Skip to content

Latest commit

 

History

History
96 lines (77 loc) · 5.14 KB

README.md

File metadata and controls

96 lines (77 loc) · 5.14 KB

eaR: An R package for perception-based auditory analysis

AppVeyor build status CRAN status lifecycle

This is a version in development of the eaR package, an open-source R tool for perception-based auditory analysis. eaR provides user-friendly and flexible tool oriented to facilitate the workflow of researchers interested in content analysis and automatic extraction of auditory and musical features. Routines are mainly designed according to the auditory short term memory (ASTM) model developed by Marc Leman and previously implemented in IPEM Toolbox.

Install the Package

Before downloading eaR, you should have installed devtools and load it:

install.packages("devtools")
library("devtools")

After that, to download eaR, you can type in the R prompt:

install_github("m-vidal/eaR")

A short glipmse to Auditory Images, the atom of eaR

The eaR package contains an adapted version of the auditory periphery model proposed by Van Immerseel and Martens (1992). The model simulates the cochlear mechanical filtering using an array of overlapping band-pass filters yielding to a set of harmonic oscillators distributed in pitch-related frequency bands. An audio signal can be represented as a primary image or auditory nerve image (ANI), which is k-dimensional vector of rate-code of neural discharges in k-frequency bands.

To calculate the ANI of the SchumannKurioseGeschichte dataset, type:

library(eaR)
data(SchumannKurioseGeschichte)
ANIs <- CalcANI(SchumannKurioseGeschichte, 22050) #auditory nerve image
PPs <- PeriodicityPitch(ANIs)
PlotImage(PPs)

After running the code, you will get an image as follows,


which is a transformation of the auditory nerve image into a more coherent pitch representation.

Version 0.2.1

The package was designed through the object class ANI, which at least must contain an image, the sample frequency of the image and a set of frequencies or periods used in their transformations. Unlike in IPEM Toolbox, the formulation of the class ANI facilitates interaction in the use of the functions and prevent disorderly data. This version includes the functions detailed in the following table. In addition, you can check whether the functions and documentation are available.

Function Status Documentation Migrated
AdaptLevel ✔️ ✔️ Yes
as.Wave -- -- No
BellShape ✔️ -- Yes
CalcANI ✔️ ✔️ Yes
CalcNoteFrequency ✔️ ✔️ No
CalcOnsetsFromANI ✔️ -- Yes
CalcProfile -- -- No
CalcRMS ✔️ -- Yes
Clip ✔️ -- Yes
ContextualityIndex ✔️ ✔️ Yes
CreateMask ✔️ -- Yes
FindAllPeaks ✔️ -- Yes
FindNearestMinima ✔️ -- Yes
InstallAuditoryModel ✔️ ✔️ No
LeakyIntegration ✔️ ✔️ Yes
OnsetPattern ✔️ -- Yes
OnsetPatternFilter ✔️ -- Yes
OnsetPeakDetection ✔️ -- Yes
OnsetPeakDetection1Channel ✔️ ✔️ Yes
PeriodicityPitch ✔️ ✔️ Yes
PlotImage ✔️ ✔️ No
RoughnessFFT ✔️ ✔️ Yes
ShepardTone ✔️ ✔️ Yes
ShepardToneComplex ✔️ ✔️ Yes

More details

eaR uses a core application written in c-code which is fully executable in R environments. Both the eaR Package code and the auditory model c-code are released under the GPL.

eaR Package was build on R 4.0.3 (2020-10-10) -- "Bunny-Wunnies Freak Out".