CODEWITHSUNDEEP
python-3.xnumpypython-imaging-library
umang
umang

Reputation: 19

How to fix 'Cannot handle this data type' while trying to convert a numpy array into an image using PIL

I am trying to visualize music into an image by using sounddevice to input the sound and then converting it to a numpy array. The array is 2D and so I convert it to 3D (otherwise I only get a single thin vertical line in the image). However when I use PIL to show the image it says 'Cannot handle this datatype' The code is mentioned below:

import sounddevice as sd

from scipy.io.wavfile import write

import soundfile as sf

import numpy

from numpy import zeros, newaxis

from PIL import Image

fs = 44100 # Sample rate

seconds = 3 # Duration of recording

myrecording = sd.rec(int(seconds * fs), samplerate=fs, channels=2)

sd.wait() # Wait until recording is finished

print(myrecording)

print(numpy.shape(myrecording))

write('output.wav', fs, myrecording) # Save as WAV file

filename = 'output.wav'

A=myrecording[:,:,newaxis]

print(A)

im = Image.fromarray((A * 255).astype(numpy.uint8))

im.show()

I expect to get an image which shows colours corresponding to the sound being inputted in

Upvotes: 0

Views: 133

Answers (1)

alexpiers
alexpiers

Reputation: 726

This depends a little bit on what you want to do.

You have two channels with n-samples ((nsamples, 2) ndarray); do you want each channel to be a column of the image where the color varies depending on what the value is? That is why you were getting a very narrow image when you just plot myrecording.

You do not really have the data to create a full 2D image, unless you reshape the time series data to be something more like a square (so it actually looks like an image), but then you sort of lose the time dependence nature that I think you are going for.

Upvotes: 0

Related Questions