an idea for a s'ware

Discussion in 'Programmer's Corner' started by eesycho, Feb 12, 2008.


  1. ...

    0 vote(s)
  2. ...

    0 vote(s)
  1. eesycho

    Thread Starter New Member

    Feb 12, 2008
    Hi,i'm a new gay just joining in. I'd really appreciate it if you can give me an idea for this software I am hopping to develop.The softoare is supposed to recive a sample human voice and use this vioce to synthesize a speech from a text given to it by a user. I'd be looking forward for your response thanks.
  2. Papabravo


    Feb 24, 2006
    I don't see how you could possibly have enough information from a sample to reconstruct an arbitrary text. If you can control the sample and get say 80% of the required phonemes with some rules about how to fill in the rest then maybe you have a shot at it.

    What exactly are we voting on -- it makes no sense to me!
  3. beenthere

    Retired Moderator

    Apr 20, 2004
    A voice midi system? And one that is able to take its samples from a few sentences spoken to it? I think we would be quite interested in your sampling methodology.
  4. Art

    AAC Fanatic!

    Sep 10, 2007
    You might want to review the first few words of your post,
    not that it's a problem unless you are misrepresenting yourself ;)

    Take a look at the open source Flite speech engine.
    It will give you some idea how a typical speech engine has worked since the '80s.
  5. Dave

    Retired Moderator

    Nov 17, 2003
    Have I read that correctly? It is effectively taking a spoken input and synthesising a completely different verbal speech based on the original input?

    Initially I thought this was a voice recognition exercise, but the objectives need to be clarified.

    I have also closed the poll.

  6. Art

    AAC Fanatic!

    Sep 10, 2007
    As I understand, it's a run of the mill speech engine supposed to learn how a particular
    Human sounds, and then mimick the sound of that Human in it's own regular speech.
    Maybe possible, but I don't think it will happen effectively given some arbitrary sample
    from the Human subject.

    A relatively tough project unless you are Miles Bennett Dyson,
    or have access to his material, or are a product of his work.
  7. thingmaker3

    Retired Moderator

    May 16, 2005
    Sounds like something from Star Trek. Specifically from A Taste of Armegeddon wherein "Anan 7" uses nearly the exactly the described technology to fake Captain Kirk's voice. The only difference is "Anan 7" spoke into a mic rather than typing.

    I suspect this project will require much reading about the sounds comprising the human voice. I've read that a human infant will, in the first x months of life, make all the sounds a human is capable of making. In the next x months, they have limited themselves to the sounds of the language(s) commonly spoken in their presence. Getting that kind of range out of a short sample may prove an insurmountable challenge. Interesting nonetheless.
  8. chesart1

    Senior Member

    Jan 23, 2006