split audio files not evenly

split.py

#!/usr/bin/env python3
# Copyright 2021 Marcin Zepp <nircek-2103@protonmail.com>
# SPDX-License-Identifier: MIT

import sys
import subprocess
import re
import math

opus, oldmagic, quiet, magic = False, False, True, True
TARGET = 1e-2
assert len(sys.argv) >= 2
for arg in sys.argv[1:-1]:
	if arg == '-opus':
		opus = True
	elif arg == '-nomagic':
		magic = False
	elif arg == '-debug':
		quiet = False
	elif arg == '-oldmagic':
		oldmagic = True
	else:
		raise Exception(f'arg `{arg}` not known')
inp = sys.argv[-1]

def nice(x):	
	return f'{int(x/3600)}:{int((x%3600)/60):02d}:{x%60:05.2f}'
def getDuration(file):
	return float(re.search(b'\nduration=([\d.]+)\n', subprocess.check_output([*'ffprobe -v error -show_format -show_streams'.split(), file])).group(1).decode())

duration = getDuration(inp)
print(f'Duration: {nice(duration)} ({duration})')


i = 2
while i <= 10 or ms > 3*60:
	print(f'{i})', end='\t')
	s = (duration-i*(i-1)//2)/i # sum(range(i))
	ms = s + i - 1
	print(nice(s), end=' \t')
	print(nice(ms))
	i += 1

parts = int(input(':'))
print()


inps = inp.split('.')
name, ext = '.'.join(inps[:-1]), inps[-1]
last = None

MAGIC_OFFSET = 0
if oldmagic:
	MAGIC_OFFSET = (duration*(1/122-(1/(parts**3))/2.4)*math.log(parts)+duration/100*8)/1000 if magic else 0
elif magic:
	print('init', end='(')
	for i in range(1, 5):
		off = (duration+MAGIC_OFFSET-parts*(parts-1)//2)/parts
		ss = (parts-1)*(off+(parts-2)/2)
		exp = off+parts-1
		ss = f'{ss}'
		sys.stdout.flush()
		outp = f'{name}-0.{ext}'
		outo = f'{name}-0.opus'
		if not opus:
			subprocess.call(['ffmpeg', '-loglevel', 'quiet', '-hide_banner', '-y', '-ss', ss, '-i', inp, '-c', 'copy', outp])
			last = outp
		else:
			# https://stackoverflow.com/a/39671913/6732111
			subprocess.call(['ffmpeg', '-loglevel', 'quiet', '-hide_banner', '-y', '-ss', ss, '-i', inp, *'-f wav -flags +bitexact -acodec pcm_s16le -ar 22050 -ac 1'.split(), outp])
			subprocess.call(['opusenc', '--quiet', outp, outo], stderr=subprocess.DEVNULL if quiet else None)
			subprocess.call(['rm', outp])
			last = outo
		dur = getDuration(last)
		subprocess.call(['rm', last])
		print(f'{"" if i == 1 else "=>"}{dur-exp:.4f}', end='')
		if abs(dur-exp) < TARGET:
			break
		MAGIC_OFFSET += (dur-exp)/i/i
		sys.stdout.flush()
		# for e in *.opus; do ffprobe -i $e -show_entries format=duration -v quiet -of csv="p=0"; done
	print(')')
print(f'Using {MAGIC_OFFSET=}.\n')

off = (duration+MAGIC_OFFSET-parts*(parts-1)//2)/parts
exp = off+parts-1
last = None

for i in range(parts):
	ss = i*off+(i-1)*i//2
	t = off+i
	e = ss + t
	ss, t = [f'{x}' for x in [ss, t]]
	print(f'{100*i//parts}%', end=' ' if quiet else '\n')
	if quiet:
		sys.stdout.flush()
	outp = f'{name}-{str(i+1)}.{ext}'
	outo = f'{name}-{str(i+1)}.opus'
	if not opus:
		subprocess.call(['ffmpeg', *(['-loglevel', 'quiet', '-hide_banner'] if quiet else []), '-y', '-ss', ss, '-i', inp, '-c', 'copy', *(['-t', t] if i != parts-1 else []), outp])
		last = outp
	else:
		# https://stackoverflow.com/a/39671913/6732111
		subprocess.call(['ffmpeg', *(['-loglevel', 'quiet', '-hide_banner'] if quiet else []), '-y', '-ss', ss, '-i', inp, *'-f wav -flags +bitexact -acodec pcm_s16le -ar 22050 -ac 1'.split(), *(['-t', t] if i != parts-1 else []), outp])
		subprocess.call(['opusenc', *(['--quiet'] if quiet else []), outp, outo], stderr=subprocess.DEVNULL if quiet else None)
		subprocess.call(['rm', outp])
		last = outo
		
print('100%\n')
if last is not None:
	dur = getDuration(last)
	print(f'Missed target by {dur-exp:.4f} s (was {dur:.4f}, expected {exp:.4f}).')