particular clearification for festival needed

[krishnakant Mane]

hello,
This is Krishnakant again.
I am perpusfully creating this as a new thread even though another
similar thread is going on in the list regarding speach synthesizers
for linux and the related impact.
I think that the topic has gone so far that I am not quite getting the
answer I am trying to search.  never the less that thread wont be
disturbed any ways.
so here is my question out of the box.
I will be proposing festival for my project to begin with.  it is an
embedded device we are building and will be like a self contained
hardware synth come screen reader.  the system will be linux kernel
2.6 or above.  I will be writing screan reader for windows to begin
with and then also for linux.  but the job of converting the text into
actual speach will be always done on the device with the linux kernel.
 the device will only take textual info from the client computer.  in
addition we have also developed modules to download doc or other text
files (formatted or non-formatted) from the client computer on to the
device for future, any time any where reading.
the modules r ready and have been tested with eflite and flite.
now all I want to know is that with 128 mb ram and more than 512 mb
storage given with intel 8086 or 8051 processor, is festival the right
choice.  Please answer on the following two points,
1. will festival run smoothly with the given hardware as mentioned above?
2.  are there good quality voices in festival available right now for
british and american english?
if I decide to build my own voices can I have better voices made from
my side which sound better then diphone voices provided?
thanks a lot,
Krishnakant.