An idea originally by sdrothrock which I've been thinking about for the past few days: An app which can be trained to recognize certain sounds. You record a few

An idea originally by sdrothrock which I've been thinking about for the past few days: An app which can be trained to recognize certain sounds. You record a few examples of that sound and then it listens and gives an alert when it hears it. Another idea, allow deaf people to "visualize" sounds. Just displaying the raw sound wave is too redundant and difficult to interpret. But what if you could display a higher level representation, perhaps obtained by unsupervised learning. Even cooler if you can attach rough labels to the sound like is done by some object recognition systems (example: http://i.imgur.com/anp7RY9.png http://i.imgur.com/anp7RY9.png). The second idea is probably unrealistic but the first is possible.