Pitch Extraction and Notes Generation Implementation using Tensor flow

Great ideas are born in the least expected place. Musical ideas might just happen while travelling, hiking or working or in any random place just not in the music studio. Every musician’s desire is to capture the flow of ideas in the moment to convert it into sheet music which can be passed on to their instrumentalists. Thus, it has become a research challenge to automatically detecttunes and recognise modes of humming melody which is an important guarantee for humming music retrieval and automatictranscription. Deep learning has played a vital role in extraction of tune features and utilising them to produce the instrumental tones although with some limitations. The data is given and it is pre-processed and explored. Our project is trained using a Deep learning algorithm, Neural Networks. The attention mechanism makes the network focus on the humming pitches and gets the note sequence by detecting most relevant tune features directly. Then, the mode of the note sequence is recognized according to the result of the tonic detection and automated tones for the specified instrument are produced.