check_used_audio.py

from lxml import etree
import glob
from pathlib import Path

SOUND_DIR = "sound/"
AUDIO_DIR = SOUND_DIR + "audio/"

unused_audio = list(map(lambda x: x.replace("/", "\\"), glob.iglob(f"{AUDIO_DIR}/*")))

for sound_info_file in Path(".").glob(f"{SOUND_DIR}*.sound.gmx"):
    sound_info = etree.parse(sound_info_file)

    sound_file_elem = sound_info.find("/origname")
    assert sound_file_elem is not None
    sound_file = sound_file_elem.text
    sound_file_obj = Path(sound_file)

    if not sound_file_obj.exists():
        audio_file_relpath = str(sound_file_obj.relative_to(AUDIO_DIR))
        unused_audio.append(f"{str(sound_info_file)} (no corresponding file {audio_file_relpath})")
        continue
    unused_audio.remove(sound_file)

for idx, file in enumerate(unused_audio):
    file = unused_audio[idx] = Path(file)
    if file.suffix != ".wav" and file.with_suffix(".wav").exists():
        file = file.with_name(file.name + " (has corresponding wav file)")
    print(file.relative_to(AUDIO_DIR, walk_up=True))