I am finally in a place where I am enjoying the process of making my (massively over-complicated) five-minute video - I'm learning a lot because I've never done anything like this. Is it smart to do something completely new for the last assessed piece of work of the MA? I really don't think so, but I don't think anyone could talk me out of it now!
Using a sort of script-cum-planning document (I wonder if my use of the word 'cum' will get flagged up? BA cum louder!) seems to be working really well for me. I put a bunch of stuff into it fairly free-form - ideas for what I want the video to be, and not be, a list of all the art I've made, and bullet-points about what I want to say about my work... The last of these developed fairly naturally into a set of statements, which I didn't try to turn into a script immediately. I think the ability to step back a bit and allow myself to do stuff without trying to turn everything into a formal process is something I've really learnt from the MA - outcomes are important, but the process is important too, and often the process reshapes the outcomes, but only if you allow it to do so.
Anyhoo, allowing the different statements to just flow really helped a different voice to develop - a more authentic voice, talking about the emotional meaning behind my art, something I've always kept quite hidden - I guess a voice closer to the voice of this blog, rather than the slightly sardonic voice my art itself often projects. In fact, I realised that the material divided quite neatly into three voices - 1) an authoritative rather formal voice describing the power of grief to disrupt our assumptions about the world 2) a more gentle, personal voice describing my direct experience and 3) a narrative voice describe a direct experience of seeing someone clearly dying of a heart attack in a local supermarket.
I hit on the idea of using AI to narrate my video - this serves a few purposes - firstly I sort of hate my voice, secondly it creates emotional distance for me to allow myself to be more authentic (I can speak more authentically hiding behind an AI voice), and thirdly, I thought the idea of a rather robotic voice describing very harrowing events would be provocative at best, and entertaining at worst... I checked out the text-to-speech options in Runway ML and I was staggered by the quality. They actually completely failed my robot-talks-about-emotions idea by sounding amazingly realistic and animated. They actually read the text with the passion I hear in my head, but which somehow disappears when translated into my spoken voice - it's like an assistive technology for someone with a chronic lack of vocal inflection!
At this point something quite interesting happened... You can probably see what's coming (or indeed cumming) dear reader, but it caught me unawares... Runway ML offers a bunch of voices, so I tried a load to see which I liked best (note to self - having them read your own text is much more informative than listening to a canned sample). Of course I realised I didn't need to pick one... I could pick different speakers for different material! A rather brisk male British voice (called 'Tom', I shit you not) for the rather stiff description of the way grief affects your life trajectory, a wonderfully lilting and animated female voice to read my personal reflections with far more passion than my own voice would convey, and a whispering, slightly detached, american female voice for the description of the supermarket death. All three are the canine's testicles.
I generated each paragraph of speech separately, then used Shotcut to stitch them all together. So far the whole narrative runs to about 5 mins 30 sec, and that's with no breaks between speakers, which is jarring... Some sections of the audio were immediate candidates for removal e.g. the initial paragraph contained some swearing that might have worked in my own voice, but sounded really harsh and jarring in the AI voice.
So now I am starting to try and pair artworks and imagery to the different sections, with the aim of making a long version of the video which I can start to edit down to the five minutes...
No comments:
Post a Comment