Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - I2C8 (Inspiring to Create)

Teaser

The I2C8 project started from the successful research carried out during the FP7-ICT project Lrn2Cre8 under grant number 610859. It brings parts of that research closer to market by exploiting results related to creativity and automatic music generation technologies. I2C8 has...

Summary

The I2C8 project started from the successful research carried out during the FP7-ICT project Lrn2Cre8 under grant number 610859. It brings parts of that research closer to market by exploiting results related to creativity and automatic music generation technologies. I2C8 has dealt with the development of a new product for music chord generation with intelligent and automatic features, aimed at the music production market.

The Lrn2Cre8 FET base project led to new ways to generate coherent chord sequences based on machine learning and musical semiotics. I2C8 is a spinoff project derived from Lrn2Cre8 dealing with the theory and implementation of a digital audio workstation plug-in for chord sequence generation. The plug-in is intended to encourage and inspire composers to explore loops through pattern definition, position locking, and generation into unlocked positions. Degrees of “diatonicity” of generated sequences can be explored and parameters for voicing the sequences manipulated.

The method implemented by the plug-in clearly separates it from other tools that are already on the market. Most of these tools could be described as chord recommendation engines or chord sequence builders, whereas we generate a set of full chord sequences that satisfy a given structural pattern. The USP of our approach is (1) the use of a machine-learning model as the foundation for the chord generation and (2) user-communication through an idiosyncratic pattern language conveyed by the user interface — not through more or less explicit implementations of music theory. Simplicity and ease of use are paramount. I2C8 was designed with quick results in mind. Our aim is to “inspire” for composing, not to deliver complete compositions.

We are not aware of another application/plug-in which uses a UI concept like I2C8. For the structural representation of chord sequences, a simple graphical, idiosyncratic pattern language was used which allows for results to be immediately viewed and listened to. This fostered an explorative and playful approach when looking for and finding new and interesting chord progressions.

We regard I2C8 as being a commendable result of applied research with the ultimate goal of successful commercialization. The principal requirements lie in the workflow optimization and not in the automated creation of complex musical patterns. I2C8 is a combination of a powerful music engine reduced to useful functionalities, with regard to specific workflow problems. Product-readiness of I2C8 could be reached within the funding period by the EU. However, it is a work in progress as ongoing user testing has been producing valuable feedback for further advancements.

I2C8 does not claim to generate long and musically complex chord progressions but predominantly short sequences of up to eight chords, or sequences within longer progressions. So the software caters to producers of loop-based, electronic genres for which the creative process is different in comparison to the ones for more traditional forms of music.

For the immediate next product iteration, we are currently evaluating the concept of extending I2C8 into a powerful arpeggiator by a flexible rhythm logic building on presets of a magnitude of musical styles. Creating arpeggiated chord progressions, i.e. successions of many notes in musically logical order, is a tedious and time-consuming process. The extended version of I2C8 could drastically simplify and shortcut laborious music production steps. This is a clear workflow benefit in addition to the assistance in the creative idea-finding process.

Considering the resources provided, we believe to have reached the utmost positive result with the I2C8 project in terms of technology transfer from an abstract, fundamental method to a tangible application, product-readiness of the software, and efficiency of project execution. With the right market preparation work, the I2C8 software is an application with commercial poten

Work performed

The project was subdivided into three focus areas:

1. Management

2. Product Development: idea finding, theory building, experimentation, technology integration, prototype design, application development, GUI design, as well as in-house and public testing, update of website backend systems (for payment, licensing management, customer support, user management, etc.)

3. Dissemination: product release and market push, user engagement and communication (including other software producers, experts, bloggers, youtubers, press and music producers), business plan update.

The I2C8 software product is currently undergoing late beta testing and is being prepared for the impending market launch (6/2018).

Final results

I2C8 makes (1) use of a machine-learning model as the foundation of the chord generation and (2) applies user communication through an idiosyncratic pattern language conveyed by the user interface. I2C8 generates chord progressions based on a specification of how long such a sequence should be and where a chord should be repeated.

In contrast, similar chord generation software tools are more or less explicit implementations of music theory. Such products concentrate on providing a corpus of musically proven chord successions, while I2C8 generates chord sequences from a statistical model which always contains a bit of randomness and unpredictability, with sometimes surprising results, i.e. simulating creative inspiration.

We regard I2C8 as a prototype implementation of the concept of “computer-supported creativity” — which aims to assist users in creative processes and places the human mind in the loop — in contrast to “computational creativity” which aims to imitate — and ultimately replace — human creativity. So by this definition, I2C8 serves as an assistant in creative processes.

The novelty of the approach in the use of semiotic variables to guide chord sequence generation into cyclic patterns. Statistical modeling with random sampling of chord sequences often leads to unexpected and therefore more interesting results, which has the potential to inspire users. Machine learning-driven music generation in general could help in the production of novel musical phrases which could become “meaningful” or “iconic” after being deployed in the context of an artistic work. Meaning is a highly socio-cultural artifact. In the context of music, it emerges during the creation and the reception of music. For I2C8, we capture existing meaning in the form of a statistical model and combine it with constraints that can be defined by the composer in order to recommend potentially interesting chord sequences. However, the decision on which chord sequence is used in a given musical context is left to the composer. I2C8 could therefore be of good help in creating (more) meaningful musical phrases.

We regard the question of the dichotomy between “computer-supported creativity” vs. “computational creativity” as significant. The difference between these two ideas may appear like a triviality, but as a matter of fact it really reflects one of the main questions we are faced with in the field of Artificial Intelligence: Do we want computers to fully replace people’s work, or do we rather want them to help and assist us?

I2C8 utilizes AI methods to provide full autonomy over creative processes while augmenting a user’s scope of action within them. The human remains in the driver’s seat in the decision-making process.

Website & more info

More info: https://www.i2c8.eu/.