parse_data.py

import struct
from pathlib import Path
import wave


def parse_data(data):
    # the first 24 bytes are the header
    header = data[:24]

    # the remaining bytes are the data
    body = data[24:]

    # parse the header into 6 32 bit unsigned ints, little endian
    (job_id, sequence_id, finished_flag, num_audio_bytes,
     num_face_bytes, num_timing_bytes) = struct.unpack('<6I', header)

    # the remaining data is split into 3 parts (audio, face, timing)
    # audio comes first, so you can use the num_audio_bytes to split it off
    audio_data = body[:num_audio_bytes]

    # we don't need it yet, but here are the other two parts
    face_data = body[num_audio_bytes:num_audio_bytes + num_face_bytes]
    timing_data = body[num_audio_bytes + num_face_bytes:]

    # the audio part itself has a header of four bytes, but just throw it away
    audio_data = audio_data[4:]
    # the remaining audio data is a series of 16 bit signed ints, little endian
    # here I'm just going to return the raw bytes, stick them together,
    # and put them in a file
    return audio_data


if __name__ == "__main__":
    # empty byte sequence
    audio_data = b''

    for i in range(1, 5):
        filename = f'event_data ({i}).txt'
        # read the raw bytes from the file
        data = Path(filename).read_bytes()
        # parse out the audio data, stick it on the end of the sequence
        audio_data += parse_data(data)

    # write the audio data to a file
    with wave.open('output.wav', 'wb') as f:
        f.setnchannels(1) # single channel, mono
        f.setsampwidth(2) # 2 bytes per sample, i.e, 16 bits
        f.setframerate(22050) # 22050 samples per second
        f.writeframes(audio_data)