Musical tone note identification method

Abstract

The invention provides a musical tone note identification method which is used for solving the problems that in existing note identification, low-and-medium-frequency fundamental tones are hard to identify, and computing complexity is high. The musical tone note identification method comprises the main steps of computing voice frequency parameters, carrying out mute judging, screening local maximum points, counting frequency difference modes, determining notes and correcting the notes. The musical tone note identification method improves low-frequency stage note identification performance, enhances identification robustness under the nonideal conditions of fundamental tone vanishment and noise interference, and is simple in algorithm, quick in processing and suitable for musical tone identification applications which are low in device processing ability and have the real-time requirement, for example, the method is used for site rating of music played on a mobile phone, a handheld computer and other mobile terminals.

Description

technical field [0001] The invention belongs to digital audio processing technology, relates to a musical tone recognition method, in particular to a musical note recognition method. Background technique [0002] Tone recognition technology has broad application prospects in the fields of digitization of music works, music creation and music database retrieval. The key to musical tone recognition is the recognition of a single note, which is usually realized by estimating the pitch frequency of the musical tone (abbreviated as fundamental frequency). Pitch frequency is a basic feature of audio, which is the number of vibrations of the entire sounding body within a certain period of time. The higher the frequency, the higher the pitch. The pitch frequency directly determines the note and melody of the tone. [0003] The existing fundamental frequency estimation methods can be roughly divided into two categories with parameters and without parameters: [0004] The fundament...

AI Technical Summary

Problems solved by technology

[0006] Although note recognition technology has made qualitative progress, due to the complexity and variability of the music signal itself, as well as the influence of various application environmental factors, the performance of existing methods is not ideal, and there are still several problems in note recognition: (1) Since the energy of the fundamental tone can be lower than the energy of its overtone, or even so low that the phenomenon of "pitch disappearance" occurs, this makes the method of using spectral energy to find the fundamental tone invalid; (2) The fundamental frequency estimation in the low frequency band is always The existing difficulties are mainly due to the small pitch interval in the low frequency band, which requires the algorithm to have a high frequency resolution; in addition, the pitch disappearance often occurs in the bass part

In practical applications, such as the recognition of recorded audio of live performances, factors such as artificial percussion noise and background noise will make the recognition of low-frequency pitches more difficult; (3) Algorithms with good recognition performance Often the computational complexity is high and cannot meet the real-time requirements

Images