find-audio

#!/usr/bin/env python3
# Copyright 2022 by Chris Osborn <fozztexx@fozztexx.com>
#
# This file is part of viddin.
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License at <http://www.gnu.org/licenses/> for
# more details.

import argparse
import os
import librosa
from scipy import signal
import numpy as np
import warnings

def build_argparser():
  parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
  parser.add_argument("sample", help="audio to search for")
  parser.add_argument("files", nargs="+", help="Video file(s) to scan")
  parser.add_argument("--flag", action="store_true", help="flag to do something")
  return parser

def main():
  args = build_argparser().parse_args()

  # Disable librosa complaining about having to use audioread
  warnings.filterwarnings('ignore')

  sample_rate = 22050
  sample, _ = librosa.load(args.sample, sr=sample_rate, mono=True)

  zeros = np.zeros(len(sample))
  silent = np.isclose(sample, zeros, atol=0.001)
  trim = np.where(silent == False)[0]
  sample = sample[trim[0]:trim[-1]]
  sample_len = len(sample) / sample_rate
  
  for path in args.files:
    within, _ = librosa.load(path, sr=sample_rate, mono=True)
    c = signal.correlate(within, sample, mode="valid", method="fft")
    peak = np.argmax(c)
    offset = round(peak / sample_rate, 2)
    base, ext = os.path.splitext(path)
    with open(base + ".aud", "w") as f:
      print("%0.3f %0.3f" % (offset, offset + sample_len), file=f)
    
  return

if __name__ == '__main__':
  exit(main() or 0)