This software is developed under such environment. It may work under other environment, but not supported. Any report about working / not working is welcome.

• Ubuntu 12.04 LTS 32 bit
• OpenJDK 1.7.0_09
• gcc version 4.7.2

This tool uses gstreamer, sphinxbase, pocketsphinx, and flite for speech recognition.

• jna: From apt apt-get install libjna-java
• gsteramer: From apt apt-get install libgstreamer0.10-dev
• sphinxbase
  • Fetch from CMU Sphinx - Browse /sphinxbase/0.8 at SourceForge.net
  • configure, make, make install
• pocketsphinx
  • CMU Sphinx - Browse /pocketsphinx/0.8 at SourceForge.net
  • configure, make, make install
  • install gst plugins

cd src/gst-plugin
make
sudo cp .libs/libgstpocketsphinx.so /usr/lib/gstreamer-0.10

• flite
  • Fetch from Index of /flite/packed/flite-1.4
  • configure, make, make install

This project uses Gradle (and gradle-wrapper) as a build tool.

Go to project home and run these commands.

./gradlew build   # build
./gradlew install # create executables of Java modules

• Move recognizer/build/lib/libaccel_recognizer.so to a directory included in $LD_LIBRARY_PATH environment variable.
• Load projectdict/clients/accelvoice.el from your emacs setting, and (require 'accelvoice)

Load client library and set projectdict path.

(setq accelvoice--root "/path/to/accelvoice")
(add-to-list 'load-path (concat accelvoice--root "/projectdict/clients"))
(require 'accelvoice)
(setq accelvoice--projectdict-path (concat accelvoice--root "/build/install/projectdict/bin/projectdict"))
(global-set-key (kbd "C-,") 'accelvoice--complete)

These configs make AccelVoice more useful on Emacs.

;; Show completion status (esp. current candidates size) in mode line
(defun insert-before-vc-mode (rest-format)
  (if (and (listp (car rest-format)) (eq (car (car rest-format)) 'vc-mode))
      (cons '(:eval (accelvoice--update-mode-line)) rest-format)
    (cons (car rest-format) (insert-before-vc-mode (cdr rest-format)))))

(setq-default
 mode-line-format
 (insert-before-vc-mode mode-line-format))

;; Start completion automatically when you type more than 1 words
(defun accelvoice--post-command-hook-function ()
  (when (and accelvoice--current-process (eq this-command 'self-insert-command))
    (let ((bounds (bounds-of-thing-at-point 'symbol)))
      (when (and bounds (< 1 (- (cdr bounds) (car bounds))))
        (accelvoice--complete)))))

(add-hook 'post-command-hook 'accelvoice--post-command-hook-function)

Take tomykaira/clockwork as a sample project. git clone [email protected]:tomykaira/clockwork.git under ~.

• M-x accelvoice--start-projectdict opens the minibuffer. Put path to the project clockwork`.
• Put "clo" into any buffer, and press C-,, or M-x accelvoice--complete.
• Then speak "clockwork" to the microphone.
• "clo" will be expanded to "clockwork".

If it gives unexpected result (for example "clock"), try several times and other words. It it seems not working, check the log file ~/.accelvoice/projectdict.log.

Here are key lines to find out the cause.

• "TIME [main] INFO io.github.tomykaira.accelvoice.projectdict.App - Starting projectdict server for /path/to/clockwork/": Emacs correctly starts AccelVoice projectdict server
• "TIME [main] INFO io.github.tomykaira.accelvoice.projectdict.Server - System initialized in ***: Server started successfully.
• "TIME [pool-1-thread-2] INFO io.github.tomykaira.accelvoice.projectdict.HttpCompletionHandler - -1938647573: complete clo from clockwork clock clone": completion is correctly started
• "INFO: callbacks.c(31): vader start ********: vocal input is recognized. If there is no line like this, ensure the mic is working.
• "ERROR: "fsg_search.c", line 1104: Final result does not match the grammar in frame ***": AccelVoice is working, but matching candidate is not found. Try again with better pronunciation ;)

If you find a problem or have a question, please report it from Issues.

If you fix a bug or implement enhancement, please send me a pull-request.

GNU General Public License version 3.

Copyright 2013 tomykaira

This file is part of AccelVoice.

AccelVoice is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

AccelVoice is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with AccelVoice.  If not, see <http://www.gnu.org/licenses/>.

See COPYING for more details.