Are the days of hardware synths and speaking from boot numbered?

[Tony Baechler]

Let me throw in my two cents here.  I just had a brand new server built in 
August from scratch.  One of my primary requirements was a serial port. 
Yes, I had no problem finding one.  The trick seems to be to look for a 
server motherboard.  I doubt if most assembly line desktops will come with 
them anymore because they aren't usually necessary, but from poking around 
on various sites while researching, there are still plenty of uses for 
serial ports in servers.  So, to answer part of your question, I don't think 
onboard serial ports are going away in the immediate future.  They will 
continue to be harder to find, but there are still plenty of old devices out 
there in the enterprise that need them.

To answer your second question, Debian Squeeze and the Squeeze live CD still 
support hardware speech.  They use kernel 2.6.32 and I was able to do a 
fresh install with my Trippletalk.  I had to spend many hours using the live 
CD to try to recover my files from a damaged hard drive.  I mostly did this 
with ssh, but I was still able to load the Speakup modules from the CD and I 
had speech at the console.  For the curious, just burn and boot the live CD 
as normal.  Press Enter when it spins down and wait for it to boot.  You 
won't get any beep or other feedback.  After about 30 seconds, run "sudo 
bash" to get to a shell.  Load the Speakup modules as normal.

To get to the subject of your message, it ultimately comes down to the 
kernel developers.  There are still some USB synthesizers out there, 
although software speech is now the popular trend.  I see very little 
practical need for hardware speech in the near future, but I wish it was 
still available in Wheezy.  I upgraded to Wheezy and kernel 3.2 and lost my 
hardware speech.  I kept the 2.6.32 kernel to fall back on if necessary, but 
that isn't a perfect solution and doesn't help with the 3.2 initrd.  I don't 
see a practical way to get software speech in an initrd just due to the 
overhead of sound drivers and ESpeak.  Wheezy does have software speech 
built into the installer and I can verify that it works.  I guess the 
ultimate advantage in ditching hardware speech is that there are a lot more 
desktops with speakers and supported sound devices than there are with 
serial ports, so anyone can plug in a USB stick and have a talking Linux 
with minimal hassle.  In the end, that's probably more important, as much as 
I really don't like giving up my good old hardware synthesizers.  It seems 
that most Windows users use software speech now as well.

In conclusion, if you're really determined to have hardware speech on a new 
box, it can be done with time and effort.  Then again, you can run DOS on 
most boxes too if you really want.  It's just a matter of time before the 
kernels change and you'll be stuck with software speech one way or the 
other.  Eventually, Squeeze won't be supported and you won't have the 
upgrade path available to Wheezy as you do now.  For now, you can install 
Squeeze, upgrade to Wheezy and make sure to not install the 3.2 kernel.

On 12/11/2012 12:55 PM, Arthur Pirika wrote:
> I think the subject pretty much speaks for itself. With serial synths, and
> especially serial ports getting harder to find, with the only serials synths
> still being made new are the venerable doubletalks, should work be ramped up
> on getting software speech, somehow at kernel level? I know there was a
> project working on this, but not much has been done on it for a while.
> Related to this, I assume it's still possible to build speakup into the
> kernel, although most distributions package speakup as modules, thereby
> allowing messages from the moment of powerup?