Ideally i am a very lazy guy, i usually end up doing what i see. 90% of the time its only doing what i see, i don't even do something that's on my todo list if it's not there in front of me. Hence i find that it becomes very important to see something right in front of us, so that we keep on picking up on things.
Something as simple as tracking the rates of my share on the market, and selling it when the time might be right.
But eventually i happen to ignore the whole of it, and suddenly one day i check my portfolio.
This made me think, that i need to the stock prices to come to me everyday and show me my portfolio status everyday, rather than me going to my portfolio.
Same thing shall happen with the goals that we have in our mind. Intermittently they should keep on coming to us, rather than we going and trying to find them.
This whole idea of getting things i want come to me and remind me about them gave me an idea. That we need an application, in which we write stuffs, and this application keeps on reminding me about them again and again, till we achieve them.
So i am planning to start writing an application which shall take your stocks and your investment date and your rate of buying and selling and it will mail you the rate at which you are making money or loosing it, i.e. %income, gross %income, etc. thrice in a day.
I am assuming mail as a message board which i shall visit and read daily, but the catch here is that you should not fill your messageboard with more than what you can handle, and the timing of receiving the messages shall also be relevent, otherwise we learn to ignore mails too.
We shall mail short term goals, or things that require quick attention daily, like stocks quotes etc.
long term goals shall me less frequent.
Need to figure out the way to keep the goals, and their reminders.
Also using this we shall be able to set priorities to things.
Later i wanna add more features to it.
Lets see lets start atleast with stock quotes mailer. ;)
Wednesday, December 8, 2010
Wednesday, February 24, 2010
Find my Song.
I was just wondering if there can be a better method to find music, songs. Finding music or songs is mainly dependent on the idea of tags. I mean we need to know tag information about the song to find a song.
Also if we have duplicate songs, which are gathered from different sources, it is difficult to find it, until all songs are properly tagged.
Say we have a song and don't have any of its tag information, how can we retrieve its information.
So I came up with idea of searching music with actual music.
The idea is very simple, we down-sample the music we need to search, and then match the pattern in original file with approximation.
Firstly we would downsample the file so that we get a rough wave of the music, we don't need accuracy cause it would be difficult to match waves exactly when music can be ripped at different sampling rate.
Now after downsampling the music, we reconstruct the wave. Now the music sample we have which we will search in the source file will also be downsampled at the same sampling rate, and the wave will be reconstructed. Now this wave will be approximately matched with the wave of the original file.
If we have a match we know which song this is, else, it doesn't match.
The drawback of this method would be that it actually requires a lot of processing. But we might be able to get way out this. Say we downsample first few seconds and store in the file itself, now when we get a start of a song, we compare this with the data, and if it matches we know that this is the same song.
The idea is very abstract right now, but i am wondering how can i test the same and actually see how feasible this option is.
Also if we have duplicate songs, which are gathered from different sources, it is difficult to find it, until all songs are properly tagged.
Say we have a song and don't have any of its tag information, how can we retrieve its information.
So I came up with idea of searching music with actual music.
The idea is very simple, we down-sample the music we need to search, and then match the pattern in original file with approximation.
Firstly we would downsample the file so that we get a rough wave of the music, we don't need accuracy cause it would be difficult to match waves exactly when music can be ripped at different sampling rate.
Now after downsampling the music, we reconstruct the wave. Now the music sample we have which we will search in the source file will also be downsampled at the same sampling rate, and the wave will be reconstructed. Now this wave will be approximately matched with the wave of the original file.
If we have a match we know which song this is, else, it doesn't match.
The drawback of this method would be that it actually requires a lot of processing. But we might be able to get way out this. Say we downsample first few seconds and store in the file itself, now when we get a start of a song, we compare this with the data, and if it matches we know that this is the same song.
The idea is very abstract right now, but i am wondering how can i test the same and actually see how feasible this option is.
Subscribe to:
Posts (Atom)