Anyway to download a new system voice?

Posted:
in macOS edited January 2014
I've been using the voice commands, and with the system voice responding.



However most of the voices sound really robotic, and not fluid. So I was wondering if I can download new system voices?





I've google, but couldn't find anything. I would like to get the voice from Knight Rider

Comments

  • Reply 1 of 6
    MarvinMarvin Posts: 14,188moderator
    http://www.cepstral.com/



    It would be good if Apple provided a tutorial on how to make them. It could start a craze like themes.
  • Reply 2 of 6
    aquaticaquatic Posts: 5,602member
    Quote:
    Originally Posted by Marvin


    http://www.cepstral.com/



    It would be good if Apple provided a tutorial on how to make them. It could start a craze like themes.



    Yeah are they hard to make? Or do they just take a lot of time and persistence (many syllables recorded, many combinations, dipthongs, etc.)
  • Reply 3 of 6
    The new voice built into leopard, Alex, sounds just as good as the Cepstral voices. And its free. And it does fast speech way better. And !!*** that may not be the only new voice added.
  • Reply 4 of 6
    On another note, if Apple (or anyone, really) could make an app that lets you easily create a voice from your own through a guided tutorial, that would be intensely amazing. It would become a new market. Everyone famous would be selling their voices for teenagers to use on their computers. Hilarity ensues...



    wouldn't that be cool? You could trade voices with your friends, and prank calling would be raised to a new level.
  • Reply 5 of 6
    aquaticaquatic Posts: 5,602member
    Definitely. I want James Earl Jones to read me my alerts and e-mail.
  • Reply 6 of 6
    kukukuku Posts: 254member
    If you watched the WWDC demo and Alex demo VS Cepstral, Apple wins.



    Sure Apple would try to cherry pick their quotes, but the extra long paragraph from WWDC is a little hard to fake completely.



    It looks like apple(and to extent all Text2Speech) spend a large amount of time hiting the often used junctions well. No one is asking text to speech sing here.



    It probably very hard and have to go through massively number crunching analyzer, and recordings. You can have a computer tell you a speech is "robotic" with enough cpu power.



    Then design a program that can change the pitch based on the forumla.



    I'm wondering if there is a certain CPU requirement for the new txt2spch. It's hinted that the sw overhaul we've heard all these yrs have finally found leopard, and a good time with Intel chips that blaze away the 2x+ optimized speeds of their ppc brothers.



    a 800mhz g3 isn't really all that usable with a cpu heavy leopard, but that's like 3% of people, after their macbook cycle comes.
Sign In or Register to comment.