Percorrer por autor "Gomes, Paulo"

A mostrar 1 - 5 de 5
  • Miniatura
    Item
    Bi-modal music emotion recognition: Novel lyrical features and dataset
    (9th International Workshop on Music and Machine Learning – MML’2016 – in conjunction with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases – ECML/PKDD 2016, October 2016, 2016) Malheiro, Ricardo; Panda, Renato; Gomes, Paulo; Paiva, Rui
    This research addresses the role of audio and lyrics in the music emotion recognition. Each dimension (e.g., audio) was separately studied, as well as in a context of bimodal analysis. We perform classification by quadrant categories (4 classes). Our approach is based on several audio and lyrics state-of-the-art features, as well as novel lyric features. To evaluate our approach we create a ground-truth dataset. The main conclusions show that unlike most of the similar works, lyrics performed better than audio. This suggests the importance of the new proposed lyric features and that bimodal analysis is always better than each dimension.
  • Miniatura
    Item
    Classification and regression of music lyrics: Emotionally-significant features
    (8th International Conference on Knowledge Discovery and Information Retrieval, 2016-01) Malheiro, Ricardo; Panda, Renato; Gomes, Paulo; Paiva, Rui Pedro
    This research addresses the role of lyrics in the music emotion recognition process. Our approach is based on several state of the art features complemented by novel stylistic, structural and semantic features. To evaluate our approach, we created a ground truth dataset containing 180 song lyrics, according to Russell’s emotion model. We conduct four types of experiments: regression and classification by quadrant, arousal and valence categories. Comparing to the state of the art features (ngrams - baseline), adding other features, including novel features, improved the F-measure from 68.2%, 79.6% and 84.2% to 77.1%, 86.3% and 89.2%, respectively for the three classification experiments. To study the relation between features and emotions (quadrants) we performed experiments to identify the best features that allow to describe and discriminate between arousal hemispheres and valence meridians. To further validate these experiments, we built a validation set comprising 771 lyrics extracted from the AllMusic platform, having achieved 73.6% F-measure in the classification by quadrants. Regarding regression, results show that, comparing to similar studies for audio, we achieve a similar performance for arousal and a much better performance for valence.
  • Miniatura
    Item
    Emotionally-relevant features for classification and regression of music lyrics
    (IEEE TRANSACTIONS ON JOURNAL AFFECTIVE COMPUTING, MANUSCRIPT ID, 2016-08-08) Malheiro, Ricardo; Panda, Renato; Gomes, Paulo; Paiva, Rui Pedro
    This research addresses the role of lyrics in the music emotion recognition process. Our approach is based on several state of the art features complemented by novel stylistic, structural and semantic features. To evaluate our approach, we created a ground truth dataset containing 180 song lyrics, according to Russell’s emotion model. We conduct four types of experiments: regression and classification by quadrant, arousal and valence categories. Comparing to the state of the art features (ngrams - baseline), adding other features, including novel features, improved the F-measure from 69.9%, 82.7% and 85.6% to 80.1%, 88.3% and 90%, respectively for the three classification experiments. To study the relation between features and emotions (quadrants) we performed experiments to identify the best features that allow to describe and discriminate each quadrant. To further validate these experiments, we built a validation set comprising 771 lyrics extracted from the AllMusic platform, having achieved 73.6% F-measure in the classification by quadrants. We also conducted experiments to identify interpretable rules that show the relation between features and emotions and the relation among features. Regarding regression, results show that, comparing to similar studies for audio, we achieve a similar performance for arousal and a much better performance for valence.
  • Miniatura indisponível
    Item
    Keyword-Based Approach for Lyrics Emotion Variation Detection
    (8th International Conference on Knowledge Discovery and Information Retrieval, 2016-01) Malheiro, Ricardo; Oliveira, Hugo Gonçalo; Gomes, Paulo; Paiva, Rui Pedro
    This research addresses the role of the lyrics in the context of music emotion variation detection. To accomplish this task we create a system to detect the predominant emotion expressed by each sentence (verse) of the lyrics. The system employs Russell’s emotion model and contains 4 sets of emotions associated to each quadrant. To detect the predominant emotion in each verse, we propose a novel keyword-based approach, which receives a sentence (verse) and classifies it in the appropriate quadrant. To tune the system parameters, we created a 129-sentence training dataset from 68 songs. To validate our system, we created a separate ground-truth containing 239 sentences (verses) from 44 songs annotated manually with an average of 7 annotations per sentence. The system attains 67.4% F-Measure score.
  • Miniatura
    Item
    Music Emotion Recognition from Lyrics: a comparative study
    (6th International Workshop on Machine Learning and Music (MML13). Held in Conjunction with the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECMLPPKDD13), 2013-09) Malheiro, Ricardo; Panda, Renato; Gomes, Paulo; Paiva, Rui Pedro
    We present a study on music emotion recognition from lyrics. We start from a dataset of 764 samples (audio+lyrics) and perform feature extraction using several natural language processing techniques. Our goal is to build classifiers for the different datasets, comparing different algorithms and using feature selection. The best results (44.2% F-measure) were attained with SVMs. We also perform a bi-modal analysis that combines the best feature sets of audio and lyrics.The combination of the best audio and lyrics features achieved better results than the best feature set from audio only (63.9% F-Measure against 62.4% F-Measure).