Overview

PyAWECore Library is a Pythonic wrapper or abstraction for AWE Core. It allows the use of AWE Core in Python scripts or programs.

...

import awecorelib
import wave

signalflow_file = "my_signalflow.awb"
wave_input_file = "my_inputfile.wav"

# Initialization of AWE Core instances and loading an AWB file
awe = awecorelib.init_from_awb(signalflow_file)
awb = awe.load(signalflow_file)

collected_data = bytearray(b'')
with wave.open(wave_input_file, 'r') as wav_in:
    data_in = wav_in.readframes(awb.layout_info.fundamental_blocksize)
    while data_in:
        data_out = awe.pump(data_in, 2)  # assuming a stereo-wav file here!
        collected_data.extend(data_out)
        data_in = wav_in.readframes(awb.layout_info.fundamental_blocksize)

print(f"Processing done: {len(collected_data)} bytes of audio data obtained")

Limitations

• For license protection purposes, default builds of PyAWECore Library impose a 3-hour timeout upon the AWE Core instance. Hence, in the code example above, users would need to re-instantiate the line awe = awecorelib.init_from_awb(signalflow_file) at least once every three hours. Please contact your sales representative for a custom build of PyAWECore Library if you need AWE instances to run for longer durations.

• As of the PyAWECore Library 2.1.2 release, the TensorFlow Lite Micro and TensorFlow Lite Micro V2 Modules are not supported with the Linux (x86) PyAWECore Library wheel. However, this module is supported with the Windows wheel. This issue will be fixed in a future release.

• As of the PyAWECore Library 2.1.2 release, the ONNX Runtime Module is not supported with PyAWECore Library. Support will be added in a future release.

Installation

Currently, PyAWECore Library is provided as Python wheel files. Those files are part of the Windows AWE Designer installation, and will be placed into a directory pyawecore inside the base installation directory; for example, under C:\DSP Concepts\AWE Designer 8.D.2.6 Pro\pyawecore.

Those wheel files have to be installed with a Python package manager like pip (see “Installation of Wheels” below).

Prerequisites

A Python installation is required. Please use a Python version that corresponds to the Python version mentioned in the wheel name, e.g., cp311.

...

For WSL (Windows Linux Subsystem) and other Linux systems, Python 3.10 is currently used.

Installation of Wheels

Typically, Python packages are installed into "environments" and not into the system-wide Python installation. There are many ways to handle those environments, like pyenv or conda or simply pip - which also comes with the Python installation by default.

...

Please note that you need to be connected to the internet when running this installation command as further Python package dependencies are retrieved.

Command-Line Scripts

Once the Python environment is activated and the PyAWECore Library packages are installed, some scripts are available:

• (aweenv) awecorelib-docs - this shows the documentation in HTML (in your system’s browser)

• (aweenv) awecorelib-filepump - this can read audio data from a WAV file and “pump it” through a design (see below)

• (aweenv) awecorelib-config - this shows the underlying AWE Core build and version information

Processing Audio

As mentioned above, PyAWECore Library's API can be used to implement programs covering many different use-cases. One of the typical tasks is to process an audio file through an AWE design. PyAWECore Library already provides a rudimentary implementation of such a program, the command-line script awecorelib-filepump mentioned above.

Sine Wave Generator

A simple sine wave generator layout (provided in pyawecore/Designs/sine_generator.png) can be used to generate a stereo WAV file. This first channel contains a clean sine tone, and the second channel is distorted with a pink noise.

...

This generates a 4-second-long audio signal (16000 / 64 = 4ms * 1000 = 4s) in the file my_sine_output.wav.

Passthrough with Attenuation

A sine stereo input signal is passed through the design pyawecore/Designs/passthrough.awj, which attenuates the first channel by minus 10dB.

...

This generates the file my_passthrough_output.wav, which should have the same content as ./Audio/output_processed_audio.wav.

Passthrough and Controlling the Attenuation

The same passthrough design is used now, but programmatically the gain on the second channel is decreased during processing.

...

You should see a step-wise decrease of amplitude in the second channel in file passthrough_output_controlled.wav, as in the following waveform plot.

...

PyAWECore Library Package Documentation

Extensive documentation for PyAWECore Library can be viewed after installation, by running the script awecorelib-docs and clicking the link to open the documentation in your web browser.

PyAWECore Library Application Notes

An application note for showing how to use PyAWECore Library with Tensorflow for machine learning feature extraction and training is provided here: (8.D.2.6) Audio Weaver Tensorflow Tutorial - Simple Keyword Recognition