[Audacity-devel] Control of Audacity by voice command using "SayPlay"

[Marty G. <jmg...@sy...>]

Dear Audacity Team:

I’m finishing my thesis “Voice Commands to Control Recording Sessions”, and I want to invite Audacity users to try out “SayPlay”.  

Here’s a link to a video I made showing some of what can be done:

http://www.youtube.com/watch?v=zFmMQlv8ehI

Let me know if you’d like to try it out, and I’ll send installers.


Changes to Audacity: 
There are a few key changes I’ve made to Audacity in order to make it work.  The real difference in the ModScriptPipe interface is that the named pipe I use is a single bidirectional named pipe rather than using the two unidirectional Named Pipes that ModScriptPipes uses. (I couldn’t get it to work using two single directional pipes).

I added the Get/SetTrackInfoCommand changes and the Get/SetProjectInfoCommand awhile back.  

Recently I added new ModScriptPipe commands to allow the selection of named labels to allow voice commands to cue up song sections for playback and overdub.  (I show this in the YouTube video).
This new code is for my local version of Audacity in the file: LabelCreateSelectScriptCommand.cpp

SayPlay Installer:  
I have a stable version of SayPlay for anyone who wants to try it out.  The main feature of SayPlay is to allow tracks to be named, and subsequent commands to refer to them by name.

I recently added the feature to allow selection of labels by voice command.  A problem here is that the labels need to use underscores in place of spaces.  This is because script commands do not allow strings with spaces, since the space is the string delimiter.

Currently I do not allow labels to be created by name, because it is hard enough setting the current selection, in order to set the start and end times a label, that may as well just be done from the keyboard.

SayPlay Icon:
I’ve based the icon for the SayPlay program on the Audacity icon, by simply adding a microphone to the headset and a balloon that says “play”.  Please let me know if whether this is acceptable.  I tried to change the color of the waveform to a greenish color similar to my program but failed.

I will update the Wiki to include a link to the two video and vice versa and I will provide an installer for each of the special version of Audacity using bidirectional NamedPipes and SayPlay to whomever wants it.

I’m debating making the whole thing open source but have not decided to follow through on this yet.

Known Problems:
There still a few bugs in the system, namely when multiple sessions are open, the scripting commands do not necessarily go to the correct one, and in fact is unknown which one should receive commands.  The current work around for this is to simply open one session and control it, however is a session is opened and then a new session is created and in the first session is closed the failure mode occurs as well.

I hope SayPlay can be useful in hands-free recording and particularly for blind users, in any case I’ve learned a lot about speech recognition.

Please enjoy,
J. Marty Goddard
Syracuse University