The problem with this concept is that there is a whole extra dimension than in the track Seek Tool, namely time.

If you weren't listening to the radio, the profile could be constructed relatively quickly, the limiting factor in this scenario is the I2C bus which is capable of about 10,000 commands per second.

If you were listening to the radio and the profile took say 10 seconds to construct, by the time you've got back to constructing the start of the profile again the signal at that frequency may have changed, but maybe only because you're driving under a bridge.

In short the key to getting this right will be good tweaking of the amount of time the values for frequencies are averaged over, this may need to be different for when you are listening to the radio and when you are not.

(The antenna issue is easy to resolve)