tyiannak / paura

Python AUdio Recording and Analysis (paura)

Welcome 👋

I am a Principal Researcher at the Multimedia Analysis Group of the Institute of Informatics and Telecommunication< of the National Center for Scientific Research "Demokritos" and a Machine Learning Director at Behavioral Signals. You can find more info about me at my web page

paura's Issues

README doesn't include need for `ffmpeg` install nor for `plotly` (pip) install

In my attempt to use paura in the environment I use (WSL Ubuntu),

bballdave025@MY_MACHINE:~/programs/paura$ uname -a
Linux MY_MACHINE 4.4.0-18362-Microsoft #836-Microsoft Mon May 05 16:04:00 PST 2020 x86_64 x86_64 x86_64 GNU/Linux
bballdave025@MY_MACHINE:~/programs/paura$ bash --version | head -n 1
GNU bash, version 4.4.20(1)-release (x86_64-pc-linux-gnu)
bballdave025@MY_MACHINE:~/programs/paura$ python3 -V
Python 3.6.9

I went through the full installation steps (including installing pyAudioAnalysis). However, I ran into the following problems during my first attempt to run paura. (I realize there's no --help option, but I wanted to see if it would run). These are my compolete paura steps.

bballdave025@MY_MACHINE:~/programs$ git clone https://github.com/tyiannak/paura.git
Cloning into 'paura'...
remote: Enumerating objects: 137, done.
remote: Counting objects: 100% (137/137), done.
remote: Compressing objects: 100% (96/96), done.
remote: Total 169 (delta 78), reused 93 (delta 40), pack-reused 32
Receiving objects: 100% (169/169), 9.92 MiB | 11.73 MiB/s, done.
Resolving deltas: 100% (94/94), done.
bballdave025@MY_MACHINE:~/programs$ cd paura
bballdave025@MY_MACHINE:~/programs/paura$ ls
LICENSE  README.md  model  modelMEANS  paura.py  paura_lite.py  requirements.txt
bballdave025@MY_MACHINE:~/programs/paura$ pip3 install -r ./requirements.txt
bballdave025@MY_MACHINE:~/programs/paura$ python3 paura.py --help
/home/bballdave025/.local/lib/python3.6/site-packages/pydub/utils.py:165: RuntimeWarning: Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work
  warn("Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work", RuntimeWarning)
Traceback (most recent call last):
  File "paura.py", line 10, in <module>
    from pyAudioAnalysis import audioTrainTest as aT
  File "/mnt/c/David/audio_annotation/pyAudioAnalysis_probablyUsed/program_wsl/pyAudioAnalysis/pyAudioAnalysis/audioTrainTest.py", line 17, in <module>
    import plotly
ModuleNotFoundError: No module named 'plotly'
bballdave025@MY_MACHINE:~/programs/paura$ 

I prefer ffmpeg, so I fixed the first part by checking for and then running the install for it.

bballdave025@MY_MACHINE:~/programs/paura$ which ffmpeg
bballdave025@MY_MACHINE:~/programs/paura$ # Nothing
bballdave025@MY_MACHINE:~/programs/paura$ sudo apt-get -y install ffmpeg

Then I fixed the second problem with

bballdave025@MY_MACHINE:~/programs/paura$ pip3 install plotly

After that, things seemed to work fine.

Increase legnth of segments

The audio segments are classified after 1-second length. I want to classify audio after 5 seconds as music/speech etc. How can i do that?

opencv?

Hi @tyiannak!

I the README.md, you require installation of OpenCV, but I couldn't find any calls to it in the repo.

Could you please explain why this is a dependency?

how does this silence detection method work (is there a documented example)?

I like this project a lot.
I've heard of silence detection using RMS
what is this method?

Activity Detection:

               #what's energy100? sum of the squares * 100 divided by the sample rate? 
               energy100 = (100*numpy.sum(midTermBuffer * midTermBuffer) 
                    / (midTermBuffer.shape[0] * 32000 * 32000))     
                if count < 10:           #sample the first few seconds of silence
                    energy100_buffer_zero.append(energy100)                    
                    mean_energy100_zero = numpy.mean(numpy.array(energy100_buffer_zero))
                else:
                    mean_energy100_zero = numpy.mean(numpy.array(energy100_buffer_zero))
                    if (energy100 < 1.2 * mean_energy100_zero): #new silence

My main question: why this

(100*numpy.sum(midTermBuffer * midTermBuffer) / (midTermBuffer.shape[0] * 32000 * 32000))

and not RMS?

Any help with this would be much appreciated!! (is it in your book??)

Will paura.py be transformed for python 3.x and will paura support windows OS?

Will paura.py be transformed for python 3.x and will paura support windows OS?

Using paura in audio-ml-lab docker environment

Hi, I would like to build small raspberry sensors for audio recording. Those should be composed by docker and store the sensed audio data as files on a network drive. Those files will be processed by several other analysis tiers to create meaningful information.
If I try to run paura.py it gives me this error:

libdc1394 error: Failed to initialize libdc1394
Press Ctr+C to stop recording
ALSA lib confmisc.c:768:(parse_card) cannot find card '0'
ALSA lib conf.c:4248:(_snd_config_evaluate) function snd_func_card_driver returned error: No such file or directory

seems like I have not set up the docker container properly?

Question: RasperryPI is supported?

Hi,
I'd like to implement a "smart object" with this library (like a stripe of led connected with RasperryPi).
Had someone tried to use it on a rasperry pi ?
Some articles/suggestion to look for?