Shop OBEX P1 Docs P2 Docs Learn Events
Female VocalTract? — Parallax Forums

Female VocalTract?

ProcessingData...ProcessingData... Posts: 208
edited 2009-07-28 14:27 in Propeller 1
Is it possible to create a female voice with vocal tract, or is it just too high of a frequency?-- ProccessingData...

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Basic Stamp,···· Propeller,·· · SX,·· FUN!


START:·
>Proccessing Data. . . .··
>Task Complete. . .·.
>Saving Data. . . .
>Entering SLEEP Mode. . . .
>Signing OFF


·

Comments

  • mctriviamctrivia Posts: 3,772
    edited 2009-07-23 06:04
    should be but would require lots of time.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

    Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

    Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
  • HannoHanno Posts: 1,130
    edited 2009-07-23 06:48
    It's a big commitment.
    Also, getting the code right will be hard- it'll change it's mind all the time.
    Once you start it'll be hard to do anything else- code "jealousy"
    And finally, the code might become dynamic and reproduce- then you lose all your free time.
    Hanno

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Download a free trial of ViewPort- the premier visual debugger for the Propeller
    Includes full debugger, simulated instruments, fuzzy logic, and OpenCV for computer vision. Now a Parallax Product!
  • MagIO2MagIO2 Posts: 2,243
    edited 2009-07-23 08:57
    But wouldn't it be nice to have a female propeller? You could put it together with a male propeller and have your own propeller production running. How long does it take to grow the little props up to 32 bits??
  • 4Alex4Alex Posts: 119
    edited 2009-07-23 13:58
    Mating 32bits propellers? Who knows, perhaps we'll get 64bits offsprings...

    In any case, a female prop would be the only reason I'd be looking for a hot IC !

    Couldn't resist. Cheers,

    Alex
  • Toby SeckshundToby Seckshund Posts: 2,027
    edited 2009-07-23 15:01
    The femail vocal tract usually tells me to stop pratting about with electronics, and to do something else.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Style and grace : Nil point
  • Mike GreenMike Green Posts: 23,101
    edited 2009-07-23 15:13
    It's not just a matter of changing the pitch (frequency). The proportions and ratios of proportions are different. You can experiment with the vocal tract parameters and see what you get.
  • 4Alex4Alex Posts: 119
    edited 2009-07-23 15:39
    Mike is right. Synthetic female speech is difficult to achieve. I've used a few chip sets and the male version always have a better sounding. Here's a link from AT&T Labs. Even with all their resources you can hear the difference using the different speakers. You can give it a try at:

    http://www.research.att.com/~ttsweb/tts/demo.php

    Cheers,

    Alex
  • mctriviamctrivia Posts: 3,772
    edited 2009-07-23 15:50
    Though if you just increase frequency will sound more female. Accidently played a bill Crosby record at wrong speed he sounded like a girl

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

    Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

    Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
  • Oldbitcollector (Jeff)Oldbitcollector (Jeff) Posts: 8,091
    edited 2009-07-23 15:54
    Chip did a fantastic presentation on speech at UPEW. You can view it on the recap page.
    www.parallax.com/tabid/784/Default.aspx

    But hey, why lower the bar... Why not a sultry female VocalTract voice?? tongue.gif

    OBC

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    New to the Propeller?

    Visit the: The Propeller Pages @ Warranty Void.
  • ProcessingData...ProcessingData... Posts: 208
    edited 2009-07-24 01:28
    4Alex said...
    Mating 32bits propellers? Who knows, perhaps we'll get 64bits offsprings...
    Hey, we might be on to something here...lol.gif

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Basic Stamp,···· Propeller,·· · SX,·· FUN!


    START:·
    >Proccessing Data. . . .··
    >Task Complete. . .·.
    >Saving Data. . . .
    >Entering SLEEP Mode. . . .
    >Signing OFF


    ·
  • Toby SeckshundToby Seckshund Posts: 2,027
    edited 2009-07-24 15:24
    Yeah, but ...

    The next thing somebody would try would be immoral, in a misguided quest for 128bits

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Style and grace : Nil point
  • mctriviamctrivia Posts: 3,772
    edited 2009-07-24 15:34
    We need 4 different voices so we can have a whole family on 1 prop

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    propmod_us and propmod_1x1 are in stock. Only $30. PCB available for $5

    Want to make projects and have Gadget Gangster sell them for you? propmod-us_ps_sd and propmod-1x1 are now available for use in your Gadget Gangster Projects.

    Need to upload large images or movies for use in the forum. you can do so at uploader.propmodule.com for free.
  • 4Alex4Alex Posts: 119
    edited 2009-07-24 17:47
    How could the quest for 128bits be misguided and immoral? Electronic selection could improve the propeller breed to a greater future. More efficient, faster, leaner, a bit (no pun intended) like the bulk of the current breeds of dogs that didn't even existed 150 years ago. Shorter leads and brainier marvels seems to be the shining evolutive path for the electronic denizens. Oh wait: I think I get your point.

    On the other lead, so to speak, we all agree that avidly seeking the enjoyment of an artificial feminine voice tract is definitely not immoral! wink.gif

    Cheers,

    Alex
  • VIRANDVIRAND Posts: 656
    edited 2009-07-27 06:47
    ProcessingData said...
    Is it possible to create a female voice with vocal tract, or is it just too high of a frequency?

    I tried to hack the Sevens and Monks demo voices for that but I got strange results,
    for example I couldn't get some of the voices to
    sing certain notes without them cracking out loud popping random static noise.
    (I wonder what causes that and how to avoid it, because it happens to me a lot when I try to
    change the voices.)

    How do the formant filters work in the vocal tract object?

    Changing the crystal seemed to only make lower voices, or silence,
    but I haven't tried 6.25Mhz crystals like those someone mentioned in an unrelated topic.

    My experience with other kinds of speech synthesis suggests female voices are NOT too high.
    (If they were, then you couldn't understand them on wire telephones, which only go up to 3Khz.)
  • John R.John R. Posts: 1,376
    edited 2009-07-27 20:40
    VIRAND said...My experience with other kinds of speech synthesis suggests female voices are NOT too high.· (If they were, then you couldn't understand them on wire telephones, which only go up to 3Khz.)
    You mean you can understand them on wire telephones?· I can't even understand them "live", much less Memorex or anything else.· Hear them, yes; understand them, no.
    ·
    FWIW: I vote for sultry with a·British accent.
    ·
    Back on topic:· I agree that it sure seems easier to get male voices, but why?· Is this an indirect result of a male dominated area of development, or are female voices really harder to synthesize?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    John R.
    Click here to see my Nomad Build Log
  • 4Alex4Alex Posts: 119
    edited 2009-07-28 12:30
    @John R.:

    I don't think 'a male dominated area of development' is to be blame here, though we've seen plenty of that happening in the past.

    My gut feeling would be to look at the significant physiological differences of the voice organs in the male and female. As young child, both genders sounds quite similar. As the body develop from infant (they all cry similarly) to early adolescence, the voice change relatively linearly. It's only through puberty that significant (and dramatic) variations occurs. Kids do sounds quite 'neutral' when young (i.e., it is more difficult to identify the speaker's gender over the phone). In more mature adulthood, the voice changes again as the organs are no longer as flexible as they used to be. The 'richness' (you call it sultry) comes from multiple, complex, and unique developmental configurations. The pulsed air is accordingly modulated into very complex minute differences almost unique to each speaker, so it becomes easier to identify them individually. Male voices use more of the lower frequencies than female voices. Perhaps there is less diversity in that sound mix, so it makes it simpler to synthesise. Or for listeners to simply recognise the end result as being of a 'typical' male voice.

    The polyphonic voices demo from Parallax really sounds like males singing, and it is definitely in the lower frequencies domain.

    So, my other gut feeling is that it shouldn't be as difficult to create a synthetic voice of kids as compared to adults. The modelisation could be a good starting point, with progressive addition of variations as the working model evolves. Great academic / science fair project. It would be interesting to see ('hear') at which point the 'real' difficulties starts. Anyway, I doubt that there would be much commercial interest in that: beside being somewhat of an annoying pitch, people might not feel comfortable enough with kid voices giving them directions...

    Cheers,

    Alex
  • LeonLeon Posts: 7,620
    edited 2009-07-28 12:47
    I used to work on the human factors of military communication systems. Lots of studies have shown that male speech is significantly more intelligible than female speech, using both natural and synthesized speech, over a given communication channel. For emergency conditions I remember a proposal that a familiar voice, such as that of a pilot's wife or child, might produce a quicker response than the usual warning tone.

    Leon

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Amateur radio callsign: G1HSM
    Suzuki SV1000S motorcycle

    Post Edited (Leon) : 7/28/2009 12:54:27 PM GMT
  • Mike GreenMike Green Posts: 23,101
    edited 2009-07-28 13:17
    On the other hand, both natural and synthesized female voices are noticed (by both male and female subjects) much more readily than male voices which is why emergency messages (like stall warnings) are in female voices.

    Apple's voice synthesis software has very nice male and female voices that come with the Mac OS as standard features.
  • Toby SeckshundToby Seckshund Posts: 2,027
    edited 2009-07-28 14:27
    I remember the voice synth software that ran on a C64 back in the mid eighties, via it's SID chip, they made great play of it's effectiveness because of the use of Walsh functions. I just looked up the wiki on them and got frightenened off with all those superscripts and subscripts. Perhaps there is a version for dummies.

    The thing was very litteral in converting text to speech and it left the prose looking like text-speak, before texting was dreamt up.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Style and grace : Nil point
Sign In or Register to comment.