There are certain programs that translate wav files into midi...but the results are often not what you want.
They don't make these midi files by "converting" the audio files (because that is impossible to do). The files are sequenced from new. To do this, you need a working knowledge of both music and the equipment/software.
Yes, you need to listen to the song over and over again, transcribe each section of it and then either write it out with music publishing software (finale, sibelious, encore) or record each track with midi software (cakewalk, sonar, digital performer etc.) by playing it back on a keyboard. Welcome to my hell!
it's hard to do so. But if we make midi like that, the sound maybe isn't good as the original track???
Nobody said it was easy The sound is down to your skills and equipment. Midi is a poor medium for communicating music, it's too dependent on the equipment being used by the listener to play the file. Six people will, almost certainly, hear it six different ways - and not one of them the way the originator heard it! Once completed, midi files should be rendered to audio. That way the listener will hear it the way intended by the writer.
MIDI is just a way for us to play back compositions and/or arrangements through synthesizers or computers. To get the same effect as the original recording you need to record the MIDI tracks, add whatever effects you want and mix it down in a home studio. I think what you may be looking for is something called a "Vocal Eliminator". This is a piece of hardware that you can play music through which it keys in on the frequency of the human voice and wipes it from the recording. I'm not sure how well they work, not having used one, but I've seen them advertised and they claim to be able to create your own Karaoke.
I'm sorry to tell you that this is not how these devices work at all. A search through some of the other posts on this board will reveal the truth, but it is all to do with adding out of phase information together and nothing to do with frequencies. As to how well they work, the answer is highly dependent on the way in which the original recording was made and the results are definitely extremely variable.... from acceptable to totally useless. Incidentally, the result is always a mono file - which is not what most people would want.
True, the "devocalized" music output is as flat as a tennis court. Even if you don't mind listening to AM radio music, you still won't like it because of artefacts created by the vocal removal process. Also, all the so-called Vocal Removal softwares use the same "phase inverting/cancelling out" technique mentioned by Graeme. So don't be too picky about choosing which one. They're basically the same, and have in common a hidden feature not mentioned in sales pitches: Quality Removal.
Well, like I said, I'm not familiar with the process at all. I've just seen them advertised and I thought that might have been what Hoangtien was thinking about. Thanks for educating me though.