| Hi Paul, (and Greg Casamento -- see the important stuff relating to the SpeechServer below),
Many thanks for the feedback/comments on your Ubuntu experience. I am currently running Debian 6.0.0 on a 2GHz AMD64, as I noted earlier, which made installing GNUstep very easy.
One can resize the Monet "Synthesis Window" with the command buttons at the top right, or (when not maximized) by grabbing one of the corners and dragging. If the window is maximised, you can also grab the title bar and drag and this has much the same effect as hitting the maximize button to toggle to the non-maximized state.
However, I can see that on a 15 inch screen none of these options may be accessible unless the window happens to display at near minimum size, in which case you can hardly make it much smaller!
In "Synthesizer" there is no resize possibility. When I designed the interface, I found it hard to fit everything in and still have things legible and operational. It just barely fits on a 15 inch screen, I admit. The other tool windows are smaller in both "Monet" and "Synthesizer" Apps.
"Monet" is not really intended as a lap-top text-to-speech facility. It is intended as a serious research tool for use in developing databases for new languages. The basic text-to-speech facility if the so-called SpeechServer that runs as a daemon and is called by other Apps that need text-to-speech service ("BigMouth", for example, and also a "Services Menu" text-to-speech facility that allows any highlighted text to be spoken -- currently only in English). The SpeechServer is up and running on the Mac OS X, but has yet to be ported to the GNUstep trunk.
There is an example-App that uses the SpeechServer on the Mac version called GnuTTSClient.
I think Greg Casamento may be in the process of porting the SpeechServer. If all you want to do is hear various synthesized utterances, or get speech output from your own App, you only need the SpeechServer. Ideally, the control for the SpeechServer should also be ported both onto the Mac and onto GNUstep, otherwise you will not be able to play with the voice quality parameters very easily. I have put images of the control panel for the ServerTestPlus App (historical name, ServerTest did not show the scrolled hidden methods) on my university web site. As there is a scroll panel within the window, there are two versions so everything is visible:
These give an exact statement of the speech controls that are available and which are also accessible in the fully ported SpeechServer. The documentation for the original NeXT version is at:
which gives detailed instructions on using the SpeechServer within the original NeXTStep development environment which, as you now, is the precursor to both GNUstep and Mac OS X xcode/InterfaceBuilder, so it should be very helpful in any work porting to, and using the Speech server on, those two newer systems.
I would be very grateful for any help you can offer, if only criticism! :-) But one thing you could do is to port the "BigMouth" App, the code for which you will find in the NeXTStep trunk. Also there is the code for the pronouncing dictionary App "PrEditor". You could have a shot at that too, though you wouldn't be able to check either out thoroughly until you could connect to a working SpeechServer. You can use your system as a sandbox to avoid making mistakes which might clutter the repository and send me anything you think I could try. You could also have a shot at the SpeechServer daemon itself. It would be a great learning experience whatever the outcome.
If you have access to a Mac, you could do some things on that with a working SpeechServer, and these would readily port to GNUstep. It is the gap between NeXTStep and the two newer systems that is the big step.
Just some ideas.
Warm regards.
david
David Hill -------- The only function of economic forecasting is to make astrology look respectable. (J.K. Galbraith) --------
On Mar 11, 2011, at 1:24 PM, Paul Tyson wrote: My experience building and running gnuspeech on an Ubuntu system are in line with Marcelo's. However, on my 15-in laptop screen the Monet synthesizer window was sized and positioned inconveniently, and I could not resize it to see the command buttons. I opened the project in gorm and was able to resize it there, save, and then run the app to synthesize text input. A first improvement might be to equip the synthesizer window with resize controls. Since I don't know how to exercise other features of the app I can't say more, but in poking around I did experience unexpected window disappearances. I am glad to see discussion on this topic. I am new to gnuspeech and gnustep but would like to contribute if I can. Regards, --Paul On Fri, 2011-03-11 at 10:29 -0800, David Hill wrote: Sorry people,
I didn't copy my reply to Greg to the list. Here's the list copy
Begin forwarded message:
From: David Hill <address@hidden>
Date: March 11, 2011 10:28:14 AM PST
To: Gregory Casamento <address@hidden>
Subject: Re: [gnuspeech-contact] GNUstep & gnuspeech
Hi Greg,
We abandoned CVS a while back -- it is all in SVN now. The CVS
repository really ought to be removed from the gnuspeech savannah
site but it did not seem to be an option at the time.
Warm regards.
david
On Mar 11, 2011, at 10:09 AM, Gregory Casamento wrote:
As embarassing as this is... is there a branch you guys created
for
building the GNUstep version because what's in CVS right now does
not
build for me.
Any special instructions?
GC
On Thu, Mar 10, 2011 at 8:00 PM, David Hill <address@hidden>
wrote:
Hi Marcelo,
This is by way of a few comments on my initial experiences with
"gnuspeech" under GNUstep.
I have done some more testing on the GNUstep version of
gnuspeech "Synthesizer". There are more problems than I first
thought, including the fact I can even crash the App by simply
choosing a radio button at times (I found more than one
example). I need to look at this carefully and sort out the
problems. However, much of what was implemented for OS X works
on the GNUstep version, the most notable omission being the
spectrograph display for speech (but it display correctly for
the test waveform input). The spectral cross-section in the
analysis window seems to work as it is supposed to. Of course
there are stubs for some of the secondary graphs which I have to
flesh out in both versions. The "About" panel shows up without
its content, but the content is available in its .pdf file.
Perhaps GNUstep doesn't handle .pdf. I have to get more
thoroughly into testing and start looking at it in the
development environment.
Is there a good guide to working with the GNUstep IDE?
(especially porting from Max OS X)
When I "openapp Monet" command it prints out a message near the
start:
Monet [2431 ] Error: No audio output device is available.
Monet [2431] API OSS
Do I need to install OSS, even though I did install
portaudio19-dev. I can synthesize to a file, and then listen to
it using the Media player but cannot produce the output
directly. Perhaps this is why. On subsequent invocations, the
message did not appear.
Later, just after ", [2431]
filename: /home/david/Library/Application/Support/GnuSpeech/pronunciations, db: 0x44e9df0, has been loaded: 1" it outputs:
Monet [2431] File NSData.m: 282. In readContentsOfFile read of
file (/home/david) contents failed -t
Monet [2431] Error: Failed to load file (nil) (nil)
but it didn't seem to stop anything working.
When synthesizing the message:
Monet [3537] NSNumberFormatter-getObjectValue:forString ... not
fully implemented
appeared -- presumably explaining the failure to output numbers
on the display properly (many appear as "NaN")
Some font "not found" messages are output .
PreMo seems to work just fine using openapp, but a curious fact:
I can fire up "Synthesizer" by double-clicking the App icon but
I cannot do that with "Monet" or "PreMo".
I am impressed by how much of the project is ported and working.
I am keen to get things to a state, as soon as possible, where I
feel comfortable with a first official release, even if it is
alpha or beta. Help anyone? Comments anyone?
Warm regards and thanks for any feedback.
david
-------
_______________________________________________
gnuspeech-contact mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/gnuspeech-contact
--
Gregory Casamento - GNUstep Lead/Principal Consultant, OLC, Inc.
yahoo/skype: greg_casamento, aol: gjcasa
(240)274-9630 (Cell)
_______________________________________________
gnuspeech-contact mailing list
address@hidden
http://lists.gnu.org/mailman/listinfo/gnuspeech-contact
|
