VR Mode - integration of GNU Emacs and Dragon NaturallySpeaking. Available at . Copyright 1999 Barry Jaspan, . All rights reserved. See the file COPYING.txt for terms of use. VR Mode integrates the features of Dragon NaturallySpeaking with GNU Emacs. You must already have and be familiar with NaturallySpeaking 3.52 and GNU Emacs 19.34 (or newer versions) in order to use VR mode. RELEASE NOTES: Release 005 This release fixes the known bugs in V005 beta 1. Changes include: - Works on Windows 2000. - Reworked voice command setup; anyone who customized their voice command list should read the help information for vr-voice-command-list for the new structure. - New voice command "repeat that <0to20> times" repeats most recent voice command 0 to 20 times. - Fixed a synchronization bug caused when a new utterance began before the previous command finished. - Minibuffer activation is now properly controlled by vr-activate-minibuffer. - Keyboard activity is automatically locked out during speech recognition. - Other miscellaneous bug fixes. RELEASE NOTES: Release 005 Beta 1 Version 005 is a fairly major upgrade to VR Mode, with numerous new features. Unfortunately, it may also be somewhat unstable; I suspect there is a memory bug lurking somewhere. Thus, this is a beta release. Please report problems or successes. Changes include: - Added support in VR.EXE for multiple simultaneous clients, so one VR.EXE process can support VR Mode on multiple machines. - Added support for simple lists as arguments to voice commands, for example "move down 5" -> (next-line 5). Thanks to Steve Freund. - Added support for repeating commands, for example "move up", "repeat yank", etc., based on code by Steve Freund. - Added numerous voice commands to vr-default-voice-command-list. - vr-win-title and vr-win-class can now contain any substring of the title and/or class name of the Windows window on which VR.EXE will activate for a given client. - "Hookified" all VR Mode protocol commands from VR.EXE, to make VR Mode extensions possible. - Made voice activation of the minibuffer optional. - Allow exclusion entries on vr-activation-list so you can voice activate, for example, "all buffers except those matching REGEXP." - Automatically shut down VR Mode when Emacs exits, without prompting the user for confirmation. - Support multiple Emacs frames, with DNS 4.0 or newer. RELEASE NOTES: Release 004 - Allow VR Mode in Emacs and VR.EXE to run on separate computers, so VR Mode can be used on Unix. - Allow voice commands to be bound to keystroke sequences, in addition to M-x commands. - Redesign the inter-process communication architecture to use TCP/IP instead of pipes. RELEASE NOTES: Release 003 This version fixes a bug in vr.el in V002 that caused Emacs to report an error message like "wrong argument type: overlayp, nil" every time the minibuffer was activated. RELEASE NOTES: Release 002 This version is a major upgrade to VR Mode, but this should still be considered alpha-mode software. VR Mode now allows dictation into any Emacs buffer and supports multiple buffer simultaneously without losing the text/voice mapping (see vr-activation-list). It also allows voice execution of Emacs commands (see vr-voice-command-list). Auto fill mode is now supported. VR Mode will still not work properly if you use multiple Emacs frames. In this release, VR Mode does actually track the changes to the active Emacs frames, but it appears that the Dragon NaturallySpeaking dictation object does not support changing Windows handles after it is created. Perhaps this will be fixed in a future release of NaturallySpeaking. Better user documentation is needed. RELEASE NOTES: Release 001 This is the very first release of this software, so don't have high expectations! VR.EXE depends in the actual Windows handle (HWND) of the Emacs frame that is active when the program is started. If you display a VR Mode target buffer in an Emacs frame other than the original frame or, worse, close the original frame entirely, VR Mode will not work until it is restarted. Auto fill mode is ignored when voice-recognized text is inserted into the buffer. As a result, all paragraphs end up as one long line. If auto fill mode is enabled, however, you can always say "Press Alt Quebec" to invoke the Emacs command fill-paragraph. The next release should contain voice-command bindings for common Emacs commands (find-file, etc.), hooks for post-dictation processing (such as by VoiceGrip), design documentation, and who knows what else.