I’m facing a loss of the transcription gigs I’d been using to pay the bills. It’ll likely have vanished entirely by mid-2018 and quite possibly before that. The AI breakthroughs in intelligent speech recognition have occurred sooner than anticipated; Microsoft’s R&D division has come up with an AI capable of understanding conversational context and automatically transcribing large volumes of audio with an error rate equivalent to the average human transcriber.
The best human audio transcribers are able to maintain error rates just below 4% given typical batches of audio – with overlapping conversations, and faint background conversations, muddy static or sudden jarring noises in the mic audio, and really obscure jargon, comprising most of the mistakes made by transcriptionists, and this newly announced program (a new form of Cortana) will have a consistent 5.1% error rate with similar audio. It’s close enough that it’s not worth paying humans to do the work anymore; the AI can do the task almost as well as any human can.
What this means for me: I was pulling the bulk – roughly 55% – of my income, from transcription gigs. Other scattered chunks of revenue came from unrelated freelance work (10%) and 35% from online sales of art products. As for where I was spending? 45% of my spending in the past year was tied to ramping up that 35% of my income. Repeat: The art sales were losing money, as in: 20%+ loss margin when the low average sale prices, the shipping, eBay fees, mailers, and other materials, were all factored in.
So basically: My effective total cash flow will collapse soon. And I was making well below $10k/year to start out with.
I need to figure out a promising new business option soon, ideally by November, or this web network and my creative work in general, is at risk of going under entirely. What I’m hoping for is consistently available work in the $2.50/hr range or better. But realistically I will settle for half that much.