AI Transcription Smack-Down: The Latest Round

After the massive media brouhaha over Rev.com lower its already skimpy payouts for transcriptionists, the company has announced its first rate increase since its founding 10 years ago: https://www.businessinsider.com/rev-petition-to-ceo-jason-chicola-demands-he-raise-wages-2019-11?r=US&IR=T?utm_source=copy-link&utm_medium=referral&utm_content=topbar.

But as someone who’s spent the last couple of years training our new robot overlords, I can point you to the bigger picture behind all this: AI-driven voice recognition systems are getting so good that’s it’s tough for mere mortals to compete. In fact Rev’s own Temi.com unit charges roughly 1/12 the rate for human-typed transcription: just 10 cents a minute versus $1.25 for Rev (which is still quite low by traditional industry standards).

Even Rev itself has been shifting to the editing model, which hands off computer-generated text to humans to clean up. But if you produce clear, easy to hear content, why would you pay 12 times more for humans to work on it all? What remains for the contractors is increasingly often almost indecipherable audio. So it doesn’t matter how you slice it, the cheese is moving.