Automatic Captioning

Happy IRS Day! I know it’s stressful, trying to get your taxes mailed off at the last minute (unless you’re a geek like Brandt and sent them off in February), so today’s post is meant to add a little humor to your day.

Last March, Google launched a service called Google Voice that automatically transcribes voicemail messages into text messages. The potential this has for assisting Deaf and Hard-of-Hearing people (and many others!) is huge. In practice, though, it still has a long way to go. Google Voice is currently available by invitation only, and Brandt the technology-geek got his hands on one a few months ago. He installed it on his iPhone during our first snow day of the season in January, and I had way too much fun testing it out.

My first voicemail message:

“Hi honey! It’s snowing and sleeting outside! Holy s***, it is so cold! We didn’t have school today, hurray! Bye!”

Google’s transcribed text message:

“Hi honey, it’s snowing ends waiting outside we shared. It is so how old the business today and I hi.”

I figured maybe my voice was too high and excited, and if I talked slower it would help. Brandt asked off-hand, “I wonder how it will do with my mother’s messages, since she talks so slow?” I thought that was a great idea, so I called ‘pretending’ to be his mother—including her thick Southern drawl.

What I said:

“Hi Brandt. How are you? We are good. Did it snow there? It sure snowed here! The horses don’t like all the ice on their feet. Well, we love you. Bye.”

What Google says I said:

“Hi Brandt, How are you. We are good. He it. It’s now and then it’s Cherice and I will be here. So courses. Don’t white all eyes on the house 8 whale. We love you bye.”

Keep in mind that with her thick accent, ice does indeed sound like “eyes,” and well is pronounced “whale.” You will see this again in a minute.

The Southern accent was so much fun, I next tried a Boston accent:

“Hi, Brahndt, pahk yah cah in Hahvahd yahd.”

What Google sent:

“Hi Barack. Hi, Yahoo, it's Ron Probably Holland.”

I love that Barack is now a part of our lexicon!

My favorite transcription of Brandt’s mother’s message has to be:

“… we just wonder how you’re coming with all of your Laotians…”

I read the text and asked him, “What word is THAT?” So he started explaining, “Laos is a country next to Vietnam…” I snapped at him, “I know what ‘Laotian’ means, I meant what word did she actually say?!” Then I laughed and called him a jackass. I wondered if she could have said “lessons,” since he was working on his homework. We listened to the message and sure enough, she’s said lessons. I like Laotians better though.

Tonight’s message from her:

“We're all in for the night and I'll lease whales and we trust you all are. Take care of your sales ways. Love you.”

I told you we’d see those whales again!

The same voice recognition technology used for Google Voice was recently added to YouTube videos, in their new Auto-Captioning feature. Notably, one of the people who worked on the project is Deaf. His team focused on accessibility for disabled people, specifically the Deaf and Hard-of-Hearing. Try watching THIS VIDEO with the Auto-Captioning turned on (click the red “CC” button on the bottom toolbar, then select “Transcribe Audio”). This is from my favorite channel on YouTube, CODA Brothers. They are Hearing children of Deaf parents, and their videos are hilarious. They speak in sign and then dub their voices onto the video later. There’s a lot going on at once, with Sign, voice, and captioning simultaneously, but it will give you a little insight into ASL/Deaf Culture in preparation for our history lesson coming up soon.

[I tried embedding the video into this post, but the captions didn’t come with it, so you’ll have to watch it directly on YouTube. Technical difficulty!]

And despite what the captions say, at no point in the video are they talking about phenolphthalein, lynching, helicopters, or negative campaign tactics; and of course the word is “Deaf,” NOT death.