How to generate a simple waveform of an entire sound file?

Core Library Questions

jonatan.v..

How to generate a simple waveform of an entire sound file?

in Core Library Questions • 5 months ago

I've been searching for a while now, and all I find is fancy visualizations, but I need to generate a simple waveform like this.

Is there any code examples of how to do something like this?

I'm not entirely sure which info is represented, is just volume?

Replies(4)

jonatan.v..

Re: How to generate a simple waveform of an entire sound file?

5 months ago

I figured out how to do it to some extent, but it would take me forever as I have to do +20 soundfiles of +5 mins as it requires the song to be playing. Is there a way to do it without playing the song?

Here's what I have, using Minim

void draw()
{
// first perform a forward fft on one of song's buffers
// I'm using the mix buffer
// but you can use any one you like
fft.forward(song.mix);
ellipse(x,height/2,2,2);
ellipse(x,height/2 + song.mix.get(0) *500,2,2);
x+=0.04;
if(x > width) x = 0;
}

jonatan.v..

Re: How to generate a simple waveform of an entire sound file?

5 months ago

I got there in the end, here's a code dump if anybody ever needs it

It's based on one of the Minim samples

/**
* This sketch demonstrates two ways to accomplish offline (non-realtime) analysis of an audio file.<br>
* The first method, which uses an AudioSample, is what you see running.<br>
* The second method, which uses an AudioRecordingStream and is only available in Minim Beta 2.1.0 and beyond,<br>
* can be viewed by looking at the offlineAnalysis.pde file.
* <p>
* For more information about Minim and additional features, visit http://code.compartmental.net/minim/
*
*/
import ddf.minim.*;
import ddf.minim.analysis.*;
import ddf.minim.spi.*;
Minim minim;
float[][] spectra;
void setup()
{
size(1000, 1000, P3D);
minim = new Minim(this);
// There are two ways you can do offline analysis:
// 1. Loading audio data fully into memory using an AudioSample and then analyzing a channel
analyzeUsingAudioSample();
// 2. Loading an AudioRecordingStream and reading in a buffer at a time.
// This second option is available starting with Minim Beta 2.1.0
//analyzeUsingAudioRecordingStream();
}
void analyzeUsingAudioSample()
{
//AudioSample jingle = minim.loadSample("bump.mp3", 2048);
AudioSample jingle = minim.loadSample("keys riffs.ogg.mp3", 2048);
// get the left channel of the audio as a float array
// getChannel is defined in the interface BuffereAudio,
// which also defines two constants to use as an argument
// BufferedAudio.LEFT and BufferedAudio.RIGHT
float[] leftChannel = jingle.getChannel(AudioSample.LEFT);
// then we create an array we'll copy sample data into for the FFT object
// this should be as large as you want your FFT to be. generally speaking, 1024 is probably fine.
int fftSize = 1024;
float[] fftSamples = new float[fftSize];
FFT fft = new FFT( fftSize, jingle.sampleRate() );
// now we'll analyze the samples in chunks
int totalChunks = (leftChannel.length / fftSize) + 1;
// allocate a 2-dimentional array that will hold all of the spectrum data for all of the chunks.
// the second dimension if fftSize/2 because the spectrum size is always half the number of samples analyzed.
spectra = new float[totalChunks][fftSize/2];
for(int chunkIdx = 0; chunkIdx < totalChunks; ++chunkIdx)
{
int chunkStartIndex = chunkIdx * fftSize;
// the chunk size will always be fftSize, except for the
// last chunk, which will be however many samples are left in source
int chunkSize = min( leftChannel.length - chunkStartIndex, fftSize );
// copy first chunk into our analysis array
arraycopy( leftChannel, // source of the copy
chunkStartIndex, // index to start in the source
fftSamples, // destination of the copy
0, // index to copy to
chunkSize // how many samples to copy
);
// if the chunk was smaller than the fftSize, we need to pad the analysis buffer with zeroes
if ( chunkSize < fftSize )
{
// we use a system call for this
Arrays.fill( fftSamples, chunkSize, fftSamples.length - 1, 0.0 );
}
// now analyze this buffer
fft.forward( fftSamples );
// and copy the resulting spectrum into our spectra array
for(int i = 0; i < 512; ++i)
{
spectra[chunkIdx][i] = fft.getBand(i);
}
}
jingle.close();
}
void draw()
{
// jump back to start position when we get to the end
background(0);
stroke(255);
// render the spectra going back into the screen
float scaleMod = (float(width) / float(spectra.length));
println("--");
println(width);
println(spectra.length);
println(scaleMod);
for(int s = 0; s < spectra.length; s++)
{
stroke(255);
int i =0;
float total = 0;
for(i = 0; i < spectra[s].length-1; i++)
{
total += spectra[s][i];
}
total = total / 10;
line(s*scaleMod,total+height/2,s*scaleMod,-total+height/2);
}
stroke(255,0,0);
line(spectra.length * scaleMod, 1000, spectra.length *scaleMod, 0+300);
}

lightbubb..

Re: How to generate a simple waveform of an entire sound file?

5 months ago

This is exactly what i was looking for.

Maybe im missing something but when i run your code it cannot find anything named 'Arrays'

Arrays.fill(fftSamples, chunkSize, fftSamples.length - 1, 0.0 );

is this where you store your offline music samples?

jonatan.v..

Re: How to generate a simple waveform of an entire sound file?

5 months ago

which version of processing do you use?

seems like you're missing an import

https://forum.processing.org/topic/array-fill-in-processing-2-0b7

I'm tweaking up the code so I'll give you a better version where you can do multiple files, in a few hours when it's done :)

Top Reply