=================================================================================================
UPDATE (22 Sep 2020)
[New!] Call for participation: Present your Findings of EMNLP paper in NLP4MusA
We would like to invite papers accepted in Findings of EMNLP to present their work at the ISMIR 2020 workshop NLP4MusA: Natural Language Processing for Music and Audio (https://sites.google.com/view/nlp4musa). Please submit title and abstract to https://easychair.org/conferences/?conf=nlp4musa by September 28 - Submissions will be lightly reviewed to determine fit to the workshop themes (see description below).
The presentation format will consist of a pre-recorded talk and two live slots for Q&A. Registration to the workshop will be free for both authors and attendees.
==================================================================================================
Workshop description
Natural Language Processing (NLP) research is typically associated with the advancement of language technologies. However, in today's growing digital landscape, where users engage with other users, products and service providers, NLP is becoming ubiquitous due to its ability to facilitate and personalize access to digital content of any nature. One case of particular relevance is the audio entertainment industry, as language is a key component of most audio content (e.g., songs, podcasts, radio shows, audio ads, etc.) as well as user-content interactions (e.g., textual queries, spoken utterances, social media activity, etc.). In this context, we propose the First Workshop on NLP for Music and Audio, a forum for bringing together academic and industrial scientists and stakeholders interested in exploring synergies between NLP and music and audio.
Topics of Interest
We welcome both academic and industry submissions at the crossroads of NLP and music and audio, including (but not limited to) topics such as:
NLP for Music and Audio Understanding
- Tagging and meta-tagging
- Knowledge graph construction
- Information extraction
- Named entity recognition and linking
- Speech to text
- Song and podcast segmentation
- Topic modeling
- Sentiment analysis
- Representation learning
- Bias in music and audio corpora
- Audio captioning
- Debiasing music entity embeddings
NLP for Music and Audio Retrieval and Personalization
- Conversational AI
- Slot filling and intent prediction
- Information retrieval
- Cross-modal retrieval
- Recommender systems
- Ads personalization
- Responsible recommendations
- Fairness and transparency
NLP for Music and Audio Generation
- Lyrics generation
- Music generation with language models
- Spoken audio/podcast generation
Submission Instructions
We accept extended abstracts of up to 2 pages or short-papers of up to 4 pages excluding references. Submissions should adhere to the ACL Anthology formatting guidelines.
Submission template: NLP4MusA-templates.zip
Only papers using the above template will be considered. Word templates will not be provided.
Papers should be submitted via EasyChair: https://easychair.org/conferences/?conf=nlp4musa
The review process will be single-blind. Accepted papers will be published at the ACL Anthology.
Double submission
To maximise the impact of work in the field of NLP for Music and Audio, we are open to the possibility of double submission, or submission of work which has been partially published elsewhere. Any double submission should however be reported to the programme committee at the time of submission.
Authors can submit works that were previously published on preprint websites like arXiv.org.
Important dates
Submission Deadline: July 24, 2020
Notification of Acceptance: August 14, 2020 September 1, 2020
Camera-Ready Deadline: September 11, 2020 September 25, 2020
Workshop Date: October 16, 2020
All deadlines are in Anywhere on Earth (AoE) timezone.
Contacts
All questions about submissions should be emailed to Sergio Oramas (soramas [at] pandora [dot] com) or Massimo Quadrana (mquadrana [at] pandora [dot] com)
Related Bibliography
The music and audio domains have sparked the NLP community’s interest in recent years. These domains pose specific challenges due to their particular register and jargon, for which current NLP pipelines are not particularly well adapted (for instance, current NER models require major domain adaptation to account for artist and song names [10]). Thus, a body of work has proposed to expand current NLP technologies to this domain, for example, by expanding coverage of entity recognition and linking systems to the music domain [10, 16], ad-hoc topic understanding [20], sentiment analysis [9], and representation learning, via either embedding musical entities [3], or modeling semantic relations between them using distributional models [1]. With these domain-specific models in place, downstream applications that stand out may include, e.g., the creation of music-specific knowledge graphs from text [11], explaining music recommendations in natural language [15], generating music reviews [17], audio captioning [6], genre translation [2], summarizing lyrics [4], or classifying music genres [12]. Conversely, NLP can contribute dramatically to the improvement of user-audio content matching and user experience through, for example: voice assistants [14] (with music as the most relevant use case), text retrieval systems [7], multimodal music recommender systems [13], podcast recommendations [19], playlists creation [8], ads targeting, and content generation [18, 5].
[1] J. Camacho-Collados, C. D. Bovi, L. E. Anke, S. Oramas, T. Pasini, E. Santus, V. Shwartz, R. Navigli, and H. Saggion. Semeval-2018 task 9: Hypernym discovery. In SemEval, 2018.
[2] E. V. Epure et al. Leveraging knowledge bases and parallel annotations for music genre translation. In ISMIR, 2019.
[3] L. Espinosa-Anke, S. Oramas, H. Saggion, and X. Serra. Elmdist: A vector space model with words and musicbrainz entities. In ESWC, 2017.
[4] M. Fell, E. Cabrio, M. Corazza, and F. Gandon. Song lyrics summarization inspired by audio thumbnailing. In RANLP, 2019.
[5] C.-Z. A. Huang, A. Vaswani, J. Uszkoreit, I. Simon, C. Hawthorne, N. Shazeer, A. M. Dai, M. D. Hoffman, M. Dinculescu, and D. Eck. Music transformer: Generating music with long-term structure. In ICLR, 2018.
[6] C. D. Kim, B. Kim, H. Lee, and G. Kim. Audiocaps: Generating captions for audios in the wild. In NAACL-HLT, 2019.
[7] M. Levy and M. Sandler. Music information retrieval using social tags and audio. IEEE Transactions on Multimedia, 11(3):383–395, 2009.
[8] B. McFee and G. R. Lanckriet. The natural language of playlists. In ISMIR, 2011.
[9] S. Oramas, L. Espinosa-Anke, F. Gomez, and X. Serra. Natural language processing for music knowledge discovery. Journal of New Music Research, 47(4):365–382, 2018.
[10] S. Oramas, L. Espinosa-Anke, M. Sordo, H. Saggion, and X. Serra. Elmd: An automatically generated entity linking gold standard dataset in the music domain. In LREC, 2016.
[11] S. Oramas, L. Espinosa-Anke, M. Sordo, H. Saggion, and X. Serra. Information extraction for knowledge base construction in the music domain. Data & Knowledge Engineering
[12] S. Oramas, O. Nieto, F. Barbieri, and X. Serra. Multi-label music genre classification from audio, text, and images using deep features. In ISMIR, 2017.
[13] S. Oramas, O. Nieto, M. Sordo, and X. Serra. A deep multimodal approach for cold-start music recommendation. In DLRS-ACM RecSys, 2017.
[14] A. Saade and et al. Spoken language understanding on the edge, 2018.
[15] M. Sordo, S. Oramas, and L. Espinosa-Anke. Extracting relations from unstructured text sources for music recommendation. In NLDB, 2015.
[16] R. Speck, M. Roder, S. Oramas, L. Espinosa-Anke, and A.-C. N. Ngomo. Open knowledge extraction challenge 2017. In ESCW, 2017.
[17] S. Tata and B. Di Eugenio. Generating fine-grained reviews of songs from album reviews. In ACL, 2010.
[18] K. Watanabe, Y. Matsubayashi, S. Fukayama, M. Goto, K. Inui, and T. Nakano. A melody-conditioned lyrics language model. In NAACLHLT, 2018.
[19] L. Yang, Y. Wang, D. Dunne, M. Sobolev, M. Naaman, and D. Estrin. More than just words: Modeling non-textual characteristics of podcasts. In WSDM, 2019.
[20] S. Zhang, R. Caro Repetto, and X. Serra. Understanding the expressive functions of jingju metrical patterns through lyrics text mining. In ISMIR, 2017.