Natural language processing of lyrics

Mahedero, Jose P. G.; Martínez, Apen Ruiz; Cano, Pedro; Koppenberger, Markus; Gouyon, Fabien

doi:10.1145/1101149.1101255

Cited by 54 publications

(25 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We have found no prior articles studying the combination of lyrics and acoustic information for this particular classification purpose. Mahedero et al [10] reported promising results in using lyrics for thematic categorization suggesting that a mood classification is possible. Neumayer and Rauber [12] have shown the complementarity of audio and lyrics in the context of genre classification, which is also encouraging.…”

Section: Related Workmentioning

confidence: 99%

Multimodal Music Mood Classification Using Audio and Lyrics

Laurier

Grivolla

Herrera

2008

2008 Seventh International Conference on Machine Learning and Applications

161

View full text Add to dashboard Cite

In this paper we present a study on music mood classification using audio and lyrics information. The mood of a song is expressed by means of musical features but a relevant part also seems to be conveyed by the lyrics. We evaluate each factor independently and explore the possibility to combine both, using Natural Language Processing and Music Information Retrieval techniques. We show that standard distance-based methods and Latent Semantic Analysis are able to classify the lyrics significantly better than random, but the performance is still quite inferior to that of audio-based techniques. We then introduce a method based on differences between language models that gives performances closer to audio-based classifiers. Moreover, integrating this in a multimodal system (audio+text) allows an improvement in the overall performance. We demonstrate that lyrics and audio information are complementary, and can be combined to improve a classification system.

show abstract

Section: Related Workmentioning

confidence: 99%

Multimodal Music Mood Classification Using Audio and Lyrics

Laurier

Grivolla

Herrera

2008

2008 Seventh International Conference on Machine Learning and Applications

161

View full text Add to dashboard Cite

show abstract

“…Pada text classification, algoritma ini memiliki dua asumsi yaitu asumsi Bag of Words yang berasumsi bahwa posisi kata dalam dokumen tidak berpengaruh dan Conditional Independence dimana masing-masing probabilitas setiap feature saling independence. Perhitungan Multinomial Naïve Bayes dirumuskan pada persamaan (2).…”

Section: Multinomial Naïve Bayesunclassified

“…Dari hal tersebut, lirik dapat digunakan sebagai objek penelitian pada text classification. Pada penelitian ini dibuat sebuah sistem yang mampu mengkategorikan lagu berdasarkan topik lagu seperti penelitian yang telah dilakukan oleh Jose P. G. Mahedero, Alvaro Martinez, dan Pedro Cano [2]. Kelas topik yang dihasilkan pada penelitian ini yaitu cinta, persahabatan, keluarga, religi, nasionalisme dan konten negatif.…”

Section: Introductionunclassified

Klasifikasi Topik pada Lirik Lagu dengan Metode Multinomial Naïve Bayes

Aziz¹

2016

Indosc 2016

View full text Add to dashboard Cite

The current system for categorizing songs on application like iTunes on the iPhone has been able to categorize songs by genre, artist, and album. However, there has been no categorization song by song topics, categorizing songs based on topic is still done manually. This research aims to create a system that categorizes songs by song topics using lyrics as research objects. This research begins by collecting lyrics from many website which serve as research dataset, then data preprocessing is done, consisting of case folding, tokenization, stopword removal and stemming. The next step is to do feature selection by using chi-square. Furthermore, feature extraction will be on the feature selection results by using bag of words. Classifier building is then performed to be used in the classification process. The next step is to clasify song lyrics on test data into a class topic by using multinomial naïve bayes method. The last step in this research is to evaluate system by calculating accuracy, precision, recall and f1-measure. Topic class in this system are love, friendship, nasionalism, family, religion, and negative content. The performance of this system in this study, the f1-measure values at 88.91% and accuracy at 96.00%.Keywords: Classification, Song lyrics, Chi square, Multinomial naïve bayes Abstrak Saat ini sistem pengkategorian lagu pada aplikasi pemutar musik seperti iTunes pada iPhone sudah mampu mengkategorikan lagu berdasarkan genre, artist, dan album. Namun belum terdapat pengkategorian lagu berdasarkan topik lagu, sehingga untuk mengkategorikan lagu berdasar topik ini masih dilakukan secara manual. Penelitian ini bertujuan untuk membuat sistem yang mampu mengkategorikan lagu berdasarkan topik lagu dengan menggunakan lirik lagu sebagai objek penelitian. Penelitian ini diawali dengan mengumpulkan lirik lagu dari berbagai website penyedia lirik lagu sebagai dataset penelitian, kemudian dilakukan preprocessing data yang terdiri dari case folding, tokenization, stopword removal dan stemming. Tahap selanjutnya yaitu melakukan feature selection dengan menggunakan metode chi-square. Selanjutnya hasil feature selection tersebut dilakukan feature extraction dengan menggunakan metode bag of words. Selanjutnya dilakukan pembangunan classifier yang nantinya digunakan pada saat proses klasifikasi. Tahap selanjutnya yaitu melakukan klasifikasi lirik lagu pada data testing kedalam sebuah kelas topik tertentu dengan menggunakan metode multinomial naïve bayes. Tahap terakhir pada penelitian ini yaitu mengevaluasi sistem dengan menghitung accuracy, precission, recall, dan f1-measure. Kelas topik lagu yang dihasilkan pada sistem ini yaitu cinta, persahabatan, nasionalisme, keluarga, religi dan konten negatif. Performansi sistem yang dihasilkan pada penelitian ini didapatkan nilai f1-measure sebesar 88.91% dan nilai akurasi 96.00%.

show abstract

“…5a) datée dans le dernier tiers du Ier siècle de notre ère avec une technique de qualité semblable à celle de la plateforme de la courtine 23 . Son installation semblerait changer la morphologie de la tour, la convertissant en une tour rectangulaire 24 , mais nous méconnaissons la hauteur totale de la plateforme, et il est probable que celle-ci n'atteigne pas la hauteur totale de la tour 25 , car le fait qu'une seule rangée s'adosse à la tour pourrait rendre fragile sa construction. La réforme de l'angle tour/ enceinte-plateforme marquerait l'un des derniers actes documentés, dédié à une réparation postérieure avec du matériel de qualités moindre par rapport aux structures antérieures.…”

Section: Introductionunclassified

Lorsque les murailles sont attaquées... par l’érosion. Quelques indices archéologiques sur les solutions apportées durant l’antiquité à Cordoue

Courault¹

2017

REUDAR

View full text Add to dashboard Cite

Erosion was a real problem for damage edifice in long term, and Roman people knew this fact, as Vitruve and Frontin ilustrated. There is not much investigation about erosion, due to the difficulties concerning the identification of the relationship between the erosion process and reparation acts during an emergency archaeological excavation. However, Cordoba presents particular interest within the investigation of its City Wall during the Antiquity. Certain aspects have not been considered yet by investigation, reason by which Cordoba offers some interesting clues to understand how Roman citizens protected their urbanism against erosion.

show abstract

Natural language processing of lyrics

Cited by 54 publications

References 6 publications

Multimodal Music Mood Classification Using Audio and Lyrics

Multimodal Music Mood Classification Using Audio and Lyrics

Klasifikasi Topik pada Lirik Lagu dengan Metode Multinomial Naïve Bayes

Lorsque les murailles sont attaquées... par l’érosion. Quelques indices archéologiques sur les solutions apportées durant l’antiquité à Cordoue

Contact Info

Product

Resources

About