USB synthesizers

[Trevor Astrope]

Greg, what you say applies to usb serial adapters in order to provide 
speech when speakup is loaded, but if usb serial ports are treated like 
the softsynth driver, speakup could access them via /dev/ttyUSB instead of 
directly. Of course, speakup would still need to know what synth was at 
the other end of the usb connection and how to communicate with that 
device, so it probably wouldn't be easy to implement.

I know there is work being done on a kernel structure for accessing serial 
devices directly, but perhaps it would be easier if it were possible to 
pass the irq and i/o address to speakup on the grub/lilo command line? It 
may not get speech right away for usb serial devices if the usb drivers 
load after speakup, but it would still allow the user to start/script 
speakup to load with the usb serial synth after the system has booted, 
provided the usb serial uarts are compatible. Or maybe that could also be 
a command line argument...

On Sat, 3 Apr 2010, Gregory Nowak wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Sat, Apr 03, 2010 at 04:26:45PM -0500, acollins at icsmail.net wrote:
>> If I were writing a wish list, I could
>> wish that the drivers for serial synthesizers could be modified to check
>> for a usb serial port with a synth on it, and use that if found.  I've
>> been told that this is a lot of work, and I'm not a programmer, else I'd
>> volunteer to help.  But could you consider adding this to the Speakup
>> to do list?
>
> Though I am a programmer, I haven't done anything involving linux
> internals, simply because I haven't had time to get around to
> learning/reading about it. So, what I'm about to say stands to be
> corrected by someone more knowledgeable than I. Some of what I'm going
> to say has already  been addressed in a general sense in the past, but
> I'm going to rain on the picnic specifically regarding your
> suggestions below.
>
>> In fact, I think the drivers, could be written to act much
>> like the softsynnth driver, since when you plug a generic usb serial
>> converter in, it gets assigned to /dev/ttyUSB and 1a number, starting at
>> 0.  In this way, Speakup could simply use a generic usb to serial
>> connection as soon as the usb system becomes available  after boot up,
>
> Ok, while this does address how to talk to the usb to serial
> converter, it doesn't tackle the problem of how to talk to the synth
> attached to the converter. As has been mentioned before, speakup talks
> to synths by directly banging on i/o ports, which have to be
> configured as standard ttyS0-ttyS3 ports, and have to emulate a 16550a
> uart as far as I know. From what I know, most (all?) usb to serial
> converters out there use their own uart types, which means that as
> things stand now, speakup would have to know how to specifically communicate with every single
> usb to serial converter supported by linux now, and in the
> future. Even if there are usb to serial converters which use/emulate a
> 16550a uart, none of them gets configured as a standard serial port
> when it is plugged in, which means that speakup has no idea how to
> talk to the synth attached to that converter.
>
> Samuel has stated that this is bound to change soon, since more people
> are wanting a structured framework for the kernel to be able to talk
> to arbitrary serial ports, and that includes speakup.
>
>> and any new drivers that needed to be written could be run in user
>> space.  It seems to me that they could be written to use the softsynthn
>> interface, but just written to output to a specifi synth on a specified
>> ttyUSB device.  Thanks for your consideration in advance.
>
> Even if the above problem weren't an issue, there's still the problem
> of translating from the doubletalk command set (that's what softsynth
> uses if I remember right), to the command set of whatever the current synth uses. That again comes back to the issue of speakup
> being specifically aware of every supported usb to serial converter, and
> synth out there, in order to know what i/o ports to bang on, and to
> know how to convert from the doubletalk command set to the native
> command set of the synth being used.
>
> While you're idea may be doable, I'm afraid it's impractical. Besides
> that, given the above factors, I think we'd find ourselves complaining
> about how responsive the whole system is, and just deciding not to use
> it in the end, in favor of software speech. Like I said, I'm by no
> means an expert, and maybe your idea isn't as impractical as it seems
> to be to me.
>
> Greg
>
>
> - --
> web site: http://www.romuald.net.eu.org
> gpg public key: http://www.romuald.net.eu.org/pubkey.asc
> skype: gregn1
> (authorization required, add me to your contacts list first)
>
> - --
> Free domains: http://www.eu.org/ or mail dns-manager at EU.org
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iEYEARECAAYFAku3+nEACgkQ7s9z/XlyUyAFWACeMce1iYy9hJai7wjNrlx+KosF
> S+4AnjCJHnwsBNxQdsVIu+UvS7+Dw1gR
> =l6e/
> -----END PGP SIGNATURE-----
> _______________________________________________
> Speakup mailing list
> Speakup at braille.uwo.ca
> http://speech.braille.uwo.ca/mailman/listinfo/speakup
>