Binaural Convolution Processing

Setup

• setup a python virtual environment with 64 bit python
• pip install -r requirements.txt

File Structure Overview

The IR-directory contains multiple subdirectories, each representing a person (e.g., Person1, Person2, etc.). Each person's directory contains:

Impulse Response Files (IRs)

• A set of stereo .wav files that correspond to the 7.1.4 channel speaker configuration.
• The file names correspond to the positions in a 7.1.4 surround system:
  • frontleft.wav, frontright.wav, frontcenter.wav
  • sideleft.wav, sideright.wav
  • backleft.wav, backright.wav
  • topfrontleft.wav, topfrontcenter.wav, topfrontright.wav
  • lfe1.wav (low-frequency effects channel)
  • voiceofgod.wav (likely an overhead source)

The top level directory contains

"out" Directory (Located in the Top-Level Directory)

• This subdirectory will store the binaural output after processing for each person

"in" File (Located in the Top-Level Directory)

• A 12-channel .wav file that serves as the input.
• This file will be convolved with the impulse response files from each person's folder to generate personalized binaural audio.

Objective

• The program should iterate through each person's directory, access their 7.1.4 stereo impulse response .wav files, and perform convolution with the 12-channel input file from the "in" directory.
• The resulting binaural output should be stored in each person's "out" directory.

Usage Instructions

1. Place the 12-channel input .wav file inside the "in" directory.
2. Assure the IR-directory is setup properly based on (file structure)[Impulse Response Files (IRs)]
3. Run the Python script to perform convolution and generate binaural audio.
4. The output files will be saved inside "out" directory.

Convolution Method

• The script uses Scipy's signal.convolve function to convolve the input file with the impulse responses.
• Impulse responses are trimmed to 0.5 seconds to remove irrelevant tail noise.
• The output is normalized before being saved.

Running the Script

python conv.py