Category: CONTEMPORARY ISSUES IN SOUND ARTS

RAVE is an audio processing, generative tool, designed by Antoine Caillon and the ACIDS research group at IRCAM. RAVE (Realtime Audio Variational autoEncoder) is a learning framework for generating a neural network model from audio data.

As an encoder, it is able to take sound as an input, and generate new sound as an output – it translates incoming audio into a set of synthesis parameters that are used to generate back the sound.

This is based on two separate processes:

• Encoding process : where a given parameter of inputed audio is transformed into a set a latent variables (128 parameters in general).

• Decoding process: inverts these 128 latent variables back into sound.

As each model is trained on a limited set of data, it will try to extract the parameters even if the input sound does not match the original database. This is why RAVE is able to perform timbre transfer. For example, if RAVE has been trained on piano sounds, and is then given a violin sound, it will try to extract synthesis parameters from it and generate it as a piano sound. This allows you to use RAVE as an audio effect through transforming incoming audio, but also as a synthesiser – by directly controlling these latent parameters.

I discovered RAVE through Michael at the LCC Creative Technology Hub and as soon as he mentioned it I recognised the importance of RAVE for myself in this project. It, itself is an AI tool that artists are using to create generative sonic work, with a strong community surrounding the tool, this is a great example of subcultures forming around AI.

RAVE can be seen as a more ethical AI alternative because it’s architectural design, giving you, the user, control over the data being inputed for the generative output. Artists providing sources are able to consent to the use of their audio, therefore if you train it on your own recordings or licensed datasets, it avoids the ‘theft’ concerns often associated with AI tools using the internet.

Because it is used more as a creative instrument rather than a replacement for human labor, the perspective of it being an AI tool is different to, say, how large corporations use AI as a tool.

RAVE’s ability to ‘timber transfer’ allows it to be used in a way which many see as a natural progression of digital music.

Like all deep learning models, training RAVE requires significant computational power, contributing to carbon emissions and the use of water – a common environmental concern for all generative AI. It is hard to say if or how this would change as I feel I currently do not know enough about the environmental impact of these supercomputers which are powering AI. I fully comprehend and empathise with the refusal to use AI due to the environmental damage it has. However, there are many contributing factors to this destruction of nature, which, in contemporary society is hard to control due to the involvement of advanced technology in our everyday lives. This is an aspect of AI that contributes hugely to the mass rejection and lack of acceptance to the use of AI in art, therefore is a key aspect that I will be researching for this essay.

RAVE is an facet of AI that I am able to cite for both elements of this project, the essay as well as the composition. It is a key development in my research for understanding the core of my essay, whilst also providing me with an inspiring, new tool which I can use to create my composition work.

With the work I have been doing this past year, (involving Pure data) I have learnt a vast amount as it was an aspect of sound arts I had no prior knowledge of, and so I have built my work from the ground up in that respect. However I found I got to this point with my work, where, I didn’t quite know which direction to take it in. There are a number of routes I am able to take with the knowledge I now acquire, and I feel using RAVE is one I feel motivated to continue in.

I feel it is important to understand the importance of AI in the contemporary artistic world as it is not leaving anytime soon, in fact the opposite. AI will continue to evolve at rapid speeds and in ways that get more complex, therefore being able to comprehend it now feels significant.

With the coding knowledge I have picked up from using Pure data (although tiny compared to others) I feel confident in continuing to learn and use code in a creative way. Inputting my own entire database of audio into RAVE to then use to create work feels very exciting to me, and is a continuous project that will really allow me to understand both RAVE itself and how AI learns and uses sources to create generative sounds.

Some examples of artists using RAVE:

Mouja: https://youtu.be/4qbK3cw3E5M?si=ef7GszYk2-AMbBYi

Tenor Boston 2023 – Concert: https://youtu.be/kuxYIYgPrTs?si=_SLhYMd4lHSRHuGC

The concept of a sonic ‘failure’ stems from the mainstream rejection of radical (of the time) sonic concepts, often occurring when creative innovations clash with commercial demands for consistency, engagement, and traditional structures. With money being at the forefront of the corporate music and arts industry, progressive concepts often get rapidly declined due to the uncertainty of their profit. There is a specific mould in which people know will make money, therefore steering away from this formation poses risk, and the mainstream isn’t risky, it is predictable, expected and formulaic. (Most of them time.)

Often times, when rejected by the mainstream world, such concepts are adopted by subcultures, forming a strong community around the discovery, and enhancing the emotional depth of the work. Subcultures are key in art and music, they supply a non-restrictive space for deeper development, allowing concepts to continue to grow in a very ambitious way. They allow for people to put passion over profit. Unlike the corporate view, seeing the work as a commodity, stripping the emotional context away from the art and the artist, leaving only a ‘product’ left.

This tendency of corporate rejection and underground adoption is continuous throughout history, and often times follows a very similar pattern, repeatedly ending in the reabsorption into mainstream media. This reabsorption occurs due to the popularity that surrounds the once ‘radical’ concept. The community formed, create new, innovative works which begin to attract a wider audience due to their experimental, unheard, and imaginative features. This popularity attracts establishments who recognise they can benefit finically from this public engagement. Frequently using rather unethical methods in order to brand the art.

There are many different examples of this pattern occurring, both prior to and in the contemporary world , sonic examples include:

Luigi Russolo’s ‘Intonarumori‘.

(“noise-intoners”)

Russolo was a key figure in the Italian Futurist movement, a patron for noise, he believed the industrial revolution changed the human ear. For Russolo, the traditional orchestra was a limiting sound, he argued that a modern soul required modern sounds and therefore the rigid definitions of music needed expanding and widening.

In 1913, challenging this definition, Russolo introduced ‘intonarumori’, a hand cranked, acoustic noise maker designed to hiss, howl, and rumble, replicating the industrial sounds of the time. Made up of large wooden boxes with a metal speaker horn on the front. Inside, a complex mechanical system of wheels, strings, and membranes producing sound when a hand crank was turned or a lever moved. There were 27 different sounding machines all together, forming their own complex orchestra. Made to represent the soundscape of modernity.

When exhibited to the public at his 1914 Milan debut of ‘intonarumori’, Russolo was confronted with physical hostility from the audience. To listeners raised on the elegance of opera, used to the sounds of Verdi or Puccini, it felt like an auditory assault, a personal insult on the culture of orchestra itself. It was entirely new perspective on performance, something people were not used to and certainly not comfortable in accepting.

This concept of noise was radical for the time and it is artist like Russolo who completely changed the evolution of sound. He impacted the modern views

Facing this reaction from a mainstream audience, Russolo realised the incapacity of fully conveying the revolutionary, transformative essence of his “art of noises” theory. However he didn’t stop and his work was accepted and welcomed with great admiration from people and artists who were able to understand the importance and relevance of this change in tradition.

‘Intonarumori’ wad destroyed in WWII, however replicas have since been produced, for example: https://youtu.be/BYPXAo1cOA4?si=7EIcGcSffRtCwpYY

Roland TR- 808 Drum Machine.

A more modern example of this ‘sonic rejection’ is the Roland TR- 808 drum machine. Launched in 1980, introduced as ‘realistic-sounding drums’

Prior to the release of the 808, earlier in the year the Linn LM-1 was released, a drum machine which used high-fidelity digital samples of real drums. Corporate studio musicians became accustomed to using a ‘realistic’ sounding machine.

This is where the commercial downfall of the 808 lay. The memory chips needed to play back sampled recordings of real drum sounds were too expensive, and so the engineers of the 808 used analog synthesis to recreate the sounds. Although a simple choice born of necessity, choosing analog over digital technology was the aspect that made the profound impact on music that 808 eventually had.

Due to Rolands choice to use analogue synthesis, every sound was therefore generated by an electrical circuit, making the audio sound thin and synthetic to artists who were used to the realistic sound of a drum kit. The 808 struggled to find its place in the mainstream music establishment, causing issues that producers and musicians didn’t want to deal with, eventually just disregarding it as a ‘worthy’ machine.

The 808 was discontinued in 1983 after only 12,000 units were made.

Due to the commercial failure of the machine, second hand shops and pawn shops became flooded with them, selling them for a fraction of the original price. They then became available to an entirely new audience, who previously, weren’t able to afford the product.

Early hip hop is what made the 808’s name.

Afrika Bambaataa and the Soulsonic Force used the 808 beat in ‘Planet Rock’, this track is is widely considered the first major hip-hop/electro hit to use it, popularising it as a staple of the genre. https://youtu.be/9J3lwZjHenA?si=bB51TRpatbsmhA7t

The 808 was adopted by subcultures of artists who liked the distinct sound of the machine, seeing it as an instrument that wasn’t trying to replicate the realistic sounds of a drum kit, but instead they appreciated the synthetic, robotic and futuristic noises it produced. Musical subcultures in areas of the US such as; the Bronx, Detroit & Miami discovered that the 808’s kick drum was a pure sine wave, meaning, unlike a real kick drum, you could sustain the decay, turning a drum hit into a bass note.

‘Underground’ artists used the 808 in ways it was originally not intended for, however created incredibly inspiring, influential work that changed the sound of genres. So although seen by Roland and the mainstream music industry as a complete failure, people do what they do best and used their creativity to take an existing concept and produce an entirely new perspective of it.

The community that formed around the 808 continued to grow due to the popularity of the genres it became used in. This resulted in the reintegration of the 808 into corporate music. 40 years later, every DAW is pre-loaded with an 808 sound bank.