Title:           Guide to Not-Completely-Shitty Text-To-Speech Step 1: Install MARY - Go h
Author:          Anonymous
Pastebin link:   http://pastebin.com/vcuqUWPq
First Edit:      Saturday 1st of December 2012 11:17:35 PM CDT
Last Edit:       Saturday 1st of December 2012 11:17:35 PM CDT

Guide to Not-Completely-Shitty Text-To-Speech
 
Step 1: Install MARY
 
 - Go here: http://mary.dfki.de/Download/mary-4.3.0-released
 - Click "Download MARY TTS 4.3.0" to download the installer.
 - Run the installer (requires Java)
 - Follow the default installer settings.  When it asks you what voices to
   install, just make sure it says that "cmu-slt-hsmm" is already downloaded
   (which it does by default) and then hit Quit.  (Installing extra voices will
   make it use several GBs of RAM, so that you have to run the program with
   special Java settings or it will run out and crash.  Also, you can re-run
   just this part of the installer later on if you want to get more voices.)
 
 
Step 2: Start the MARY Server
 
 - Look for the start menu item to run the MARY server.  If you can't find it,
   find the directory where you installed MARY and run
   "MARY TTS/bin/maryserver".
 
 
Step 3: Using MARY
 
 - Start a browser and go to http://localhost:59125/
 - Make sure "Voice" is set to "cmu-slt-hsmm en_US female hmm".  (This should
   be the default.)
 - Put text in the left box and hit the "SPEAK" button to hear it.
 - Don't forget to close the server when you're done.
 
 
Other stuff:
 
 - I found the default speaking rate was too fast.  To slow it down you have to
   switch the input type from TEXT to RAWMARYXML.  Then put these lines into
   the box:
 
        <?xml version="1.0" encoding="UTF-8" ?>
        <maryxml version="0.4" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
        xmlns="http://mary.dfki.de/2002/MaryXML" xml:lang="en-US">
 
        <p><prosody rate="-40%">
 
        <!-- INSERT TEXT HERE -->
 
        </prosody></p></maryxml>
 
   Put your text after the line that says "INSERT TEXT HERE".  (Don't put your
   text inside the <!-- ... --> markers, or it will be ignored.)  To adjust the
   speaking rate, change the -40% in the <prosody> tag to something else.
   (The rate setting is the percent change from the default rate, so -50% is
   slower than -40%, and -30% is faster.)
 
 - Don't put in more than 2 or 3 paragraphs at a time, or it will take a long
   time to produce the audio when you press SPEAK.
 
 - If you get no sound when you press SPEAK, there is probably some bad
   punctuation in your text.  Try deleting different parts of the text until it
   starts working, then narrow down the broken part and fix it.
 
 - You will get different kinds of pauses between words depending on what
   punctuation you use (comma vs. period - note that "..." is the same as just
   a single period).  Also, a period will produce different effects depending
   on whether the first letter of the following word is capitalized or not.