_s{1,2}_u[01-18].wav</code> <ul> <li><code>t[1-6]</code> represents the talker who produced the stimulus: <code>t1</code>, <code>t2</code>, <code>t3</code>, <code>t4</code>, <code>t5</code>, or <code>t6</code></li> <li><code>{emotion}</code> is the label of the emotion that was produced: <code>neutral</code>, <code>happy</code>, <code>angry</code>, or <code>sad</code></li> <li><code>s{1,2}</code> is the pseudospeech sentence that was used: <code>s1</code> for "Koun se mina lod belam." <code>s2</code> for "Nekal ibam soud molen."</li> <li><code>u[01-18]</code> is the utterance number: Number ranging from <code>u01</code> to <code>u18</code></li> </ul> For instance, <code>t1_happy_s2_u01.wav</code> is utterance 1 of talker <code>t1</code> producing emotion "happy" using sentence 2. Talker demographic information The table below gives an overview of the voice characteristics from the talkers who produced the EmoHI test stimuli. <table> <thead> <tr> <th scope="col">Talker</th> <th scope="col">Age (years)</th> <th scope="col">Gender</th> <th scope="col">Height (m)</th> <th scope="col">Mean F0 (Hz)</th> <th scope="col">F0 range (Hz)</th> </tr> </thead> <tbody> <tr> <td>t1</td> <td>48</td> <td>f</td> <td>1.72</td> <td>253.14</td> <td>179.97 – 421.81</td> </tr> <tr> <td>t2</td> <td>36</td> <td>f</td> <td>1.68</td> <td>302.23</td> <td>200.71 – 437.38</td> </tr> <tr> <td>t3</td> <td>27</td> <td>m</td> <td>1.85</td> <td>166.92</td> <td>100.99 – 296.47</td> </tr> <tr> <td>t4</td> <td>45</td> <td>m</td> <td>1.90</td> <td>149.41</td> <td>96.97 – 274.72</td> </tr> <tr> <td>t5</td> <td>25</td> <td>f</td> <td>1.63</td> <td>282.89</td> <td>199.49 – 429.38</td> </tr> <tr> <td>t6</td> <td>24</td> <td>m</td> <td>1.75</td> <td>167.76</td> <td>87.46 – 285.79</td> </tr> </tbody> </table> Supporting data The behavioural data from the PeerJ article is accessible at <a href="https://doi.org/10.34894/BDMX6D">https://doi.org/10.34894/BDMX6D</a>.