2   The Sprache Mission

2.1 Overview of the Sprache Mission Components and Topics
2.2 Launching the Sprache Mission
2.3 Topic #1: Ensuring Proper Headphone/Microphone Setup
     2.3.1 Mac OS X
2.4 Topic #2: Using Speech Instead of GUI Buttons

The Sprache mission involves one vehicle that is commanded via speech recognition. The primary capabilities highlighted are the uSpeechRec and uDialogManager. The uSpeechRec is able to recognize basic sentences configured for particular missions. The uDialogManager interacts with the user by confirming user speech and triggering events based on pre-defined sentence-variable pairs.

2.1   Overview of the Sprache Mission Components and Topics    [top]

Here is a bit of what the Sprache mission should look like:

Behaviors:Waypoint, Station
Vehicles:betty
MOOS Apps:pHelmIvP, pLogger, uSimMarine, pBasicContactMgr, pMarinePID, pNodeReporter, pMarineViewer, uSpeechRec, uDialogManager, pShare
Primary Topics:(1) Ensuring proper headphone/microphone setup.
(2) Commanding a vehicle through voice recognition.
Side Topics:(1) uSpeechRec is configured for a small set of commands.
(2) uDialogManager responds to the small set of commands by sending variables to the vehicle.
(3) The Helm and its behaviors are activated by the uDialogManager.

2.2   Launching the Sprache Mission    [top]

The Sprache mission may be launched from the command line: 

  $ cd moos-ivp-aquaticus/trunk/missions/sprache/
  $ ./launch.sh --warp=10

This mission is similar to the s1_alpha mission but has been augmented with the uSpeechRec and uDialogManager apps for using speech in addition to the pMarineViewer GUI buttons. This should bring up a pMarineViewer window like that shown in Figure ???, with one vehicle, Betty, initially in the "PARK" state. After hitting the DEPLOY button in the lower right corner, the vehicle enters the "WAYPOINT FOLLOW" mode and begins to proceed along the designated waypoints as shown. The example mission is configured to have the vehicle follow the waypoints and then return to its original location. The user can return the vehicle to its starting location by hitting the RETURN button. Toggling between the DEPLOY and RETURN buttons has the vehicle alternate between the "WAYPOINT FOLLOW" and "RETURN HOME" modes, respectively.

2.3   Topic #1: Ensuring Proper Headphone/Microphone Setup    [top]

It is highly recommended that a user wanting to use speech to interact should use a headphone/microphone combination or headset. The reason is that a loop can be created between uDialogManager and uSpeechRec in which it starts talking to itself and the user can not interrupt the loop. In the following sections we review the proper setup of the microphone and earphones. This is a typical first step as improper volume levels lead to poor or at worst case no speech recognition.

2.3.1   Mac OS X    [top]

These instructions were verified with OS X Yosemite (10.10.5). Should you have something different the process should be relatively the same. Click on the Apple icon in the upper left of your screen. Then choose SYSTEM PREFERENCES. Click on the Sound icon to get a screen similar to Figure 2.1 ?. At the top of the screen we see Sound Effects, Output, and Input.

Figure 2.1: The Mac OS X Sound Input Settings.

First select the Input button. Here is where we can test the microphone levels. Ensure that the microphone you wish to use is highlighted as the source, for example we highlight External Microphone. What we are looking for is that as you speak the Input level registers what you are saying. In Figure 2.1 ? the speaker is mid word and registers as dark grey ovals. The key knob to turn is the Input volume slider. Adjust the Input volume slider so that your speech registers but that ambient noise is not trigerring. Selecting the Use ambient noise reduction depends greatly on your environment and equipment.

    At the top of the Sound screen choose the Output button to get a screen similar to Figure 2.2 ?. Make sure to select the correct output device. For our example we chose the Headphones that are plugged in.

Figure 2.2: The Mac OS X Sound Output Settings.

2.4   Topic #2: Using Speech Instead of GUI Buttons    [top]

As described before, we can validate that the vehicle has the proper autonomy setup by hitting the DEPLOY button to place the vehicle in "WAYPOINT FOLLOW" mode. The user can also have the vehicle switch to "RETURN HOME" mode by hitting the RETURN button. Toggling between the two buttons also toggles between the two modes. However, let's use speech to interact with the vehicle. In this case we assume that either the vehicle has not started moving, it is in the "PARK" mode or is in the "RETURN HOME" mode. We will get the robot into the "FOLLOW WAYPOINTS" mode by speaking into the microphone. In the following, the labels are USER and iSAY for indicating that either the user or uDialogManager through iSay is communicating. The ideal interaction we are looking for is found in Listing 2.1 ?.

Listing 2.1 - Using Speech to get Betty to Deploy. 

 USER: "BETTY DEPLOY"
 iSAY: "Did you mean, Betty Deploy?"
 USER: "YES"
 iSAY: "Command Sent!"

Notice that uDialogManager confirms what uSpeechRec thinks the user has said. The user can acknowledge with a YES at which point uDialogManager sends the appropriate variables to get Betty into "FOLLOW WAYPOINTS" mode. If the user cancels the command by saying NO then uDialogManager will not send any variables and respond with: 

 iSAY: "Command Canceled."

Page built from LaTeX source using the texwiki program, developed at MIT