Merge branch '10-bs1770demo-can-produce-nan-as-loudness-result' into 'main' (0105e07f) · Commits · IVAS Codec Public Collaboration / IVAS Processing Scripts

ivas_processing_scripts/audiotools/wrappers/bs1770.py

+30 −4

Original line number	Diff line number	Diff line
		@@ -33,6 +33,7 @@
		import copy
		import logging
		import re
		import subprocess as sp
		from pathlib import Path
		from tempfile import TemporaryDirectory
		from typing import Optional, Tuple, Union
		@@ -53,6 +54,7 @@ logger.setLevel(logging.DEBUG)
		def bs1770demo(
		input: audio.Audio,
		target_loudness: Optional[float] = -26,
		rms: Optional[bool] = False,
		) -> Tuple[float, float]:
		"""
		Wrapper for ITU-R BS.1770-4, requires bs1770demo binary
		@@ -82,6 +84,16 @@ def bs1770demo(
		else:
		binary = find_binary("bs1770demo")

		# checking if the new binary (with '-rms') is used
		with TemporaryDirectory() as tmp_dir_test:
		tmp_dir_test = Path(tmp_dir_test)
		result = sp.run(str(binary), shell=True, stdout=sp.PIPE, stderr=sp.PIPE)
		result_stdout = result.stdout.decode("utf-8")
		if "-rms" not in result_stdout:
		raise RuntimeError(
		'An bs1770demo executable without RMS support (cmdl option "-rms") was detected. Please update the bs1770demo executable. See bin/README.md for details.'
		)

		if not isinstance(input, audio.BinauralAudio) and not isinstance(
		input, audio.ChannelBasedAudio
		):
		@@ -131,6 +143,9 @@ def bs1770demo(
		# write temporary file
		write(tmp_file, tmp_sig, 48000)

		# using rms if true
		if rms:
		cmd.insert(1, "-rms")
		# run command
		result = run(cmd, logger=logger)

		@@ -150,6 +165,7 @@ def get_loudness(
		input: audio.Audio,
		target_loudness: Optional[float] = -26,
		loudness_format: Optional[str] = None,
		rms: Optional[bool] = False,
		) -> Tuple[float, float]:
		"""
		Loudness measurement using ITU-R BS.1770-4
		@@ -196,7 +212,9 @@ def get_loudness(
		convert.format_conversion(input, tmp)
		else:
		tmp.audio = input.audio

		if rms:
		return bs1770demo(tmp, target_loudness, rms=True)
		else:
		return bs1770demo(tmp, target_loudness)


		@@ -204,6 +222,7 @@ def loudness_norm(
		input: audio.Audio,
		target_loudness: Optional[float] = -26,
		loudness_format: Optional[str] = None,
		rms: Optional[bool] = False,
		) -> np.ndarray:
		"""
		Iterative loudness normalization using ITU-R BS.1770-4
		@@ -232,7 +251,10 @@ def loudness_norm(

		while np.abs(measured_loudness - target_loudness) > 0.5 and num_iter < 10:
		measured_loudness, scale_factor_new = get_loudness(
		input, target_loudness, loudness_format
		input,
		target_loudness,
		loudness_format,
		rms=rms,
		)

		# scale input
		@@ -257,6 +279,7 @@ def scale_files(
		loudness: float,
		fs: Optional[int] = 48000,
		in_meta: Optional[list] = None,
		rms: Optional[bool] = False,
		) -> None:
		"""
		Scales audio files to desired loudness
		@@ -291,6 +314,9 @@ def scale_files(
		audio_obj = audio.fromfile(fmt, file, fs)

		# adjust loudness
		if rms:
		scaled_audio = loudness_norm(audio_obj, loudness, rms=True)
		else:
		scaled_audio = loudness_norm(audio_obj, loudness)

		# write into file

ivas_processing_scripts/processing/preprocessing_2.py

+3 −1

Original line number	Diff line number	Diff line
		@@ -149,7 +149,9 @@ class Preprocessing2(Processing):
		)[: len(audio_object.audio)]

		# scale background noise to desired loudness based on output format
		noise_object.audio = loudness_norm(noise_object, loudness_noise, out_format)
		noise_object.audio = loudness_norm(
		noise_object, loudness_noise, out_format, rms=True
		)

		# add array to signal
		audio_object.audio = noise_object.audio + audio_object.audio