Working in ambient voice technology for some time, I’ve noticed a few ironies and foibles of working in voice…
- You get many unintentional but correct false triggers. Seems contradictory? If you’re saying “Alexa” a lot while on the phone, or “Google”, when you’re discussing Alexa Voice Service, it can get annoying when your Echo responds. The Echo is working properly even if you didn’t mean to trigger.
- It takes longer to explain voice interaction than it does to demonstrate voice interaction. Sure, there’s an easy-out to explain to fellow Trekkies… “it’s like the Star Trek computer”. But how do you explain the Star Trek computer?
- Demos rarely work. If the demo works, it’s because hundreds of demos that you tried before that didn’t.
- You need to act humble that you knew voice was going to be the next big interface for a long time.
- For voice, good enough means perfect. Having a conversation with a device that provides 80% accuracy is awful.
- You’re working on the cutting edge. There are a lot of solutions that are half baked and the real challenge is taking them from prototype to MVP to reliable product.