We are working on a project which need a sound recognition function. The sound file will be a WAV file, and it will be a human voice, singing a melody. We need to convert it to MIDI file (or a txt file which lists the musical notes and duration of the human voice).
You can see a very simple recorded voice
[login to view URL]
and you can test with your own recorder also. Please keep in mind that always there will be some noises. It will be a user-recorded wav file.
There is no possibility of multiple sound. This will be always one human voice.
The project is written in Delphi, so a Delphi function will be preferable, but if the function is written with another programming language, its ok if the communication routines are clearly defined.