Hacker News Reader

Comment by MSM | Hacker News Reader

A little bit of a side story about something that really opened my eyes last week.

I ordered a package that was shipping through UPS and when I called I was dismayed that there wasn't a way to enter my tracking number (It was 15ish alpha-numeric characters) besides speaking. So I thought "Okay, AAyyy--Fiivvee--Geeee..." I was about half way through slowly saying the tracking number when I was interrupted "Package not found, try again?"... Okay. "I must have to rattle this off pretty quick" I thought, so I blew threw it. It immediately found my package, told me where it was and when it was expected to arrive.

It was amazing. Voice recognition was good enough that UPS deems it fail-safe enough to have it be the only way to find a package by phone. That was the first time I really thought there might actually be a chance for an almost fully speech operated phone without a traditional UI.

View on HN

In the jet-pack future I've always hoped that there was some way to build a side-channel on voice to send over some kind of data so there's something you can do with your phone other than talk into it and push DTMF tones.

There have been experiments with throwing side-band text into the phone for a while now, but they never amounted to much. Nortel experimented with this in the 1990s before IP telephony took off but because it used the call-display encoding mechanism, it was slower than dirt.

I had to use a similar system from Amtrak that could barely understand a 6 digit ID number, though it was a number that had 3's, B's and E's in it. When it supposedly understood the number, it for some reason tossed me over to an operator to finish the process, and the operator had me read the number to them again. Either UPS has a far better system, or you got lucky with a tracking code that had very distinctive letters.

As someone who doesn't have a very good english accent I find the speech recognition isn't that good. People understand me just fine, but sometimes, no matter how much effort I put into it, the software will not.

Most of them fall back on a human operator after 5+ tries ;-) Yeah that's how I know... :(

I can't personally say I feel your pain; however, my mother does. She has a thick middle-eastern accent. It's one of the funniest things seeing her try to request directions by voice on her Nexus.

One tip that might help her is to do an exaggerated impression of an american accent when speaking to her phone.

When using my nexus I regularly mispronounce words and names so that the voice recognition can make sense of it.

Thanks that's helpful, I'll pass it on.

I've read a lot of sci-fi that had a dermal mic taped to your neck, and you kinda whisper with your mouth shut to control a computer.

Looks like we might nearly have that tech?

Not sure what you're referring to but it sounds cool.

Good story albiet, quite off topic :)

Reply on news.ycombinator.com