When it came to the production of the creative work I knew that I wanted to use my voice as the main source of sound in the piece, using the RAVE models to then manipulate and orchestrate it. I wanted to include vocals to have this contrast of human and machine, on one side is this completely natural element with no integration of synthetic noise, and on the other is an artificially machine-made ‘voice’ adding the unnatural element to the work.
When beginning the process of making this piece, I definitely struggled at first to find the direction I wanted to take the work in. I was confident in using my voice but when I initially tried to record, I felt lost and unsure on how to use RAVE alongside my voice effectively. This led to prioritising the other aspects of the project over the composition and dedicating more of my attention to the written work. This is a process that is common for me, I usually complete the practical aspect once the writing is complete as, I feel the knowledge I learn from the research is needed to provide context to the practical piece. Although this structure does work well for me, in this case I think I should have provided more time to focus on the sound work.
One problem that repeatedly occurred during this project was Logic crashing due to the processing size needed for RAVE to run. I had multiple tracks using RAVE models at the same time, and even when increasing the buffer size on Logic, it would still overload the system and constantly crash. This became very annoying and off-putting to using RAVE. In order to combat this issue I realised I needed to use RAVE in increments, muting it on playback, and turning down the volume of every track that had it on when not playing, this also eliminated any latent noise that the models were outputting. This did help, however did not solve the issue entirely, and I still faced many issues with a ‘system overload’, having to close Logic and reopen it many times to restart it. In the end I was able to use RAVE on all the tracks I wanted, producing the desired effect. However, in order to export it I had to bounce it offline as when trying to bounce in realtime it started to make some glitchy sounds that were not in the piece – this was from the overload of RAVE.
Next time, in order to combat this issue, I would try and use a computer that has more processing power, defining what it is I want to make from an early stage, and how to best use the abilities of theAI models.
All that being said, I was actually quite pleased with the final outcome of the practical work. I kept to my original plan of using only my voice as the input, with no other instruments or synths being used to create the audio. I felt very happy about this aspect as like I said, I initially felt unsure of how to execute the unaccompanied use of my voice.
I used the QuickSampler in Logic to input my voice and delegate keys to certain letters, words and phrases, meaning I was able to form sentences using the keyboard. Using a sampler within a DAW is something I am previously familiar with as I find working with samples in this way inspiring. However, using my own voice and slicing the the sample up to this extent was something new to me. This whole process of using just my voice was very interesting as I had to break down the understandings I had on making a composition and look at the production from a different perspective. I am very happy with the outcome as it is very different to previous works I have made using samplers, and it has inspired me to continue creating, using both my voice and other RAVE models.
Taking inspiration from the artists I had researched, like Pamela Z, Herdon, and Dryhurst, whose work uses voice through the manipulation of technology, I feel this piece reflects the work of my essay well. It includes the elemental aspects of human and machine, whilst being an ethically viable use of AI. Both Pamela and Herdon use voice to build a united relationship between human and machine and this is what drew me to their work. Therefore, I feel it is right to have made a piece of work that does the same, from my own interpretation of the subject matter.
I used the RAVE models to create a kind of background atmosphere to the piece, and in fact, without RAVE this work would be rather empty. But that was the whole point, I viewed the noises the models produced as the other instrument of this work, accompanying my voice and at times taking on the voice.
The control I had over the models was an interesting aspect of this project, through inputting my voice and having the ability to manage certain parameters like the wet to dry signal, I was able to manipulate the movements of the sounds. However, the specific noises that the models created was out of my control, deriving from the data set in a semi randomised format. This made it fun to work with as it was like having another entirely independent input, except one that I did have some levels of control over. This did however also at times become tricky as I wanted the models to produce certain sounds that it just wouldn’t, I then realised I had to let go of the orchestrating aspect to the AI and let it run freely, which in turn ended up sounding very intriguing, as I made something I couldn’t.
Overall, I am pleased with the final outcome of this project, I feel both element 1 + 2 are cohesive with each other and give context the thematic points of the work. The use of RAVE has inspired me to continue making work with AI in ethical ways, and I now want to build my own model through knowing how they operate in a DAW. I feel I have made a sonically deep piece of work that explores both the themes of my research as well as my personal interests, making this specific practice influential in where I plan to take my work next.