Display sound spectral view with scipy FFT and matplotlib

detect_tones.py

#!/usr/bin/python3

# detect tones in sound spectrum with scipy FFT
#   here sound source is a USB microphone with ALSA (channel 1)

from collections import deque
import struct
import sys
import time
import threading
import alsaaudio
import numpy as np
from scipy.fftpack import fft

# some const
# 44100 Hz sampling rate (for 0-22050 Hz view, 0.0227ms/sample)
SAMPLE_FREQ = 44100
# 20000 samples buffer size (454 ms)
NB_SAMPLE = 20000


class Sampler(threading.Thread):
    def __init__(self, cardindex=-1):
        # init thread
        threading.Thread.__init__(self)
        self.daemon = True
        # init ALSA audio
        self.inp = alsaaudio.PCM(alsaaudio.PCM_CAPTURE, alsaaudio.PCM_NORMAL, cardindex=cardindex)
        # set attributes: Mono, frequency, 16 bit little endian samples
        self.inp.setchannels(1)
        self.inp.setrate(SAMPLE_FREQ)
        self.inp.setformat(alsaaudio.PCM_FORMAT_S16_LE)
        self.inp.setperiodsize(2048)
        # sample FIFO
        self._s_lock = threading.Lock()
        self._s_fifo = deque([0] * NB_SAMPLE, maxlen=NB_SAMPLE)

    def get_sample(self):
        with self._s_lock:
            return list(self._s_fifo)

    def run(self):
        while True:
            # read data from device
            l, data = self.inp.read()
            if l > 0:
                # extract and format sample (normalize sample to 1.0/-1.0 float)
                raw_smp_l = struct.unpack('h' * l, data)
                smp_l = (float(raw_smp / 32767) for raw_smp in raw_smp_l)
                with self._s_lock:
                    self._s_fifo.extend(smp_l)
            else:
                print('sampler error occur (l=%s and len data=%s)' % (l, len(data)), file=sys.stderr)


if __name__ == '__main__':
    # start sound sampler thread for sound card 1 (USB microphone)
    spr = Sampler(cardindex=1)
    spr.start()
    # main loop
    while True:
        # read samples
        samples = spr.get_sample()
        # compute FFT
        y_freq = fft(samples)
        # level axe at each frequency:
        # yf between 0.0 and 100.0 for every xf step
        y_level = 1.0 / (NB_SAMPLE / 2) * np.abs(y_freq[0:NB_SAMPLE / 2]) * 100
        # find higher level frequency
        index_max = np.argmax(y_level)
        max_level = y_level[index_max]
        max_freq = index_max * SAMPLE_FREQ / NB_SAMPLE
        print('max f=%i Hz (lvl=%.02f %%)' % (max_freq, max_level))
        # detect 330 Hz and 440 Hz tones 
        level_330hz = y_level[round(330 * NB_SAMPLE/SAMPLE_FREQ)]
        level_440hz = y_level[round(440 * NB_SAMPLE/SAMPLE_FREQ)]
        print('level at 330 Hz %.02f %%' % level_330hz)
        print('level at 440 Hz %.02f %%' % level_440hz)
        if (327 < max_freq < 333) and (1.5 < max_level < 3.0):
            print('%s: TONE DETECT (level=%.02f %%)' % (
            time.strftime("%d/%m/%Y %H:%M:%S", time.localtime()), max_level))
        time.sleep(0.5)

how-to-alsa-mic.txt

# bash GUI to see or change cards params (press <F6> to select USB sound card, <F4> to select Micro/Capture view) 
alsamixer

# command line for set Mic gain at 16 db for sound card 1 :
amixer -c1 set Mic 16dB

# backup all sounds params set with alsamixer or amixer :
sudo alsactl store

setup.sh

#!/bin/bash

# vars
NAME=$(basename $0)

# check root
[ $EUID -ne 0 ] && { printf "ERROR: $NAME needs to be run by root\n" 1>&2; exit 1; }

# install alsa audio python module
apt-get install libasound2-dev python3-pip
pip3 install pyalsaaudio

# install matplotlib and scipy
apt-get install python3-matplotlib python3-scipy

spectrum.py

#!/usr/bin/python3

# display sound spectral view with scipy FFT and matplotlib
#   here sound source is the system microphone with ALSA (channel 1)

from collections import deque
import struct
import sys
import threading
import alsaaudio
import matplotlib.pyplot as plt
import matplotlib.animation as animation
import numpy as np
from scipy.fftpack import fft

# some const
# 44100 Hz sampling rate (for 0-22050 Hz view, 0.0227ms/sample)
SAMPLE_FREQ = 44100
# 66000 samples buffer size (near 1.5 second)
NB_SAMPLE = 66000


class Sampler(threading.Thread):
    def __init__(self, cardindex=-1):
        # init thread
        threading.Thread.__init__(self)
        self.daemon = True
        # init ALSA audio
        self.inp = alsaaudio.PCM(alsaaudio.PCM_CAPTURE, alsaaudio.PCM_NORMAL, cardindex=cardindex)
        # set attributes: Mono, frequency, 16 bit little endian samples
        self.inp.setchannels(1)
        self.inp.setrate(SAMPLE_FREQ)
        self.inp.setformat(alsaaudio.PCM_FORMAT_S16_LE)
        self.inp.setperiodsize(512)
        # sample FIFO
        self._s_lock = threading.Lock()
        self._s_fifo = deque([0] * NB_SAMPLE, maxlen=NB_SAMPLE)

    def get_sample(self):
        with self._s_lock:
            return list(self._s_fifo)

    def run(self):
        while True:
            # read data from device
            l, data = self.inp.read()
            if l > 0:
                # extract and format sample (normalize sample to 1.0/-1.0 float)
                raw_smp_l = struct.unpack('h' * l, data)
                smp_l = (float(raw_smp / 32767) for raw_smp in raw_smp_l)
                with self._s_lock:
                    self._s_fifo.extend(smp_l)
            else:
                print('sampler error occur (l=%s and len data=%s)' % (l, len(data)), file=sys.stderr)


def plot_anim(i):
    # read samples
    samples = spr.get_sample()
    # compute FFT
    y_freq = fft(samples)
    # frequency axe in Hz:
    # 0.0 to max frequency (= sample rate/2), number of step is half of NB_SAMPLE
    x_freq = np.linspace(0.0, SAMPLE_FREQ / 2, NB_SAMPLE / 2)
    # level axe at each frequency:
    # yf between 0.0 and 1.0 for every xf step
    y_level = 1.0 / (NB_SAMPLE / 2) * np.abs(y_freq[0:NB_SAMPLE // 2]) * 100
    # wipe and redraw
    ax1.clear()
    ax1.set_axis_bgcolor('grey')
    ax1.set_xlabel('Frequency (Hz)')
    ax1.set_ylabel('Level (%)')
    ax1.plot(x_freq, y_level, 'b', lw=2)
    # find higher level frequency
    index_max = np.argmax(y_level)
    freq_max = x_freq[index_max]
    print('max level at f=%i Hz (lvl=%.02f %%)' % (freq_max, y_level[index_max]))
    #print('sample: max %.04f, min %.04f' % (max(samples), min(samples)))


if __name__ == '__main__':
    # start sound sampler thread
    spr = Sampler()
    spr.start()
    # init a dynamic plotter
    fig, (ax1) = plt.subplots(nrows=1, ncols=1)
    fig.canvas.set_window_title('Sound spectral view')
    ani = animation.FuncAnimation(fig, plot_anim, interval=1000)
    plt.show()

In setup.sh I think pithon3-scipy should be python3-scipy...

Thanks, it's fixed.