Transcribing voiced musical notes for creating, practicing and sharing of musical harmonies

A method for accurate transcription of voiced musical notes, comprising:

• receiving a user specified timing information for a musical composition to be transcribed;
• using a microphone to convert sound waves comprised of monophonic audio produced by a voiced rendition of the musical composition, into an electronic audio signal;
• storing the electronic audio signal in a data storage device;
• processing the electronic audio signal in an electronic processing device to determine a true pitch of each musical note of the voiced rendition;
• and
• wherein the true pitch is determined in the electronic processing device by
• segmenting the electronic audio signal into a plurality of musical note segments;
• sampling each note segment to obtain a plurality of audio samples;
• for each audio sample, computing an autocorrelation to determine a probability value of each audio frequency contained within the audio sample;
• identifying at least one local maxima in the energy associated with one or more audio frequencies comprising the audio sample;
• applying a corrective function to the eliminate one or more of the local maxima which are determined to be associated with octave errors;
• computing edge weights of a plurality of adjacent nodes i, j comprising a graph, where each node represents one of the musical note segments, and where the edge weight between adjacent nodes i and j is a negative log likelihood of the probability that a next musical note segment has a pitch associated with node j, assuming that node i was selected;
• and
• determining a shortest-path through the graph to select the true pitch to be assigned each of the musical note segments.