Computer Science > Sound
[Submitted on 30 Dec 2023]
Title:AI and Tempo Estimation: A Review
View PDFAbstract:The author's goal in this paper is to explore how artificial intelligence (AI) has been utilised to inform our understanding of and ability to estimate at scale a critical aspect of musical creativity - musical tempo. The central importance of tempo to musical creativity can be seen in how it is used to express specific emotions (Eerola and Vuoskoski 2013), suggest particular musical styles (Li and Chan 2011), influence perception of expression (Webster and Weir 2005) and mediate the urge to move one's body in time to the music (Burger et al. 2014). Traditional tempo estimation methods typically detect signal periodicities that reflect the underlying rhythmic structure of the music, often using some form of autocorrelation of the amplitude envelope (Lartillot and Toiviainen 2007). Recently, AI-based methods utilising convolutional or recurrent neural networks (CNNs, RNNs) on spectral representations of the audio signal have enjoyed significant improvements in accuracy (Aarabi and Peeters 2022). Common AI-based techniques include those based on probability (e.g., Bayesian approaches, hidden Markov models (HMM)), classification and statistical learning (e.g., support vector machines (SVM)), and artificial neural networks (ANNs) (e.g., self-organising maps (SOMs), CNNs, RNNs, deep learning (DL)). The aim here is to provide an overview of some of the more common AI-based tempo estimation algorithms and to shine a light on notable benefits and potential drawbacks of each. Limitations of AI in this field in general are also considered, as is the capacity for such methods to account for idiosyncrasies inherent in tempo perception, i.e., how well AI-based approaches are able to think and act like humans.
Current browse context:
cs.SD
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.