Could someone read it to me? A human, please
I believe that A.I. can help in many fields. It should not be imposed everywhere
An afternoon after a working day. Usual browsing through newsletters and subscriptions. Sorting out, what to read immediately, what in the evening, what long reads to postpone till the weekend. A couple of articles feature audio: perfect, I’ll be able to set my eyes free from the screen, listen to while tidying up my desk and study. Who will be reading this piece? A voice I know and like, or someone who demands more focus and attention to understand a learned language from hearing?
A voice starts, clear as I would expect from a pronunciation dictionary. But the intonation sounds like an indiscriminately applied technique rather than an emotional bond stemming from the understanding of the text. The voice does not come out from a human: it is A.I. generated. Disillusioned, I switch off the audio. Most likely, I will not read the article, too.
I’ve been a listener rather than a watcher all my life. Whatever occupation allowed it, I had a radio on. I knew speakers’ voices rather than faces. Masterpieces of Polish literature, not always riveting when on school obligatory reading lists, sounded more compelling, when read by actors.
It took me long years to learn English well enough to unbind my sight from a written text and try to rely on hearing only; it still remains an ability to improve. In a similar way as decades ago with Polish radio, these days, too, I associate names of narrators with the kind of rendition to expect.
It was in 2018 when I came across a TED talk dated five years before, from a scientist, who worked on synthesizing voices for people who lost their ability to speak.1
The foundation idea of VocaliD2 was to record voice samples from people throughout the world, collect from a prospect patient whatever sound she or he could produce, and combine the two into a bespoke synthetic voice, suited well for that person and featuring an indelible mark of her or his personality.
It was an idea contrary to the trend we are witnessing today: how to add human components to a technology, instead of substituting human features with artificial imitations. It resonated with me so much, that I submitted a test audition, got encouraged despite being neither native nor even fluent speaker, and then recorded a portion of readings. Maybe someone in the world uses a synthetic voice which contains a sliver of sound from my speech.
During a decade, VocaliD has adapted A.I. technology and has became part of Veritone.3 I stopped recording, self-conscious about my flawed pronunciation, and didn’t follow the development of VocaliD. Looking at their website today, I believe that the tenets of the enterprise have remained the same: adding the human factor to a technology, not the other way round. I hope that A.I. helps in the engineering work and does not replace the source of sounds, which remains human.
As a mathematician and science presenter Hannah Fry proves in her book “Hello World. How to be Human in the Age of the Machine”4 (2019), algorithms do a good job when they help humans, and struggle, when they try to substitute us.
“There is the news of the world, and the news of the heart” sings Carrie Newcomer in her recent poem-song.5 I dare to paraphrase the line to: “There are tools to serve us, and there are pieces of art”. Whatever too complex, too laborious procedures, or too dangerous tasks A.I. can perform, let it do it:
Just leave the kingdom of art to humans.
There are tools to serve us, and there are pieces of art.
Just leave the kingdom of art to humans
What can we do, to resist the hubris and arrogance of all Googles and the likes, who try to supersede humans in thinking? We can adapt a version of boycott and divestment:
refuse to accept any algorithmic playlists,
cancel subscription to services which try to feed us with A.I. too blatantly
appreciate and promote human talents and skills
Related reading:
The founder and CEO of VocaliD, Rupal Patel, recalls her TED talk from a decade ago, in a post on Medium:
“Memories from ten years ago today…”
https://medium.com/@rupal_31149/memories-from-ten-years-ago-today-36cae3af7f4e
If you have any thoughts on the above, please leave a comment. Your remarks may improve my future writing.
Do you think some of your friends might like it? Feel free to share Eyeore Ponders with them:
Thanks for reading; until next time!
Press “like” if you will, it is the kindest expression of encouragement.
“Synthetic voices as unique as fingerprints”
https://www.ted.com/talks/rupal_patel_synthetic_voices_as_unique_as_fingerprints
VocaliD https://vocalid.ai/about-us/
Veritone https://www.veritone.com/
Hannah Fry “Hello World. How to be Human in the Age of the Machine”
https://hannahfry.co.uk/book/hello-world/
“A Great Wild Mercy” by Carrie Newcomer, 2023