Sample extraction script for Roland SPD-SX PRO factory reset data

cleanup.py

"""
Sample extraction script for Roland SPD-SX PRO factory reset data

Resources:
- Factory Reset Data: https://www.roland.com/us/support/by_product/spd-sx_pro/updates_drivers/fe09f0e5-87a8-42e7-b1b8-24cf6280e93b/
- Spec sheet: https://www.roland.com/us/products/spd-sx_pro/specifications/

This code is public domain under the Unlicense: https://unlicense.org
"""

import re
from pathlib import Path
import subprocess

with open("wavePrm.bin", 'rb') as f:
    # File stops having relevant info after this many bytes
    NUM_BYTES = 0x3E330
    database_bytes = f.read(NUM_BYTES)

# Extract sample names, idents from database

# First group captures the sample name with a bit of whitespace,
# the second the ID number
PATTERN = rb"((?:\w|\s|[!?+-.])+)PRELOAD (\d{5})"

samples = dict(
    (id_num.decode('ascii'), samp_name.decode('ascii').strip())
    for samp_name, id_num
    in re.findall(PATTERN, database_bytes)
)

# Process samples
# SMP files are raw PCM. Roland specs told me that we're looking
# at 16-bit signed, 48k samples. I guessed that it was little endian
# and mono.

base = Path.cwd() / "Roland SPD-SX Pro Samples"
base.mkdir(exist_ok=True)

for d_num in range(16):
    d_folder = base / ("D" + str(d_num).rjust(3, '0'))
    d_folder.mkdir(exist_ok=True)

data_folder = Path.cwd() / 'DATA'

for (id_num, d_folder) in ((s.stem[1:], s.parent.stem) for s in data_folder.glob('**/*.SMP')):
    subprocess.run(
        [
            'ffmpeg',
            '-f',
            's16le',
            '-ar',
            '48k',
            '-ac',
            '1',
            '-i',
            str(data_folder / d_folder / f'W{id_num}.SMP'),
            str(base / d_folder / f'{samples[id_num]} (W{id_num}).flac'),
        ],
        check=True
    )