Phonebox - perl text-to-speech synthesizer

The most current version of this document should be at http://www.cs.cmu.edu/~lenzo/phonebox.html .
This stuff will be repackaged as a perl module Text::Speech with a sample program.


Using Phonebox

You should be able to use phonebox without any modification. The easiest way to do this is to start it up at the command line, let it load up, and type some text into it by typing

phonebox

You may have to change a parameter or two for the path names. If so, just copy phonebox to a directory where you have write permission, set the paths, and run it from there.

At the phonebox prompt, you can enter arbitrary text for synthesis. However, the word "quit" is special, and will cause the program to exit. Also, anything entered at the prompt that begins with a slash ("/") is interpreted as a slash command.

phonebox -h will get you a description of command-line options:

  usage: phonebox [options] [subject]

    OPTION         DEFAULT    DESCRIPTION
    -d|-dataDir    <dataDir>  location of dictionary, phone, word, map
    -s|-soundDir   <soundDir> location of sound files
    -e|-ext                   extension of sound files
    -c|-caps                  force capitalization of input
    -v|-verbosity  0-3        degree of output text
    -m|-map        <mapfile>  map file for text-2-phoneme
    -w|-word       <wfile>    word file
    -p|-phone      <pfile>    phone file
    -server                   server mode
    -client                   client mode
    -host                     host server for client mode
    -standalone               standalone mode (default)
    -stdout                   output audio to stdout

    [voicename]    mrlw       name of the voice
However, this is not the complete list, nor completely accurate. I'll update this and the documentation at some point.

phonebox -v 3 is useful for generating the most verbose traces.


Phonebox structural outline


kevin lenzo (lenzo@cs.cmu.edu)