Google Voice's Speech-To-Text is Really Speech-to-Garble

my avatar

I'm Brad Cooper, a user experience designer and front-end programmer with a passion for realizing ideas both visually and technically.

bradleyjcooper@gmail.com

(856) 316-7128

Over this past weekend, my GrandCentral account was finally upgraded to Google Voice. I had been anxiously awaiting for the integration to Google – I’d been a user for a long time before they were acquired and I was getting a bit worried that nothing would come of the acquisition.

GrandCentral (now Google Voice) is a great service with some really impressive features

I’ve always really liked GrandCentral and have found it pretty useful – the ability to screen messages as they’re being recorded and the fact that I can have different voicemail messages for different ‘groups’ of contacts comes in really handy. For those of you that aren’t aware of what it has to offer you should look over the features.

What’s new in Voice?

Now that it is officially Google Voice, there are a number of new features that I’m going to like.

The first is SMS Messaging. Text Messaging through the browser has been pretty lacking – there were some good tools to send them, but none that I found to receive replies back. I spend part of my day attached to a computer and the rest of the day attached to my phone. I don’t want to have to pick up the phone to do a text message when I’m already on a computer… this solves that.

The second is the transcription of voicemails. I’m a big fan of speech-to-text technology (I’m hoping that one day most of my blog posts could be started in this manner) but my first couple tests of it haven’t been that promising. There is a lot of potential here but right now it barely comes close.

The Speech-to-Text feature has some real potential but seems pretty useless so far

To test out the system, I left myself a message that said:

“So over the weekend, I finally got my upgrade to GrandCentral, which is now Google Voice. It’s become part of my Google account and what most interests me is the Speech to Text transcripts – as I’m very interested to see how that will go in the future.”

Voice translated it back as:

“so that weekend i finally got an upgrade to grand central which is now google a voice it’s become part of my will come out and what most interesting the in these speech that tax the transcripts so i’m very interested to see how that’s going to go and future”

It does show which words it had trouble translating by showing them in grey text. Each transcribed message also has a “Was this transcript useful?” button to give feedback.

I typically mumble like crazy but I was REALLY damn clear in this message. I’d say that, so far, this feature is pretty much useless.

We’ll see how smart the system becomes as it learns, and I’m optimistic that this could actually becoming useful.

Didn’t get an invite three years ago?

For those of you that weren’t lucky enough to get a beta invite years ago when they were giving them away, I’m sure the service will be open for everyone in no time. Maybe by the time they let you in, they’ll have hammered out some of these issues.