PiGlow FFT audio visualizer | https://youtu.be/lJcLoAAGdoo

piglow_fft.py

#!/usr/bin/env python
# by djazz, using various bits of code found over the web

# works with both python2 and python3
# requires: python-alsaaudio, python-numpy, python-smbus, piglow
# install piglow: curl get.pimoroni.com/piglow | bash

# usage:
# this script accepts raw audio in this format: S16LE 44100 kHz Mono
#  script-that-outputs-audio | python piglow_fft.py
#  echo raw.pcm | piglow_fft.py
# see usage suggestions below

# examples:
#  mpg123 -s --no-seekbuffer -m -r 44100 http://icecast.djazz.se:8000/radio | python piglow_fft.py
#  avconv -i http://icecast.djazz.se:8000/radio  -f s16le -acodec pcm_s16le -ar 44100 -ac 1 - -nostats -loglevel info | python piglow_fft.py
# you can replace the url with a local file path
# currently it's pointing to http://radio.djazz.se


import sys
import io
from time import sleep
from struct import unpack
import numpy as np
import alsaaudio as aa

# comment out all lines referencing piglow
# if you want to try just the fft analyzer
import piglow


# Return power array index corresponding to a particular frequency
def piff(val):
  return int(2 * chunksize * val/sample_rate)

# Initialize the arrays
bins       = [0,0,0,0,0,0]
smoothbins = [0,0,0,0,0,0] # for smoothing
weighting = [1, 1.5, 3, 4, 6.5, 9] # Change these according to taste

# Precalculate weighting bins
weighting = np.true_divide(weighting, 1000000)

# audio settings
sample_rate = 44100
no_channels = 1 # mono only
chunksize = 2048

# stdin acts as a file, read() method
stdin = getattr(sys.stdin, 'buffer', sys.stdin)

# setp up audio output
output = aa.PCM(aa.PCM_PLAYBACK, aa.PCM_NORMAL)
output.setchannels(no_channels)
output.setrate(sample_rate)
output.setformat(aa.PCM_FORMAT_S16_LE)
output.setperiodsize(chunksize*no_channels)

print("Loading...")
data = stdin.read(chunksize*5)

print("Playing...")
output.write(data)

counter = 0

while data!='':
  data = stdin.read(chunksize)
  output.write(data)

  npdata = np.array(unpack("%dh"%(len(data)/2), data), dtype='h')
  fourier = np.fft.rfft(npdata)
  fourier = np.delete(fourier, len(fourier) -1)
  power = np.abs(fourier)

  lower_bound = 0
  upper_bound = 64

  for i in range(len(bins)):
    mean = np.mean(power[piff(lower_bound) : piff(upper_bound):1])
    bins[i] = int(mean) if np.isnan(mean) == False else 0
    # if counter == 0:
    #   print([piff(lower_bound), piff(upper_bound)])
    lower_bound = upper_bound
    upper_bound = upper_bound << 1


  bins = np.divide(np.multiply(bins,weighting),5)

  a = ''

  for i in range(6):
    m = bins[i]
    smoothbins[i] *= 0.93
    if m > smoothbins[i]:
      smoothbins[i] = m
    value = smoothbins[i]
    if value > 1:
      value = 1
    brightness = int(value*255)
    if brightness < 1:
      brightness = 1
    piglow.ring(i, brightness)

    s = "|"
    s = s.ljust(int(value*20), "#")
    s = s.ljust(20)

    a += s
  a += "|"

  # print bars
  # if counter % 1 == 0:
  #   print(a)

  piglow.show()
  
  counter += 1


print("Finished")

When i run this i get a error message saying syntax is wrong have looked at it and it is exactly what it says here.Thev error message comes up where 35 is the return comes up as a error any help on this would be appreciated

Going to try this code over the weekend, but a couple of comments:

1. "smoothbins[i] *= 0.93" seems redundant, isn't smoothbins array all 0?
2. Am I correct that the piff function converts the frequencies: 64Hz, 128Hz, 256Hz, .. 2048Hz to power array indicies? Why did you use those particular ranges?

Cheers

I was able to get this up and running on my Raspberry Pi 3 (running a fresh install of Jessi). The only trouble I had was that I also needed to install avconv in order to run the 2nd example command. You can do that by running: apt-get install libav-tools