I finally got festival text 2 speech to work. I followed a set of instructions but had to make edits to set it up to work (albeit a few warnings). At the end I tried using quit with double quotes inside parentheses and without but could not stop the program (i.e. comes back with "festival>"). Tried CONTROL-C but that didn't work but by accident one time CONTROL-D did work. Does someone know if that is correct? Did I leave out something?
Here is how I did it:
FESTIVAL Installation
The following files are available as of 18 September 2011 and their
corresponding explanations follow:
Files:
1) ANNOUCE-2.1 25-Oct-2010 16:32 3.0K
2) festival-2.1-release.tar.gz 05-Nov-2010 11:21 769K
3) festlex_CMU.tar.gz 16-Nov-2008 16:17 1.8M
4) festlex_OALD.tar.gz 20-Apr-2010 07:56 1.4M
5) festlex_POSLEX.tar.gz 20-Apr-2010 07:56 238K
6) festvox_cmu_us_awb_cg.tar.gz 19-Apr-2010 11:53 4.2M
7) festvox_cmu_us_rms_cg.tar.gz 19-Apr-2010 11:53 5.0M
8) festvox_cmu_us_slt_arctic_hts.tar.gz 25-Oct-2010 03:18 1.0M
9) festvox_kallpc16k.tar.gz 20-Apr-2010 07:55 3.9M
10) festvox_rablpc16k.tar.gz 20-Apr-2010 07:55 5.1M
11) speech_tools-2.1-release.tar.gz 05-Nov-2010 11:11 1.6M
12) festival-1.4.3.ps.gz 27-Dec-2002
1) Some requirements and brief overview
2) Full English (British and American English) text to speech
C++ source for modules, Scheme Interpreter One Defun (SIOD) library
3) Lexicons based on CMULEX
4) Lexicons based on OALD (restricted to non-commercial use only)
5) Contains reverse probabilities of the words with special tags
6) Scottish English Male (with US frontend) clustergen
7) American Male using clustergen
8) American Female, HTS
9) American English Male residual LPC diphone
10) British English Male residual LPC, diphone
11) Edinburgh Speech Tools, low level C++ library
12) festival-1.4.3.ps.gz (Festival Manual ver 1.4)
As regular "user"____________________________________
Suggested downloads from above for US text to speech:
festival-1.4.3.ps.gz
festival-2.1-release.tar.gz
festlex_CMU.tar.gz
festlex_OALD.tar.gz
festlex_POSLEX.tar.gz
*festvox_cmu_us_rms_cg.tar.gz
*festvox_cmu_us_slt_arctic_hts.tar.gz
*festvox_kallpc16k.tar.gz
speech_tools-2.1-release.tar.gz
* Select 1 or more for use, e.g. festvox_kallpc16k American English Male
Should save all per a release.
gunzip/tar into directory "/opt/TTS"
gunzip <name>.tar.gz
tar -xvf <name>.tar
This produces 2 directories:
festival and
speech_tools
As "root"__________________________
Change to "speech_tools" directory:
cd speech_tools
NOTE: in speech_tools/INSTALL file it is suggested that most users
should uncomment the line SHARED = 1 in the "config" file.
This line does not exist there (now) and seems to be taken
care of in the script. So this action is ignored.
Run:
./configure then
make
NOTE: this takes approximately 10 min. on a 486 machine with nothing
else running on the AMD 22000+ chip with a 133 Meg front end..
Check the results using:
make test
NOTE: last line of this shows: Test OK
Change to the "festival" directory
cd /home/TTS/festival and run:
make
As regular "user"___________________
Change to "festival" directory
cd /home/TTS/festival and run:
bin/festival
and which produces:
Festival Speech Synthesis System 2.1.95:beta July 2011
Copyright (C) University of Edinburgh, 1996-2011. All rights reserved.
For details type `(festival_warranty)'
festival>
From here the following should work:
festival> (SayText "hello world") # Include parentheses
To leave festival enter: Control-D
To vocalize from a test file:
echo "Good morning, welcome to Festival" > fesTest
/opt/TTS/festival/bin/festival --tts fesTest
<End of Install>
---------- Post updated at 04:02 PM ---------- Previous update was at 02:31 AM ----------
Found it. Yes CONTROL-D is what is to be used to exit festival and return to the shell. Also (quit) does work to do the same guess I was getting tired.