Awesome

qoa

QOA is fast. It decodes audio 3x faster than Ogg-Vorbis, while offering better quality and compression (278 kbits/s for 44khz stereo) than ADPCM.

https://qoaformat.org/

This R-package is a wrapper to handle QOA files with R. The package is completely written in base R with no dependencies and pure C-code. To handle the raw audio data the R-package tuneRis highly recommended.

See this blogpost for an introduction https://phoboslab.org/log/2023/02/qoa-time-domain-audio-compression

Installation

You can install the development version from GitHub with the following command:

if (!require("devtools")) install.packages("devtools")
devtools::install_github("JohannesFriedrich/qoa4R")

Usage

There are just two main functions: readQOA() and writeQOA().

readQOA(): Takes an qoa-format audio file and decodes it into it´s raw audio data. Return value is a list with decoded values represented as matrix, used channels, sample rate and samples.
writeQOA(): Takes an matrix with every channel in new column and the sample rate as input and and encodes it into an qoa audio file.

`readQOA()`

Let´s read in an QOA-file delivered with this package:

library(qoa)
qoa_file <- system.file("extdata", "58_guitar_sarasate_stereo.qoa", package = "qoa")
qoa_data <- readQOA(qoa_file)

qoa_data is a list with the following elements: * data: a matrix with all the samples. The dimension of the matrix is samples x channels * channels: Number of channels * samples: Number of samples per channel * samplerate: Sample rate of the file

It is possible to write these samples and information with the tuneR package back to an wave file:

tuneR::writeWave(
  WaveMC(
    data = qoa_data$data, 
    samp.rate = qoa_data$samplerate, 
    bit = 16), 
  "wave_from_qoa.wav", 
  extensible = FALSE)

`writeQOA()`

With this function it is possible to write an qoa-file. The input arguments are:

samples: a matrix with dimensions samples x channels
samplerate: samplerate of the audio file
target: Either name of the file to write, a binary connection or a raw vector indicating that the output should be a raw vector (the hex-interpretation of the qoa-file).

If no second argument is given, the returned value is the hex-interpretation of the image in the QOI-file format.

wav_original <- tuneR::readWave("wave_from_qoa.wav", toWaveMC = TRUE)
writeQOA(samples = wav_original@.Data, 
         samplerate = wav_original@samp.rate)

If an second argument is given as character the image is saved to this name:

writeQOA(samples = wav_original@.Data, 
         samplerate = wav_original@samp.rate, 
         target = "test.qoa")

If the second argument is of type connection the hex interpretation of the audio file will be send to this connection, e.g. via sockets or FTP.

file <- file("file.qoa", "wb")
writeQOA(samples = wav_original@.Data, 
         samplerate = wav_original@samp.rate, 
         target = "test.qoa")
close(file)

Acknowlegment

This package would not exist without the following persons/homepages/tutorial/…: