Jaunty -RT testing

[Eric S. Johansson]

Joseph Reagle wrote:
> I need to use NaturallySpeaking in a virtualized Win2K environment which
> requires real-time like performance. I run a KDE desktop.

I know this is going off topic a bit but I'm going to throw my two cents worth
and then bow out unless there is further interest in discussing this topic.

I'm in the same position.  I run Ubuntu 8.04 as my host and XP as my guest.
virtualization software is VM Ware.

AFAIK, Speech recognition needs very clean audio without any dropouts in order
to function properly.  One of the most consistent ways of getting clean audio is
through USB.  It eliminates most if not all of the audio flakiness, bugs,
inconsistency, internal noise etc.   it's a cheaper solution than what you can
get with a sound card.  Functional, reliable, cheap what more can you ask for
when it comes to USB audio?  How about a good set of linux drivers?

when I tried to use wine to run NaturallySpeaking, I discovered that 8.04 didn't
even recognize my USB microphone(vxi b200).  Then I remembered the last time I
tried USB audio with OSS, it didn't really work very well and it prevented me
from using the built-in soundcard at the same time.

Fortunately, VM Ware proxies the USB audio device into the guest so that
NaturallySpeaking can attach to it.  Since my goal was to have working speech
recognition without counting on Windows as a host OS, I stuck with VM Ware (i.e.
functionality trumps politics).

unfortunately, this virtual system solution isn't sufficient and naturally
speaking plus wine works better everyday.  Someday, I (and other crypts like
myself) will need speech recognition quality audio from USB devices and it would
be nice if a no hassle solution was available.  Believe me, even though I can
use my hands to do cryptic stuff, I would rather save them for more important
tasks related to work or having a life.  Seriously, it sucks having to decide
how you're going to use your hands instead of just using them.