is there more to say? (2023)
This project is a collection of poetry generated from my voice library.
I made it with SpeechBrain when I was playing around with their ASR encoders and decoders. I started to get tired of the run-of-the-mill tests I was doing, and started to wonder how well the ASR would function when I ran through some of my not-so-word-y vocalisations.
Here’s an example of an audio clip I ran through the code, featuring me burping (y’know, just singer things...):
And here is what the ASR understood from my burp:
NKS PSYSSE SSSESSIBS
And another clip, this time of me crackling some sound and forcing air back through my nose (I know it sounds like a fart, but I promise it’s not!):
And here is what the ASR understood from my non-fart-sounding nose crackle:
THE ROYAL AWE CLAUDER COME ALL ALL ALI PRAGUE
So naturally, with results like this, I decided to run the entirety of my voice library through my code. All 1,357 clips. Of course, some clips didn’t parse (Still investigating why) and I had a failure rate of 11.28% (Why so high? I’m still investigating...). BUT! 1,194 clips parsed. So that means I have 1,194 poems to share. I’m going to divide everything into 3 volumes and I’m experimenting with ways of collating this into a tactile, tangible format for others to experience. So stay tuned!