Speech to text software

I need someone (or a team) to create a speech to text software for me. These are the key differences between what I need, and similar software that already exist : - The recognized words must be written to text files, as opposed to an open txt file or word document. - It must be able to recognize voices generically. By this I mean, I should not have to create a profile and train it to recognize a specific voice. It should be able to recognize the words of any voice it hears without the need for training. - This software needs to be able to work with quieter voices as well as loud ones. - Ideally, it should be able to read words when there are quieter voices or music in the background. The loudest voice should always be the one it reads. I know that Speech to text is not flawless yet, so don't worry if you can't create a software with 100% accuracy, though obviously, the higher the accuracy, the better. Here are the remaining specifics : - Although the software's main purpose is to write the information to a txt file, I also need it to display the words it recognizes on some sort of interface. Since it may be reading a lot of words, it should only display, say, the last 100 words recognized, not every word since it began running. - When the recognized words are written to a text file, there must be one word per line. - Ideally, the software should learn as it operates, becoming better at recognizing words the longer it is used. - This software will mainly be used to recognize words spoken in an English or American accent. - In making this, you can use any previously developed code, or anything from any open source software you are aware of. I don't mind. - The software will need to run in windows 7. - The budget specified here is flexible. I am open to offers with a higher price.