eSpeak - RISCOS Version,   v3.04
=======================

This package only contains the RISCOS-specific files.  You should also
download the speak-1.08-source package, which contains the data files and
documentation, and also source files for use in re-compilation.

NOTE: This particular version (3.04) does contain all the source files,
      which have been updated since the speak-1.08-source package was
      released.


Requirements
============

RISCOS 3.6 or later with 16 bit sound system enabled.

On my 200MHz StrongArm, it takes 51 sec to process 6m 3s of speech output,
so it will probably run OK on an ARM7 processor also. 



Installation and Use
====================

1.  Download the speak-1.08-source package and copy the following directories
    from it into the eSpeak directory from the speak-1.08-riscos package:
       espeak-data   (data needed by the speakmod)
       src           (only needed to re-compile the speak module)
       docs          (user information)
       
    Note that for RISCOS versions earlier than 4.0, rename "espeak-data"
    to "espeak-dat" to conform to the 10 character filename restriction.
    Either "espeak-data" or "espeak-dat" are recognised by the module.
   

2.  To run the speak module:

    a. Double-click on !Boot. This sets the <eSpeak$Dir> system variable,
       This must be set before speakmod is run, so that it can find its data.
       
    b. Double-click on "speakmod" to run the module.


3.  *help speak
      should now show version 3.**

    *speak -h
      shows the available commands
            
    *speak "this is some text"
      speaks the string "this is some text"
      
    *speak -v en-f "this is some text"
      speaks with a different voice
      
    *speak -f textfile
      speaks the text from file "textfile"
      
    *speak -a6 "this is some text"
      speaks more quietly (level 6 in range 0 to 20)
      
    *speak -s200 "this is some text"
      speaks more quickly (200 words per minute)
    
    The improvement in speech quality over Speak v.2 are most clearly
    heard when listening through a good sound system (eg. domestic
    stereo system) rather than small computer speakers.
    
Use with Pluto
==============

    !Pluto will use the new module if speakmod has already
    been run.  Pluto's Speak dialog should show the new voices.
    If you wish, you can give them more meaningful names by
    renaming the files in espeak-data.voices
    
    To install permanently in Pluto, copy "speakmod" and "espeak-data"
    into !Pluto and add the following line to !Pluto.!Run
       Set <eSpeak$Dir> <Obey$Dir>
    before the speakmod is run.
          

To Re-Compile the Speak Module
==============================

1.  Run the "copysource" script in the eSpeak directory.  This copies
    the non-RISCOS-specific files from the "src" directory into
    "cpp" and "h".
    DO NOT do step 1 FOR THIS VERSION (3.04) - all the source files are
    already in place and are later than the versions in the Source package.
    
2.  Run the Makefile

This should produce a new copy of the "speakmod" module.


Problems
========

1.  It uses the sound system directly rather than through the
    SharedSound module, would allow the system volume control apply
    to speech.
    
2.  The module still uses some floating point operations, which it
    seems can cause crashes in some versions of RISCOS.  I've changed
    most to fixed-point, and those remaining during initialisation
    don't seem to cause a problem.  The one that may be a problem is
    a sqrt() call in set_frame_rms() in synthesize.cpp
    
3.  There is no session management implemented, so changes to speed,
    amplitude etc by one user of the module will affect the others.
    