I have two ideas about the project, they are some kind of similar because both of them based on a smartphone have gyroscope, accelerator, and a good ability to computing, but the first one is focus on Graphics and the second one focus Audio.
Let’s begin at the first one:
The Invisible “I love you”
One day, Marry received a message from her boyfriend, “Look at the sky over the city center!” She ran the app “Magic city”, used the camera of the phone to look at the sky, and found the big “I love you” made by the clouds floating on the sky, like this:

(Well, I know it’s “Merry Xmas!”
)
The other day, on a celebration of Newcastle, Marry found in the 3D virtual space of Newcastle “Magic City”, a huge logo of Newcastle floating upon the city and the flowers dancing in the sky, like this:

(OK, I know, there is no flowers dancing.
)
This idea is came from this video:
http://www.youtube.com/watch?v=iikd_hucZbY
But what quite different is, the 3D virtual space of the city could not only used to navigate, even navigate, we could also use some icons in the sky to make it more gorgeous, like make the hundreds of icons floating like a pyramid, the highest one is the city’s logo, then some bigger one lower are some important place – like the nearest police station, then others… Even we could erect a huge monument like this:

An important point is, this idea may run faster then lots of augment reality programs on the cellphone, because:
1. It didn’t based on the analysis of the image, so the speed to build 3D models could be fast.
2.GPS’s an important disadvantage is it is not precisely enough, but our application is not used to navigate. After the first time the cellphone received the location information from GPS, it started to use gyroscope and compass to computing the location of the virtual object, it could be faster than always driven by GPS(It is a little similar to Google’s Skymap) and because the virtual object of this app alway be hundreds meters tall or big, only several meters imprecision won’t affect the result heavily.
The Second Idea: The Invisible Bell
The city beside my hometown has a big bell, everyday the bell ringing to tell people the time for thousands of years (a little similar to Newcastle’s bell rings, but louder). So could it be like this:
One day, I wear a headphone, and clearly sound a huge bell ring came from the sky over the city center (like the bell hang on a cloud in the image). Even when I change some direction of my head, the sound always came from there.
To make this come true, we must assisted by 3D sound( headphone ), the Binaural Recording is the best way I have ever know, like this:
http://www.youtube.com/watch?v=8IXm6SuUigI
but actually, it’s still hard to build the model of auricle in computer, so we could also use this:
http://www.youtube.com/watch?v=lqqiIR017SQ
Then if we add a gyroscope and a accelerator on the headphone, we could know how our head and the virtual sound source located in the virtual city. The data would be analysed by the smartphone and get a output to the headphone. Then, we could have a virtual huge bell tower in the city center.
Plus, the GPS+Radio is also a great point, some people said, Radio is a soul of a city, different with the TV, Radio is closely related to the city we lived. What could it be combined with GPS? Maybe this street select this song? The other street sell some old things broadcast another style of music? It could strengthen the atmosphere of the street much.

Also, because the source of the sound is far away, the imprecision could be some kind ignored. Furthermore, as the city is also 3D itself ( some cities already have a 3D model of themselves ), based on the texture, even the sound effect like how sounds broadcast in the environment has lots of building or forest could also be simulated.

At last, the city’s virtual space is an important topic today, as the more convenient head-wear display devices’ improvement, we would get rid of the camera on the smartphone’s restriction, at that time, how the virtual city based on the reality would looks like? What is this city’s “identity”? Would be more and more significant.
PS: I still have no experience on programming on sound, so I plan to do some work on the first idea first. My cellphone is HTC G1 (Android) and the coding work would be based on the GPS , gyroscope and accelerator methods which have already included in the SDK published by google.