spectrograph.py

# Realtime sound spectrograph built with Python, PyAudio, NumPy and OpenCV
# Cosmin Gorgovan <cosmin AT linux-geek.org>, Apr 16 2011
# Released into the public domain by the copyright holder

import pyaudio
import sys
import numpy
from cv import *

# No. of input samples used for a single FFT
CHUNK_SIZE = 1024

# Sampling rate
RATE = 44100

# Spectrogram's width in pixels
WINDOW_WIDTH = 700

# Don't change. Spectrogram's height in pixels
HEIGHT = CHUNK_SIZE/2

p = pyaudio.PyAudio()

stream = p.open(format = pyaudio.paInt16,
                channels = 1,
                rate = RATE,
                input = True,
                frames_per_buffer = CHUNK_SIZE,
                output_device_index = 0,
                input_device_index = 0)


spectrogram = CreateImage((WINDOW_WIDTH, HEIGHT), IPL_DEPTH_16U, 1)
Set(spectrogram, 0)

while (True):
    data = stream.read(CHUNK_SIZE)
    data = numpy.fromstring(data, 'int16')
    freq = numpy.fft.rfft(data)
    
    tmp = CreateImage((WINDOW_WIDTH, HEIGHT), IPL_DEPTH_16U, 1)
    
    # Copy last WIDTH-1 columns from spectogram to the first WIDTH-1 columns in tmp
    SetImageROI(tmp, (0, 0, WINDOW_WIDTH-1, HEIGHT))
    SetImageROI(spectrogram, (1, 0, WINDOW_WIDTH-1, HEIGHT))
    Copy(spectrogram, tmp)
    ResetImageROI(tmp)
    
    for i in range(1, CHUNK_SIZE/2):
      rvalue = abs(int(numpy.real(freq[i])))
      
      Line(tmp, (WINDOW_WIDTH-1, HEIGHT-i), (WINDOW_WIDTH-1, HEIGHT-i),  rvalue)
      #freq = int(RATE * (max_index / float(chunk)))
    spectrogram = tmp
    
    ShowImage("Spectograph", spectrogram)
    if (WaitKey(10) == ord('q')):
      break
      
stream.stop_stream()
stream.close()
p.terminate()