There are a couple of ideas floating around in my mind from some time now. I think these are brilliant ideas and if anyone implements them soon, it should be really useful to the pubic. I dont have the cost analysis, but will hopefully do some research to find its profitability. However, in todays world, more than often it happens that whatever idea you have, someone has already thought of it before and at times even implemented it. An my humble self has not done any googling to find if the solution exists. If you know of some products that solves the problems mentioned below do let me know.
Now, if you think that a wireless speaker already exists, you are right you can purchase a good set of wireless speakers for about $100. Even a good wireless headphone comes for about $90. But my problem is exactly that. While I can buy both wireless speaker and headphones separately, it is inefficient as
- The ‘wireless’ part in both speakers and headphones is common
- I cant use my existing speakers and headphones
- Its expensive and there is no wireless version of ‘cheap’ headphones or speakers for travel use.
My solution to this problem is to have a wireless unit decoupled from the speaker. The base will connect to your sound device and it will have a receiver in which you can plug in any speaker or headphone.
The technology for this already exists and just by doing behind the envelope math, I think it could be easily sold for $10 – $20 which in my opinion is the right price for a ok quality product useful for the masses.
For those of you who dont know pandora – its an online personalized radio station. They have paid professionals to tag thousands of songs and then do matching based on those tags. You specify a song or an artist and they will play streaming audio, trying best to match the properties of your chosen artist/song. But
- It only for english songs
- I cant use it on my personal collection
My solution to this is to have a simple software that captures basic features of a song – beat, tempo, instruments, style – and matches them with other songs on my local harddisk. Now this one is a bit challenging. There is a lot of research going on in finding what are the best features to use to find similarity of two songs, but those people like most researchers are trying to ‘solve’ the problem, but what I need is a ‘noisy’ solution. It does not matter to me who composed the song or who sang it or what are the instruments playing, I just want something that automatically selects the songs based on ‘some’ basic features. Something like, “Give me the product right now and keep improving it later”
An extension of this could be used in personal parties where you need a DJ to find high tempo songs.
Super-resolution using videos
Now this one is half baked idea which just came to my mind today, and a thing of the future. The problem here is that how do I take photo at the right moment! Its difficult to time the shot in fast motion – like a person falling. And many motions cant be repeated with ease. Is it possible that you take a video of the motion and to rebuild one frame using information from the adjacent frames?