VO-BB - 20 YEARS OLD! Forum Index VO-BB - 20 YEARS OLD!
Established November 10, 2004
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Machine Language: How Siri found its voice

 
Post new topic   Reply to topic    VO-BB - 20 YEARS OLD! Forum Index -> Chat
View previous topic :: View next topic  
Author Message
Scott Pollak
The Gates of Troy


Joined: 01 Jun 2010
Posts: 1903
Location: Looking out at the San Juan mountains

PostPosted: Tue Sep 17, 2013 11:07 am    Post subject: Machine Language: How Siri found its voice Reply with quote

Similar to the thread Bruce started, but much more informative and featuring our own September Day:

http://www.theverge.com/2013/9/17/4596374/machine-language-how-siri-found-its-voice
_________________
Scott R. Pollak
Clients include Pandora, NPR Atlanta, Wells Fargo, Cisco, Humana, Publix, UPS, AT&T, HP, Xerox and more.

www.voicebyscott.com
Back to top
View user's profile Send private message Visit poster's website
Mike Harrison
M&M


Joined: 03 Nov 2007
Posts: 2029
Location: Equidistant from New York City and Philadelphia, along the NJ Shore

PostPosted: Tue Sep 17, 2013 12:05 pm    Post subject: Reply with quote

Excellent, Scott. Thanks for sharing.

In February of 2008, I spent a week in the studio of a European TTS company, recording exactly the type of material described in this article. That company is now owned by another company, but the NDA prevents me from saying anything more. Unfortunately, even I don't know what the recordings of my voice were being used for.

But it was a wonderful experience, being part of an emerging technology.

I was glad to see this article made a point of mentioning context, which obviously plays a big part in how some words are pronounced and/or some syllables are inflected. In the some blog discussions I particpated in a few years ago, extolling the virtues of using computer-generated voices in eLearning, a point I was trying to make was that no matter how 'real' a CG voice might sound, there will always be the issue of genuine inflection. The goal of CG speech in eLearning is to save money. But if it's necessary to go through every script and play with HTML code to create the proper inflections, where are the cost savings?
_________________
Mike
Male Voice Over Talent
I have taken leave of my sensors.

Back to top
View user's profile Send private message Send e-mail Visit poster's website
Kristin Lennox
Flight Attendant


Joined: 30 Apr 2011
Posts: 858

PostPosted: Tue Sep 17, 2013 3:18 pm    Post subject: Reply with quote

Cool article!

Ryleigh went through exactly this experience when she recorded for the Proloquo 2Go speech aHobo Tounge -- she was recording for six hours a day, for four days straight.

Quote:
Scratching the collar of my neck, where humans once had gills.


And the sentences were all like that or worse -- they made absolutely no sense. Additionally, if a sentence ended with a question mark, she had to read it normally but inflect up only on the last word -- not easy to remember, but she got the hang of it a few hours in.

The producers/directors also picked out a phrase at the beginning that was correct in tone and pacing, and if Ryleigh ever started losing her cadence, they would play that for her to mimic and get back into the groove.

A really cool experience, especially now that Ryleigh can program her own voice on her grandmother's speech device...

cool
_________________
Always look on the bright side of life.

Dee doo. Dee doot doot doo dee doo.

my website
Back to top
View user's profile Send private message Send e-mail
FinMac
Lucky 700


Joined: 14 Jan 2013
Posts: 707
Location: In a really cool place...Finland!

PostPosted: Wed Sep 18, 2013 12:22 am    Post subject: Interesting thread Reply with quote

That was an interesting article, Scott!

It reminded me of when I was working as a Chess Consultant in the 1980s. My work was with a Chess Computer programmer who worked for a company based in Hong Kong.

At that time I was a rated Chess Master and my work involved developing heuristics for the program, as well as the opening book library, etc.

It seems like I worked myself out of a job as we got the little plug-in computer to play "Expert" level chess. It was still about 150 rating points or so below my rating and I could always beat the computer. But to improve on the play my boss had to rewrite the entire program. Thus ended my career as a Chess Consultant.

Will we work ourselves out of a job?

My boss and I used to debate who would be better, the computer or the human. I always took the side of the human while he choose the computer.

It was a sad day when Deep Blue beat the world chess champion, Gary Kasparov. Perhaps that day is coming for VO talent as well. Though it might still be a long way off. And in the meantime there is still a lot of work to do!

Mac
_________________
www.scottsvoiceover.com - An American voice in Finland

"If you want to get to the top, you have to get off your bottom". (Unknown)
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Scott Pollak
The Gates of Troy


Joined: 01 Jun 2010
Posts: 1903
Location: Looking out at the San Juan mountains

PostPosted: Wed Sep 18, 2013 6:31 am    Post subject: Reply with quote

Fin, having grown up learning to play chess from my dad (a mechanical engineer with an insanely analytical mind), I'm extremely impressed by your former chess rating. I was never anywhere near that, even in my dreams. I remember the Big Blue/Kasparov match well.

I used to think we wouldn't automate ourselves out of a job, but I look at the advances CG has made in movies.

The difference is that with CG, you can create unreal scenes that would almost impossible to create otherwise, such as outer space scenarios, superheroes, etc. But will it ever become competitive to replicate a human voice, either financially or artistically? That remains to be seen.
_________________
Scott R. Pollak
Clients include Pandora, NPR Atlanta, Wells Fargo, Cisco, Humana, Publix, UPS, AT&T, HP, Xerox and more.

www.voicebyscott.com
Back to top
View user's profile Send private message Visit poster's website
Scott Pollak
The Gates of Troy


Joined: 01 Jun 2010
Posts: 1903
Location: Looking out at the San Juan mountains

PostPosted: Fri Oct 04, 2013 6:12 am    Post subject: Reply with quote

* Bump *

AND>>>>>>>> Here is the REAL voice of Siri, another Atlantan, BTW:

http://www.cnn.com/2013/10/04/tech/mobile/bennett-siri-iphone-voice/
_________________
Scott R. Pollak
Clients include Pandora, NPR Atlanta, Wells Fargo, Cisco, Humana, Publix, UPS, AT&T, HP, Xerox and more.

www.voicebyscott.com
Back to top
View user's profile Send private message Visit poster's website
Display posts from previous:   
Post new topic   Reply to topic    VO-BB - 20 YEARS OLD! Forum Index -> Chat All times are GMT - 7 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group