Figure - available from: Applied Sciences
This content is subject to copyright.
In a simplified example, a “universe” of three chords (equivalent to groups in the presented approach) is considered. T includes the transitions of Song B, which incorporate only chords 0 and 1, and transitions are only possible from one to the other (not to themselves). The “support” space includes all three chords, and transitions are possible between all chords (S also includes the probabilities of T). Some melodic segments are compatible with some chords at time step t+i (as in ON×K); compatible chord-melody matches are indicated in grey, incompatible in white. In case a single support transition probability is used, S is only employed to facilitate a transition between chords 0 and 1, which is not possible in T; all other paths will be eventually erased by backtracking. If double (or multiple consecutive) supports are employed, then all supports, except from the final in the series, could possibly incorporate chord 2, which is not available in Song B.
Source publication
This paper presents a methodology for generating cross-harmonizations of jazz standards, i.e., for harmonizing the melody of a jazz standard (Song A) with the harmonic context of another (Song B). Specifically, the melody of Song A, along with the chords that start and end its sections (chord constraints), are used as a basis for generating new har...
Similar publications
The intersection between poetry and music provides an interesting case for computational creativity, yet remains relatively unexplored. This paper explores the integration of poetry and music through the lens of beat patterns, investigating whether a byte-based language model can generate words that fit specific beat patterns within the context of...
The dead internet theory is a conspiracy theory that states that all interactions and posts on social media are no longer being made by real people, but rather by autonomous bots. While the theory is obviously not true, an increasing amount of posts on social media have been made by bots optimised to gain followers and drive engagement on social me...
This work discusses the use of machine learning for extending the boundaries of architectural concepts. The use of deep learning algorithms is exploited for transferring stylistic attributes from one visual (architectural) domain to another (architectural or otherwise). The combination of various semantic features into hybrid results helps locate p...
The pervasive adoption and use of Generative AI tools prompted a discourse on determining the optimal balance between automa- tion and human agency in processes executed by users with the assistance of this technology. This paper presents a user-centered co-creativity framework that identifies key dimensions responsible for the modulation of agency...
Being able to evaluate aesthetics automatically is one of the fundamental needs for creating robust and autonomous computational creativity systems. In Graphic Design (GD), many aesthetic features might need to be considered simultaneously to properly evaluate GD artefacts, e.g. their visual relation to the concept of the work, legibility, innovati...
Citations
... This generative combined approach has led to the development of the CHAMELEON melodic harmonization assistant, which has been evaluated by composers of different levels, as an assistive tool for making melodic harmonizations, [33]. Except for that, a similar approach has been examined for the cross-harmonization of jazz standards, [34], where the melody of one song is harmonized with the harmonic space of another song. Similar approaches have been examined for melody and drums rhythm, generation. ...
... melody of one song reharmonized based on the harmony of another), indicated that there are still details in the generative part (i.e. the part that applies chords to the given melody) that need to be corrected. Figure 10 (taken from [34]) shows that in some cross-harmonizations, some produced chords harmonize melodic segments that include notes that are "incompatible" with the chord; identifying those incompatibilities, however, would require the formulation of extensive sets of rules for each chord. It is a future challenge to examine how deep learning models would help toward implicit learning of such rules from data. ...
The Department of Music Technology and Acoustics of the Hellenic Mediterranean University offers a unique higher education program in Greece, addressing the growing demand for specialists in music technology, sound technology, and acoustics. It aims to educate specialized professionals in the rapidly advancing scientific fields of music technology and acoustics, mainly driven by the swift progress in electronic technology. The Department aims to address a gap in the professional market by producing highly skilled graduates, capable not only of keeping up with the latest scientific and technological developments but also of leading the way by introducing innovative approaches and methods. The Department combines art, science, and technology, focusing on sound recording, analysis, synthesis, and music production. Music technology encompasses various cutting-edge fields such as network music performance, artificial intelligence in music, and music embodiment. Acoustics refers to fundamental aspects of sound as well as its generation, transmission, and related phenomena. It includes research fields such as physical acoustics, optoacoustics, and vibroacoustics. This overview presents the research activities, methodologies, and results. A discussion of future research works and pointers to future technological evolution towards real-world music and acoustics applications is also provided.
... When user clicks the 'Suggest' button of Fig. 6a after clicking on a chord, the request sends the currently displayed kern along with the index of the chord that needs to be replaced with automatic suggestion. In this case, the reharmonization server calls pretrained jazz standard-focused version the Chameleon melodic harmonization assistant [13], which substitutes the userspecified chord with a new chord, based on probabilistic inference. The harmonic and structural information of the piece in the initial request now includes a new chord in the place of the old one. ...
... The intersection of quantum computing and music generation was explored by the authors of [44], who delved into the use of partitioned quantum cellular automata for the creation of original music compositions, highlighting the potential impact of emerging quantum computing technology on the music industry. In the domain of music harmonization, the work reported in [45] introduced an innovative methodology for generating new harmonizations of jazz standards by combining melodies and chords from different songs. These studies demonstrate the potential of computational creativity and music education in diverse musical applications. ...
The musical key serves as a crucial element in a piece, offering vital insights into the tonal center, harmonic structure, and chord progressions while enabling tasks such as transposition and arrangement. Moreover, accurate key estimation finds practical applications in music recommendation systems and automatic music transcription, making it relevant across academic and industrial domains. This paper presents a comprehensive comparison between standard deep learning architectures and emerging vision transformers, leveraging their success in various domains. We evaluate their performance on a specific subset of the GTZAN dataset, analyzing six different deep learning models. Our results demonstrate that DenseNet, a conventional deep learning architecture, achieves remarkable accuracy of 91.64%, outperforming vision transformers. However, we delve deeper into the analysis to shed light on the temporal characteristics of each deep learning model. Notably, the vision transformer and SWIN transformer exhibit a slight decrease in overall performance (1.82% and 2.29%, respectively), yet they demonstrate superior performance in temporal metrics compared to the DenseNet architecture. The significance of our findings lies in their contribution to the field of musical key estimation, where accurate and efficient algorithms play a pivotal role. By examining the strengths and weaknesses of deep learning architectures and vision transformers, we can gain valuable insights for practical implementations, particularly in music recommendation systems and automatic music transcription. Our research provides a foundation for future advancements and encourages further exploration in this area.
Current computer tools aimed at the process of automatically harmonizing melodies have been reported here. These tools were divided into two lines of description: Plug-ins and Programs. It was observed that Plug-Ins are small lines of instruction inserted inside software that have more generic functions, while Programs are more robust lines of instruction that are designed solely for the purpose of enabling the harmonic automation of melodies. In order to carry out the research, searches were made using the keyword "automatic harmoni-zation" on platforms, such as ResearchGate (R G), SciELO (Scientific Electronic Library Online), Academia.edu, Google Scholar and other databases. After selecting the most relevant computational harmonization tools, it was found that the results obtained were derived from different mathematical methods. In order to verify the efficiency of an automatic process of accessible harmonizers, a manually harmonized authorial melody was used to compare the subsequent computer harmonization processes performed by the internal Sibelius Plug-ins and Band-in-a-box. It was found that all of them can generate convincing harmonies that give different meanings to the melody. The results were briefly discussed , so that they could serve as a reference for composers and arrangers interested in the subject.
