Mumble for Deaf People

Anonymous
2012-09-15
2012-10-30

  • Anonymous
    2012-09-15

    You have a few things that could be converted to make this possible. Recently
    a new person joined our guild/clan who has hearing difficulties and I thought
    that Mumble could help

    • Overlay shows who is talking could it carry text as well ?

    • Mumble can convert Speech to Text so could that system also reverse and convert voice to text instead ( at least for one person on the channel chosen by the deaf receiver lets say the chosen person is a group leader - his voice could be selected as primary Voice to Text )

    There are no current comms software packages to my knowledge that can or even
    attempt to do this - this would be a challenge but one that would make
    partially deaf if not deaf players join in with the rest of us as we laugh and
    joke and play.

    Yes we could type or even stop and type what others are saying but its a great
    opportunity to and niche possibility.

     
  • Stefan H.
    Stefan H.
    2012-09-15

    Hi John,

    currently the overlay does not carry any text messages. It could and most
    likely will in the future. It would be quite a bit of work to integrate that
    properly though so I'm not sure when we'll get to it.

    Unfortunately, in contrast to Text-to-Speech, speech recognition is still a
    pretty hard problem. I'm not aware of any open source solutions that work well
    and would be easily usable (I think there's Sphinx but that isn't exactly an
    out of the box solution and I'm unsure about its performance). Windows since
    Vista ships with a somewhat usable speech recognition that can be accessed
    over SAPI and provides access to such functionality afaik, not sure how
    flexible the API is though. No idea about MacOSX. Since we not try to be one-
    sided with features this is somewhat of a concern.

    Though in the end I'm not sure how much good it would do. You have to speak
    very deliberate to get good recognition rate with current applications and
    many of them only work really well with training. Conversational/Agitated
    speech from random speakers like you'd encounter in Mumble is pretty much
    impossible to get decent detection rates on (Google is already insanely good
    yet their automatic captioning on Youtube still not much more than hilarious
    most of the time).

    I'd love to experiment with this though, probably using SAPI at first because
    it's easiest and I don't need extra software. Only problem is my list of stuff
    I want to play with is pretty long so I wouldn't hold my breath ;-)

    Thanks for your suggestions. I'm sorry I can't really offer a proper timeline
    or anything but our core team isn't that big and it's just a labor of love for
    all of us. I'll ask around whether someone wants to give it a shot.

    Regards,
    Stefan