A Music Cognition–Guided Framework for Multi-pitch Estimation

Abstract

As one of the most important subtasks of automatic music transcription (AMT), multi-pitch estimation (MPE) has been studied extensively for predicting the fundamental frequencies in the frames of audio recordings during the past decade. However, how to use music perception and cognition for MPE has not yet been thoroughly investigated. Motivated by this, this demonstrates how to effectively detect the fundamental frequency and the harmonic structure of polyphonic music using a cognitive framework. Inspired by cognitive neuroscience, an integration of the constant Q transform and a state-of-the-art matrix factorization method called shift-invariant probabilistic latent component analysis (SI-PLCA) are proposed to resolve the polyphonic short-time magnitude log-spectra for multiple pitch estimation and source-specific feature extraction. The cognitions of rhythm, harmonic periodicity and instrument timbre are used to guide the analysis of characterizing contiguous notes and the relationship between fundamental frequency and harmonic frequencies for detecting the pitches from the outcomes of SI-PLCA. In the experiment, we compare the performance of proposed MPE system to a number of existing state-of-the-art approaches (seven weak learning methods and four deep learning methods) on three widely used datasets (i.e. MAPS, BACH10 and TRIOS) in terms of F-measure ($${F}_{1}$$ F 1 ) values. The experimental results show that the proposed MPE method provides the best overall performance against other existing methods.

Variational Inference of Structured Line Spectra Exploiting Group-Sparsity

“…See e.g. [54] and references therein. However, these methods are usually data-driven and specifically tailored to multi-pitch estimation whereas our algorithm is the best purely model-based algorithm without being specifically tailored to the problem at hand.…”

“…(54) Note, that C S in the calculation of ln Z([ θk, θ k ], [ ẑk, 0]) equals Ck in the calculation of ln Z([ θk, θ k ],[ ẑk, 1]…”

Siamese Residual Neural Network for Musical Shape Evaluation in Piano Performance Assessment

“…A Constant-Q transform (CQT) [26] is employed to generate the pseudo-color spectrogram for each audio recording. The reason for selecting CQT is because it has been frequently utilised as time-frequency transform in many MIR tasks [27]. Theoretically, any time-frequency transform methods can be fitted into our model.…”

Novel welch-transform based enhanced spectro-temporal analysis for cognitive microsleep detection using a single electrode EEG

“…With cognitive AI, machines get trained to be able to think, reason, and make decisions much like humans do! [19][20][21] Although existing research work had restricted scopes due unavailability of real-world datasets and problematic experimental conditions, researchers yielded results using indirect methods like observing subject's typical sleep patterns in the vehicle [22,23,23], or experimented with decreased cognitive response time using alcohol [24,25] or conducted repetitive tasks to elicit tiredness [26,27,27]. When it comes to sleep, there is no such thing as a typical state of being.…”