Python read output sound not input
Asked Answered
U

3

12

Any clue on what the easiest way to read the system sound level is?

I'm planning on creating an equalizer made out of short led strips and hooking them up to an Arduino or RaspberryPi. I've seen a lot of examples how to do it with the input, mic or such but not that much with the overall output. At this point I can't care about program specific or system wide, I'm happy when I actually get readings. Any advise would be welcome.

-- EDIT --

based on How get sound input from microphone in python, and process it on the fly? I changed the PCM_CAPTURE to PCM_PLAYBACK which results in python saying read is not possible on the output/playback.

I'm now attempting to create a loopback device, the idea is to pipe the playback back to a separate capture and read it then but no luck so far and I don't know how neat/clean this way is. I can't really imagine that this is the way to go.

Underlinen answered 3/5, 2013 at 21:2 Comment(3)
Please add more information on what you've attempted so far.Rapparee
based on #1937328 I changed the PCM_CAPTURE to PCM_PLAYBACK which results in python saying read is not possible on the output/playback. I'm now attempting to create a loopback device, the idea is to pipe the playback back to a separate capture and read it then but no luck so far and I don't know how neat/clean this way is. I can't really imagine that this is the way to go.Underlinen
Please update your question instead of commenting to it when possible. This way new readers can more easily analyse the issue and answer.Rapparee
M
3

Somebody achieved exactly this using Python and PulseAudio, see this blog posting. The author wrote a script that polls the system-wide peak sound level to feed this to a VU-meter. He also made the source available on Bitbucket, here.

I've tested it on Ubuntu and it works for me (though only from the command line, not in an interactive Python session). Some extra work involved is installing a ctypes PulseAudio wrapper for the interfacing with Python (link in source) and maybe setting the SINK_NAME to the correct value in the script.

What I gather from the blog PulseAudio is very well suited for this job, but maybe there are better ways.

Maybellmaybelle answered 13/5, 2013 at 22:32 Comment(3)
I have seen this and tested it but it didn't work great. With my sink name in place I only get 0's. After setting up a combined device with pulseaudio pref program (paman) on Ubuntu I was able to read something. Strange this was, even with paman I wasn't able to read out the sound level on almost any device except the combined devices. I am going to try this on different setups but I haven't had the time yet.Underlinen
If you want to go with PulseAudio, it seems a prerequisite that pavumeter shows the volume of the sound that you're playing. Are you sure your sound plays through Pulse and not directly to Alsa or something?Maybellmaybelle
Yes, I am but even if something is actually wrong with my system (which may very well be) I can access multiple systems with different setups to test thisUnderlinen
B
0

Almost any sort of sound mixer (software or hardware) should be able to reroute sound output to input. For Windows there's the built-in Stereo Mix, and for Linux there are several solutions (I have heard good things about PulseAudio, but I can't testify myself - some of the steps in this tutorial may help).

That way you can route it to Python - the examples of reading microphone levels you have read should then work just as well.

Bronchial answered 4/5, 2013 at 2:35 Comment(1)
From my understanding pulseaudio is a software layer on top of alsa and not every distro uses pulse but must of them do use alsa. Since I would like to be able to run it on a RasberryPi with XBMC as well alsa is the way to go I think.Underlinen
M
0

I found this github solution:

mserve mouse button hover.gif

I setup the python program as a separate instance so it wouldn't lag my main program. Then poll the IPC pickles every 33 milliseconds (30 fps). I modified the code a bit as you can see below:

#!/usr/bin/env python
# -*- coding: utf-8 -*-

#==============================================================================
#
#       vu_meter.py - Listen to microphone left/right and generate vu level
#
#==============================================================================
"""

FROM: https://github.com/kmein/vu-meter

.gitignore          NOT INSTALLED
LICENSE             NOT INSTALLED
README.md           NOT INSTALLED
amplitude.py                        THIS MODULE
record.py           NOT INSTALLED
vu_constants.py                     THIS MODULE
vu_meter.py                         THIS MODULE

ENHANCEMENTS:

    1) Support Python environment variable for automatic python 2 / 3 selection
    2) Add UTF-8 to support Python 2 (not necessary in this program though)
    3) Reset maximal between songs (gap when 10 samples of zero strength)
    4) Remove console output and write values to ramdisk (/run/user/1000)
    5) Optional 'stereo' parameter to measure left & right channels
    6) Utilize numpy which is usually auto-installed on distros
    7) Include separate amplitute.py & vu_constants.py in main
    8) Remove unused functions from Amplitude class

"""

from __future__ import print_function       # Must be first import
from __future__ import with_statement       # Error handling for file opens

import pyaudio
import numpy as np              # January 24, 2021 support separate 2 channels
import sys
import math
import struct
#from amplitude import Amplitude

RATE = 44100
INPUT_BLOCK_TIME = 0.05
INPUT_FRAMES_PER_BLOCK = int(RATE*INPUT_BLOCK_TIME)
SHORT_NORMALIZE = 1.0 / 32768.0
# Mono output
VU_METER_FNAME  = "/run/user/1000/mserve.vu-meter-mono.txt"
# Stereo output (Left and Right)
VU_METER_LEFT_FNAME  = "/run/user/1000/mserve.vu-meter-left.txt"
VU_METER_RIGHT_FNAME  = "/run/user/1000/mserve.vu-meter-right.txt"


class Amplitude(object):
    ''' an abstraction for Amplitudes (with an underlying float value)
    that packages a display function and many more
    
    January 25, 2021 - Remove unused add, subb, gt, eq, int & str functions
    '''

    def __init__(self, value=0):
        self.value = value

    def __lt__(self, other):
        return self.value < other.value

    def to_int(self, scale=1):
        ''' convert an amplitude to an integer given a scale such that one can
        choose the precision of the resulting integer '''
        return int(self.value * scale)

    @staticmethod
    def from_data(block):
        ''' generate an Amplitude object based on a block of audio input data '''
        count = len(block) / 2
        shorts = struct.unpack("%dh" % count, block)
        sum_squares = sum(s**2 * SHORT_NORMALIZE**2 for s in shorts)
        return Amplitude(math.sqrt(sum_squares / count))

    def display(self, mark, scale=50, fn=VU_METER_FNAME):
        ''' display an amplitude and another (marked) maximal Amplitude
        graphically '''
        int_val = self.to_int(scale)
        mark_val = mark.to_int(scale)
        delta = abs(int_val - mark_val)
        # print(int_val * '*', (delta-1) * ' ', '|',mark_val,int_val,delta)
        # January 23, 2021: Write values to ramdisk instead of displaying
        with open(fn, "w") as vufile:
            vufile.write(str(mark_val) + " " + str(int_val))


def parse_data(data, channel_ndx, channel_cnt, maximal):
    '''
        Process data from one channel
    '''
    data = np.fromstring(data,dtype=np.int16)[channel_ndx::channel_cnt]
    data = data.tostring()
    amp = Amplitude.from_data(data)
    gap = amp.value      # For signal test below.
    if amp > maximal:
        maximal = amp
    return amp, maximal, gap


def main():

    # January 24, 2021 separate left and right channels
    parameter = 'mono'
    if (len(sys.argv)) == 2:
        parameter = sys.argv[1]     # Null = 'mono', 'stereo' = Left & Right

    audio = pyaudio.PyAudio()
    reset_baseline_count = 0
    try:
        stream = audio.open(format=pyaudio.paInt16,
                            channels=2,
                            rate=RATE,
                            input=True,
                            frames_per_buffer=INPUT_FRAMES_PER_BLOCK
                           )

        maximal = Amplitude()
        maximal_l = maximal_r = maximal

        while True:
            data = stream.read(INPUT_FRAMES_PER_BLOCK)

            # January 24, 2021 separate left and right channels
            if parameter == 'stereo':
                ampl, maximal_l, gap = parse_data(data, 0, 2, maximal_l)
                ampr, maximal_r, gap = parse_data(data, 1, 2, maximal_r)
                if maximal_r < maximal_l:
                    # A momentary spike to left channel inherited by right
                    maximal_r = maximal_l
                if maximal_l < maximal_r:
                    # A momentary spike to right channel inherited by left
                    maximal_l = maximal_r
            else:
                # Mono - processing all dadta
                amp = Amplitude.from_data(data)
                gap = amp.value      # For signal test below.
                if amp > maximal:
                    maximal = amp

            # New code January 23, to reset next song's maximal during gap
            if gap == 0.0:
                reset_baseline_count += 1
                if reset_baseline_count == 10:
                    maximal = Amplitude()
                    maximal_l = maximal_r = maximal
                    # print('maximual reset', maximal.value)
            else:
                reset_baseline_count = 0

            # January 24, 2021 separate left and right channels
            if parameter == 'stereo':
                ampl.display(scale=200, mark=maximal_l, fn=VU_METER_LEFT_FNAME)
                ampr.display(scale=200, mark=maximal_r, fn=VU_METER_RIGHT_FNAME)
            else:
                # Mono processing one channel combined sound
                amp.display(scale=200, mark=maximal, fn=VU_METER_FNAME)

    finally:
        stream.stop_stream()
        stream.close()
        audio.terminate()

if __name__ == "__main__":
    main()
Malar answered 20/6, 2021 at 14:45 Comment(0)

© 2022 - 2024 — McMap. All rights reserved.