Indian classical musical instrument classification using Timbral features

Summary Musical instrument classification becomes effective when the music signal arrives with profound characteristics. This urged the researchers to develop an automatic system of recognizing the music signals and classify the instruments interplayed through the music. Thus, this paper proposes a...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Concurrency and computation Ročník 33; číslo 21
Hlavní autori: Gulhane, Sushen Rameshpant, Shirbahadurkar, Suresh Damodar, Badhe, Sanjay Shrikrushna
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Hoboken, USA John Wiley & Sons, Inc 10.11.2021
Wiley Subscription Services, Inc
Predmet:
Classification
cuckoo search
dragonfly algorithm
Fourier transforms
fractional Fourier transforms
Indian classical music
Kurtosis
Music
musical instrument
Musical instruments
Optimization
Search algorithms
Signal classification
Spectra
stack autoencoder
Timbral features
ISSN:1532-0626, 1532-0634
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Summary Musical instrument classification becomes effective when the music signal arrives with profound characteristics. This urged the researchers to develop an automatic system of recognizing the music signals and classify the instruments interplayed through the music. Thus, this paper proposes a model for the Indian music classification system using the optimization‐based stacked autoencoder. The significance of this research is based on the proposed Cuckoo‐dragonfly optimization (CuDro)‐based stacked autoencoder, where the proposed CuDro optimization trains the stacked autoencoder for acquiring accurate classification results. The proposed CuDro technique is the combination of the standard Cuckoo search (CS) and the Dragonfly algorithm (DA) that renders optimal weights for training the stacked autoencoder (SAE). Moreover, the musical instrument classification using the proposed CuDro‐based stack autoencoder is based on the compact features, such as Timbral features and proposed FrMkMFCC features, which further add value to this research. The Timbral features like Spectral flux, spectral kurtosis (SK), Spectral skewness, Spectral pitch similarity, Roughness, In harmonicity are added in the research for efficient musical instrument classification. The proposed FrMkMFCC feature is the integration of the Fractional Fourier transforms and Multi kernel method, and Mel Frequency Cepstral Coefficient (MFCC) features. The analysis using the developed classification methodology confirms that the proposed method acquired the maximum accuracy of 96.16%, the sensitivity of 86.86%, and specificity of 92.85%, respectively.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1532-0626
1532-0634
DOI:10.1002/cpe.6418