Audiosockets

Prerequisites

Basic Compilers

You might not have gcc or make, so make sure to install them

sudo apt install gcc make

Install PortAudio

In case you do not have it, you will need to install PortAudio. To do that, first run

sudo apt install libasound-dev

Then download the latest PortAudio at its Downloads Page, e.g. (2021-04-23)

wget http://files.portaudio.com/archives/pa_stable_v190700_20210406.tgz

Then unzip it (tar -xvf <file>), go into the directory and run

./configure
make
sudo make install

sudo apt install portaudio19-dev

yay -S portaudio

Installation

Clone the repository, go into the repository and run

pip install -e package

Quickstart

Configuration File

Here are the instructions for using v3. First create a configuration JSON file which contains the following structure

{
    "PORT": 5050,
    "HEADER": 64,
    "FORMAT": "utf-8",
    "DISCONNECT_MSG": "DISCONNECT",
    "logging_format": "%(asctime)s - %(message)s", 
    "logging_level": "info"
}

This configuration file establishes a few parameters:

• PORT: The port number where the server will be running, and where the clients will access
• HEADER: A large enough byte size for the header to receive a header
• FORMAT: A decoding format for the strings
• DISCONNECT_MSG: A disconnecting message that will be agreed upon every node in the network to understand when something will disconnect from the server
• logging_format: A format for viewing the logging module
• logging_level: The level for the logging module. That may be info, debug, warning, error, critical or any of those choices with any letter capitalization

Usage

For using a recorder, simply start up a server on a script by running

from audiosockets import MailmanSocket

server = MailmanSocket()
server.start()

Start recording something with

from audiosockets import RecorderSocket

recorder = RecorderSocket()
recorder.start()

Finally, if you wish to process the data, create a new processor by inheriting from BaseProcessorSocket and overwrite the method process_data(data) in order to process the data in whichever way you wish.

from audiosockets import BaseProcessorSocket
from audiosockets.utils import LogMelSpectrogram

class MyProcessor(BaseProcessorSocket):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

    def process_data(self, data):
        audio, fs = data["data"], data["fs"]
        lms = LogMelSpectrogram(fs)(audio)
        print(lms)