This is a modified version of python-espeak (previous version here). It is a Python binding over the eSpeak speech synthesizer C library and does not simply make calls to the espeak binary.

Some highlights of the modifications.

• File output instead of only playback (espeak.init(playing=False)).
• Support for multiple instances of speakers (the espeak.Espeak class). They have to take turns and not simultaneously speak.
• Easily setup multiple callbacks (espeak.add_callback).
• Wave file content sent through callbacks (for playing=False only since this is how the C library works).
• Slightly more uniform bindings in some places (espeak.const).
• Optional start and end position for synthesis.

Unfortunately some of the renamings make this version not backwards compatible. (Maybe this library should be named differently. I will try to contact the original author about that. Their email "rainct ubuntu com" seems defunct.) The big one is that espeak.init() now has to be called before anything else.

This library should (still) work with both Python 2 and 3.

Requires espeak and its libraries to be installed espeak/speak_lib.h should be in your include path somewhere. Install with

python setup.py install

or python setup.py build to get the library in the build without installation.

Simple usage

import espeak
espeak.init()
speaker = espeak.Espeak()
speaker.say("Hello world")
speaker.rate = 300
speaker.say("Faster hello world")

See espeak.const['parameter'] for all the values other than rate that can be changes.

Callback usage

def print_callback(*args):
    print args

import espeak
espeak.init()
speaker = espeak.Espeak()
speaker.add_callback(print_callback)
speaker.say("Hello world")

This should print something like

('', 2, 1, 0, 0, None)
('', 0, 1, 0, 0, None)
('', 1, 1, 5, 0, None)
('', 0, 1, 5, 0, None)
('', 1, 7, 5, 0, None)
('', 0, 7, 5, 0, None)
('', 5, 11, 0, 0, None)
('', 0, 11, 0, 0, None)

The arguments of callback functions are wave_file_fragment, event, current_pos, length, num_samples, name.

• event can be reverse looked up in (or compare to) espeak.const['event'].
• wave_file_fragment is a piece of a wave file (only in playback=False mode, otherwise its always the empty string).
• current_pos is event->text_position from speak_lib the C library
• length is event->length from speak_lib
• num_samples is event->length from speak_lib
• name is either
  • (event->id).name for play and mark events,
  • event->id.number (for sample_rate events) or
  • None otherwise.

GPL v3. See COPYING for full text.