Tip:
Highlight text to annotate it
X
Remember that Speak & Spell toy from back when we were kids?
Ugh! It was so hard to get a perfect score on that thing, not because spelling was that
hard but because it was so fricking hard to understand what the darn thing wanted you
to spell. [synthetic speech] Wrong! Try again.
Was that toy designed to troll children? [synthetic speech] You are correct.
[ ♪♪♪ ]
Hi guys. Welcome to a new episode of Vlogservations.
I'm Roberto. Today's topic: text-to-speech.
I've had a lot of experience with text-to-speech. Being visually-impaired it was thanks to text-to-speech
that I was able to access the wonderful world of computers.
I remember those old Apple IIe computers with that BEX word processing software
[synthetic speech] And with a speech synthesizer that sounded something like this.
[synthetic speech] Pretty cool, huh? Okay so it was not cool but at least it got
the job done. But text-to-speech technology has evolved
a lot since the late 80s, early 90s. Today we have access to near human sounding
voices. Actually the voices are quite human.
[with weird intonation] Their speech patterns are not necessarily that human-like,
[normal voice] but they're getting better. And thanks to the proliferation of in-car
GPS navigation systems, and Siri, text-to-speech has gone mainstream.
YES! Now I don't have to feel like a freak for
having my computer or my phone talk to me. [text-to-speech from the phone]
Yes, it's about freaking time! But as much as text-to-speech technology evolves,
you know some things never change. Text-to-speech engines can mispronounce a
lot. They take some names and do a whole massacre
with them. Should we really expect the technology to
improve that much? I mean, we mispronounce names often too!
Hey, nice to meet you "Crayg". Uh, it's Craig.
Oh, oh, I'm so sorry! This is awkward.
Although in text-to-speech defense, sometimes depending on the application and the platform
being used, you can actually teach a text-to-speech engine how to pronounce words and names correctly.
I've learned to not fully trust a text-to-speech engine.
I mean, can you imagine a sentence like this? Ke dollars ha had a live performance scheduled
for tomorrow but it was announced via facebook that it was canceled.
Good thing because without autotune that would not have ended well for her or for the audience.
But how about you? How has been your experience with text-to-speech?
Does it mispronounce your name? Does it mispronounce the name of your city,
your country or any other name you care about? Have you been able to fix the pronunciation?
Please, speak up in the comments below. If you missed the last episode, don't sweat
it. There's an annotated link in the in-screen
to my right and there is also a link in the description below.
Thanks for watching this episode of Vlogservations and I'll see you guys next weekend.