• View
• Edit
• History
• Print

FacelessPhone

So, if we wanted a phone with a) no display and b) no buttons, how would we interact with it.

And, btw, this is Michael's crazy request, but if it takes off, I'm taking full credit. ;)

Registration

First of all, we're still dealing with a Jabber-over-WiFi solution, so all that stuff is still in place. Thus, we'll need to register a JID to the phone. This could be done via a dedicated website and a serial number on the phone. So, when you get the phone, you go to the site and enter the serial number. This then asks you for the JID and password for the JID. The phone must be in a wifi hotspot at the time of registration, as entering the info on the site will then somehow "push" to the device... not sure exactly how this will work, but it needs to do something to set the phone to log into that JID when in a hotspot.

Commands

All the commands would be voice activated, since we obviously don't have any buttons or display. There should only be a few commands, and they should be short and differ greatly from one another in sound (so as not to confuse the speech recognition engine). So, we might have:

• Answer
• Call _________ (where _________ is the name of the person to call)
• End (Call)
• Record
• Stop (Record)
• Save (voice mail)
• Delete (voice mail)
• Skip (voice mail) -- may want to rename to Pass, as Skip and Stop are fairly similar sounding

Because all of these are common English words, we don't want to trigger the command processor everytime someone says one of those words in conversation. There are two ways to handle this. The first is the precede the command with a trigger word. The second is to use some type of physical interaction with the phone itself. I prefer the second option because the first seems too error prone and processing expensive (accidentally using the trigger word during normal conversation, the command processor "missing" the trigger word, needing the command processor to be continually listening, etc).

Squeeze

So, my proposal for the physical interaction is to squeeze the phone. A single squeeze tells the command processor to start listening. It should stop listening as soon as it understands the command and provides audio feedback (it should probably repeat the word back, so the user has confirmation that the correct action was taken). Jen and I decided a momentary switch is probably a better solution. Squeezing the device triggers the switch, which causes the command processor to start listening. Releasing the device causes the command processor to stop listening and to attempt to match input to a command. It should still repeat the command back to the user for confirmation.

Visual Feedback

There are four cases where I think it's important to provide visual feedback: when the phone is ringing, when a call has been missed, and to indicate the strength of the wifi signal and the battery. These could all be done via LEDs, similar to this phone. It's also important to indicate who is calling, but I think that could be determined with a voice command (or, perhaps, whenever the phone is ringing, a single squeeze speaks the name of the caller.

Contact's Names

Because we'd presumably be using the contact's name's rather than the JID, we need some way to indicate that a particular name is associated with a JID. The name would also probably have to be said by the user, then some type of pattern matching would have to be done after a Call command. Not totally sure how to handle this yet.

Hardware

Hardware we'd need to build this:

• Gumstix Connex 200 - $99
• cfstix board - $25
• breakout-gs - $27.50
• Netgear MA701 802.11b Wi-Fi card - $75
• Headset
• Something to squeeze -- like a shell over the board container that completes a circuit
• Custom LEDs