aixander / realtime_pyaudio_fft Goto Github PK

Hello, I noticed when I push the close button, it doesn't work and I tried to track it but I failed.

I'm a beginner so the code level is kind of advanced and I want to make a function contains a loop for making some classes for the range of the frequency like a classifier. So is there any hint for making this? Or where could I use it?

Thank you

Howdy!!! I just wanted to say thank for this repo and the time you spent putting it together. I'll be looking at bringing this into BirdNET-Pi for realtime spectrograms and the development I'm doing around bats :)

My very best regards,
Patrick

Error occured trying to open in Windows 10 with Git Bash

Simpler to just remove those emojis?

Mac os 10.15.2
Python 3.7.7

I can't read the system audio for the visualizer. How did you do it in your teaser? I would like my program to act like this. Thanks!

i use virtualenv:
virtualenv -p /usr/bin/python3.6 venv-3.6
source venv-3.6/bin/activate
pip3 install -r requirements.txt

i changed the requirement file to include versions (i suggest to update the file with versions)
numpy==1.18.4
matplotlib==3.2.1
scipy==1.4.1
pygame==1.9.6
PyAudio==0.2.11

but pyAudio give me error
Collecting PyAudio==0.2.11
src/_portaudiomodule.c:29:10: fatal error: portaudio.h: File o directory non esistente
#include "portaudio.h"
^~~~~~~~~~~~~
compilation terminated.

solved installing portaudio via apt:

sudo apt install libasound-dev portaudio19-dev libportaudio2 libportaudiocpp0

Is it possible to change the x axis to logarithmic? So that you can see Basslines, Vocals more clearly and the hihats and cymbals dont take up 3/4 of the screen.

Hello!

I would like to use this visualizer but with a WAV file as an input instead of a microphone stream. It should be pretty easy to edit the code to use a Wave_read object (from the wave package) instead of the current input, but I can't manage to do it. Perhaps someone here can help me?

Thanks for your wonderful job btw, this is a really good ressource :)

These could be added with a simple implementation of argparse.

On windows is get

It says it's listening from my airpods mic but even when I speak or play music out loud the visualization doesn't change.

Does it listen in from the microphone?

PyAudio was giving some problems doing pip install on W10, i found this post:
i-cant-install-pyaudio-on-windows-how-to-solve-error-microsoft-visual-c-14

Basically:

1. python --version -V
2. Download for your version: https://www.lfd.uci.edu/~gohlke/pythonlibs/#pyaudio
3. pip install PyAudio-0.2.11-cpXX-cpXXm-win_XX.whl

I test it on raspberry cm4 lite
changed the window to 320x240
but the fps is just 10
then I run it on laptop fps is about 50
how can I improve fps on raspberry?
3q

There is a typo in the readme file regarding the installation. The correct command should be:

pip install -r requirements.txt

Unfortunately doesn't work with:

wf = wave.open("my.wav", "rb")

self.stream = p.open(
    format=p.get_format_from_width(wf.getsampwidth()),
    channels=wf.getnchannels(),
    rate=wf.getframerate(),
    output=True,
    stream_callback=self.callback,
)

Hello, I'm running this in Windows 10 with python 3.11.0
I first had issues with the requierements file but removing the required version numbers fixed it (amazing).
The issue is that I want to see my speakers output that are the number 5 in the list, so I use the --device 5 but then this message appears
Input sound settings for device 5 and samplerate None Hz not supported, using defaults...
And then the algorithm defaults to my microphone and works just fine.

I tried changing the sample rate to match the actual device rate (check it in the windows sound config menu) but it says the same "not supported" so I was wondering if theres anything else I'm missing I should change.

Thank you

I saw in your source code that you plan to integrate beat detection. Thus, I have uploaded one of my PhD class projects on music tempo visualization (at https://github.com/xuancong84/TempoVis) for your reference. Unfortunately, my code was in C/C++/Assembly, but it runs extremely fast with little CPU overhead.

site-packages/numpy/core/numeric.py", line 223, in ones
a = empty(shape, dtype, order)
TypeError: 'float' object cannot be interpreted as an index

caused by line 130 of "src/stream_reader_sounddevice.py".

solved by adding a hard cast to int.

self.data_windows_to_buffer = int(data_windows_to_buffer)

Following the README.md operation will run out of this error:
Traceback (most recent call last):
File "/home/junyiwu/Realtime_PyAudio_FFT/Realtime_PyAudio_FFT/run_FFT_analyzer.py", line 4, in
ear = Stream_Analyzer(
File "/home/junyiwu/Realtime_PyAudio_FFT/Realtime_PyAudio_FFT/src/stream_analyzer.py", line 51, in init
self.stream_reader = Stream_Reader(
File "/home/junyiwu/Realtime_PyAudio_FFT/Realtime_PyAudio_FFT/src/stream_reader_sounddevice.py", line 52, in init
with sd.InputStream(samplerate=self.rate,
File "/home/junyiwu/.local/lib/python3.10/site-packages/sounddevice.py", line 1381, in init
_StreamBase.init(self, kind='input', wrap_callback='array',
File "/home/junyiwu/.local/lib/python3.10/site-packages/sounddevice.py", line 777, in init
_get_stream_parameters(kind, device, channels, dtype, latency,
File "/home/junyiwu/.local/lib/python3.10/site-packages/sounddevice.py", line 2571, in _get_stream_parameters
info = query_devices(device)
File "/home/junyiwu/.local/lib/python3.10/site-packages/sounddevice.py", line 569, in query_devices
raise PortAudioError('Error querying device {0}'.format(device))

want to know how to improve
Ubuntu 22.04.1

I run on Pi 4 8GB Ram (64 bit)
Linux 6.1.21-v8+ aarch64 GNU/Linux
LCD 5 inche

Using command: export DISPLAY=:0 && python3 run_FFT_analyzer.py
Show it OK, spectrum not play.

I check sound of pippopeppy is OK, Pygame work with pippopeppy is OK.

My log:
##################################################################################################

Defaulted to using first working mic, Running on:

MIC 1:
defaultHighInputLatency: 0.034829931972789115
defaultHighOutputLatency: 0.034829931972789115
defaultLowInputLatency: 0.008707482993197279
defaultLowOutputLatency: 0.008707482993197279
defaultSampleRate: 44100.0
hostApi: 0
index: 1
maxInputChannels: 32
maxOutputChannels: 32
name: pippopeppy
structVersion: 2

##################################################################################################
Recording from pippopeppy at 41000 Hz
Using (non-overlapping) data-windows of 42 samples (updating at 976.19fps)

Applying temporal smoothing to the FFT features...
Smoothing buffer contains 3 FFT windows (sigma: 1.000) --> min_contribution: 13.534%
Filter weights:
03: 0.233
02: 1.045
01: 1.722
Using FFT_window_size length of 2460 for FFT ---> window_size = 60ms
##################################################################################################

-- Starting live audio stream...

Starting spectrum visualizer...

Super cool program. Congrats.

I'm going to try this with Ubuntu 16.04 (Python 2.7.12) so will have to do a little tweaking.

Tweak #1:

Error message on startup:

~/python/audio$ python run_FFT_analyzer.py

Traceback (most recent call last):
  File "run_FFT_analyzer.py", line 2, in <module>
    from src.stream_analyzer import Stream_Analyzer
ImportError: No module named src.stream_analyzer

To fix this error I used:

~/python/audio$ cd src

~/python/audio/src$ echo "" > __init__.py

TWEAK #2:

I'll be revising this comment from time to time as I run into new issues. For example I will need to add program header stub for UTF-8 encoding due to this new error that popped up after fixing above error:

$ python run_FFT_analyzer.py
Traceback (most recent call last):
  File "run_FFT_analyzer.py", line 58, in <module>
    run_FFT_analyzer()
  File "run_FFT_analyzer.py", line 45, in run_FFT_analyzer
    window_ratio = window_ratio  # Float ratio of the visualizer window. e.g. 24/9
  File "/home/rick/python/audio/src/stream_analyzer.py", line 50, in __init__
    from src.stream_reader_sounddevice import Stream_Reader
  File "/home/rick/python/audio/src/stream_reader_sounddevice.py", line 131
SyntaxError: Non-ASCII character '\xf0' in file /home/rick/python/audio/src/stream_reader_sounddevice.py on line 131, but no encoding declared; see http://python.org/dev/peps/pep-0263/ for details

This is solved by inserting one line:

# -*- coding: utf-8 -*-

TWEAK #3:

The next error that appears:

$ python run_FFT_analyzer.py
Traceback (most recent call last):
  File "run_FFT_analyzer.py", line 58, in <module>
    run_FFT_analyzer()
  File "run_FFT_analyzer.py", line 45, in run_FFT_analyzer
    window_ratio = window_ratio  # Float ratio of the visualizer window. e.g. 24/9
  File "/home/rick/python/audio/src/stream_analyzer.py", line 50, in __init__
    from src.stream_reader_sounddevice import Stream_Reader
  File "/home/rick/python/audio/src/stream_reader_sounddevice.py", line 5, in <module>
    import sounddevice as sd
ImportError: No module named sounddevice

This is more complicated a fix as described here:

• Python import sounddevice as sd (ImportError: No module name sounddevice)

Because the solutions provided use pip or pip3 in Ubuntu which won't work because I only sudo apt install the real solution will be more complicated.

Device 14 are my headphones. I don't want to use the visualizer for my mic. This is what it gives me:
Input sound settings for device 14 and samplerate 44100 Hz not supported, using defaults...
Defaulted to using first working mic, Running on mic 1

Is there any way to make a unsupported device supported?
Or is there anything I'm doing wrong?