System of generating model for estimation, and estimation system

A system of generating a model for estimation that generates the model for estimation used for estimation based on information about musical pieces, the system comprising circuitry configured to:

• acquire learning information that is learning data used for machine learning and in which information about audio data of a plurality of musical pieces having sequences is aligned in a dimension of the sequences in order of the sequences;
• and
• generate the model for estimation by performing machine learning using the learning information as an input to the model for estimation performing convolution using a kernel in the dimension of the sequences, wherein:
• the learning information includes time, frequency and sequence data extracted from the audio data of the musical pieces,
• the circuitry is configured to perform convolution in the time, frequency and sequence dimensions,
• the circuitry generates the model for estimation performing pooling in time and frequency dimensions without performing pooling in the dimension of the sequences, and
• the information about the audio data is a spectrogram.