Monday, March 22, 2010
Voice Synthesis
Voice is an amazing component for a robot. Voice synthesis can be adapted to tell you what is going on inside the robot in a huge variety of ways; (1) when a battery is low, (2) where the robot is headed (destination), (3) when the robot has encountered an obstacle, (4) what the robot is passing, according to its map, to its left, to its right, (5) today’s date and/or what the time of day is, (6) what a specific sensor is reading; sonar, wheel encoders, temperature, triangulator, ammeter (current to each motor), (7) the results of its latest system checkup (diagnostic), (8) how close you are standing in front of it. It can sing a song, hum a tune, mimic a train, or snoring, and just all out entertain you. It can read (speak) your e-mail to you, or relay other immediate messages to you from anyone you know anywhere on the web or from other robots on the web.

We don’t have the e-mail functionality released for the SR4 Robot yet, and we haven’t really tested snoring. But everything else that has been mentioned above has already been demonstrated at one point or another or is just plain easily done with the SR4 Voice board.

The Voice Synthesis board on the SR4 Robot utilizes the SpeakJet chip by Magnevation. The SpeakJet converts a series of commands composed of complex sounds or allophones, the most basic parts of spoken word, into an audio stream which is played on the robot’s on-board amplifier-speaker.

A speech application program that accompanies the SR4 Voice synthesis board provides a simple text to speech interface based on words available in a database on the robot. A line of text can be entered in a text box, and when the speak button is selected; the text will be spoken by the robot. Raw phonetic codes can also be entered into the text box with the use of a backslash (\), then the name of the phoneme, such as '\OHIH' for a long I sound. Non-text sounds may also be entered in this fashion.

In speaking words, the software relies solely on the word dictionary to convert words to their phonetic combinations. If a word is not contained within the dictionary, it cannot be spoken and is subsequently skipped when the phrase is spoken by the robot. There are approximately 1400 English words in the dictionary initially provided. The SR4 Developers manual provides instructions for how to add words to the dictionary, as well as how to modify words already in the dictionary.

The code for this sample voice application is provided so that you can use it as an example for writing voice action into any application that you are writing for the robot. Once you form the habit of incorporating voice feedback from your robot in every application you write, you will never want to go back to working with a robot that stays silent all the time. Of course, do everything in moderation.

The SR4 Voice board is listed as an option whenever you select the Buy Now button of the SR4 Robot model that you are interested in. Just selecting Buy Now, does not automatically get the robot into shipping for you. You do have an opportunity to see the list of options available for the model you are considering, and of course quantity, shipping and payment information before you are truly 'buying now'. So don’t be shy about checking out the options with each robot. You can also see these options in the compare sheet. Finally, you can purchase the SR4 Voice card and 10 watt amplifier-speaker as a standalone purchase for use directly with and as an external peripheral to your PC or Laptop (connected via USB), if you wish.