Recommendation method and recommendation system on basis of sound discrimination

The invention is applicable to the technical field of audio frequency discrimination, and provides a recommendation method on the basis of sound discrimination. The recommendation method includes acquiring sound which is generated when a user reads specific characters; extracting the sound to acquire a frequency waveform graph of the sound and the time; transforming the frequency waveform graph to obtain a frequency waveform graph in a set time range, and subtracting vectors of the frequency waveform graph in the set time range from vectors of a preset standard frequency waveform graph to obtain vector difference between the frequency waveform graph in the set time range and the preset standard frequency waveform graph; querying a preset mapping table of vector difference and states according to the obtained vector difference to obtain a state corresponding to the user. The recommendation method in a particular implementation mode has the advantage that the physical state of the user can be discriminated according to the sound generated by the user.