Let me start with this:
Why do I believe that sound would be easiest? Because it can bounce on
walls and mostly anything to get to a reciever and because it's the
easiest way and most likely cheapest way to do it
You have made a lot of assumptions there that I think has your perspective on your design flawed.
So, you say it can bounce on walls and mostly anything to get to a receiver, but why is this good? I would actually argue that it is not good at all. Ideally you will want to remove as many reflections as possible because multipath will just make it more difficult for your receiver to interpret what was sent. Likely you will have to have a very low data rate in order to overcome this.
You also say that it is the cheapest. Like Leon has said, in order to produce such low frequencies the size of the transducer has to be rather large. The cost of this can start to get pretty high compared to going with other options. Also, communication of this sort is rare so the cost of the parts will be higher just because of the lack of quantity of the item.
So what are you options?
Well, IR is an option like a few people have mentioned in comments, but this isn't a great option for you because it wont go through walls and also wont bounce off of anything. So I will go ahead and rule that one out.
Ultrasound is another option that was mentioned in comments. This might be a little bit of a better option then IR, but if you are trying to go as cheap as you can this might not be great either. I say this mostly because ultrasound also doesn't have as many people using it for communication.
RF - This includes your radio, wifi, and 3g. 3g isn't probably a great option as you have to pay a cell phone company to use the 3g. When you say radio, I assume you mean like AM and FM music radios, if so this could be an option, except that in this band you will have to find a channel that isn't being broadcast on and will have to limit you radiated power. Wifi will also be an option, but let me propose another option:
The easiest option will probably be using something like an Xbee. The reason it will be the easiest is that tons of people have done it before with Arduinos and there is lots of help available for using it. The modules might sound a bit on the expensive side to begin with, but if you start to consider the cost of designing an antenna and going with any custom board, you will very quickly add up costs. If you are designing something to be mass marketed, you will probably want to roll your own design though. I would recommend trying to stick with the 2.4 GHz spectrum as it is probably the easiest to work in as far as regulations go.