Merge branch '100-unable-to-scale-blank-hoa3-file-in-osba-item-generation' into 'main' (0375de2a) · Commits · IVAS Codec Public Collaboration / IVAS Processing Scripts

ivas_processing_scripts/generation/generate_ismN_items.py

+32 −5

Original line number	Diff line number	Diff line
		@@ -30,6 +30,7 @@
		# the United Nations Convention on Contracts on the International Sales of Goods.
		#
		import logging
		import sys
		from itertools import groupby, repeat
		from pathlib import Path

		@@ -249,6 +250,23 @@ def generate_ismN_scene(
		if isinstance(scene["level"], list)
		else scene["level"]
		)

		# check if level is "-Inf" string or "N/A" or "n/a" or "NaN"
		if isinstance(level, str):
		if level.lower() in ["-inf"]:
		level = -np.inf
		elif level.lower() in ["none", "no", "", "null", "n/a", "na", "nan"]:
		level = None
		else:
		# convert to float
		try:
		level = float(level)
		except ValueError:
		logger.error(f"Error: Level {level} is not a valid entry!")
		sys.exit(-1)
		elif isinstance(level, (int, float)):
		# do nothing, it is already a number
		pass
		else:
		level = -26

		@@ -270,7 +288,16 @@ def generate_ismN_scene(
		x.fs = cfg.fs

		# adjust the level of the audio source file (need to convert to MONO first)
		x_temp = audio.ChannelBasedAudio("MONO") # create a temporary mono audio object
		if np.isinf(level):
		# set all channels to zero
		x.audio = np.zeros_like(x.audio)
		elif level is None:
		# do not change the level of the audio source signal
		logger.info("-- Level of the audio source signal is not changed")
		else:
		x_temp = audio.ChannelBasedAudio(
		"MONO"
		) # create a temporary mono audio object
		x_temp.audio = x.audio.copy()
		x_temp.fs = x.fs
		x_temp.audio, _ = loudness_norm(x_temp, level, loudness_format="MONO")

+25 −1

Original line number	Diff line number	Diff line
		@@ -247,6 +247,23 @@ def generate_MASA_scene(
		if isinstance(scene["level"], list)
		else scene["level"]
		)

		# check if level is "-Inf" string or "N/A" or "n/a" or "NaN"
		if isinstance(level, str):
		if level.lower() in ["-inf"]:
		level = -np.inf
		elif level.lower() in ["none", "no", "", "null", "n/a", "na", "nan"]:
		level = None
		else:
		# convert to float
		try:
		level = float(level)
		except ValueError:
		logger.error(f"Error: Level {level} is not a valid entry!")
		sys.exit(-1)
		elif isinstance(level, (int, float)):
		# do nothing, it is already a number
		pass
		else:
		level = -26

		@@ -299,6 +316,13 @@ def generate_MASA_scene(
		x = reverb_hoa3(x, IR, mode=None)

		# adjust the level of the FOA/HOA2/HOA3 signal
		if np.isinf(level):
		# set all channels to zero
		x.audio = np.zeros_like(x.audio)
		elif level is None:
		# do not change the level of the audio source signal
		logger.info("-- Level of the audio source signal is not changed")
		else:
		x.audio, _ = loudness_norm(x, level, loudness_format="STEREO")

		# ensure the length of the audio source signal is a multiple of 20ms

+25 −1

Original line number	Diff line number	Diff line
		@@ -247,6 +247,23 @@ def generate_MC_scene(
		if isinstance(scene["level"], list)
		else scene["level"]
		)

		# check if level is "-Inf" string or "N/A" or "n/a" or "NaN"
		if isinstance(level, str):
		if level.lower() in ["-inf"]:
		level = -np.inf
		elif level.lower() in ["none", "no", "", "null", "n/a", "na", "nan"]:
		level = None
		else:
		# convert to float
		try:
		level = float(level)
		except ValueError:
		logger.error(f"Error: Level {level} is not a valid entry!")
		sys.exit(-1)
		elif isinstance(level, (int, float)):
		# do nothing, it is already a number
		pass
		else:
		level = -26

		@@ -299,6 +316,13 @@ def generate_MC_scene(
		x = reverb_hoa3(x, IR, mode=None)

		# adjust the level of the FOA/HOA2/HOA3 signal
		if np.isinf(level):
		# set all channels to zero
		x.audio = np.zeros_like(x.audio)
		elif level is None:
		# do not change the level of the audio source signal
		logger.info("-- Level of the audio source signal is not changed")
		else:
		x.audio, _ = loudness_norm(x, level, loudness_format="STEREO")

		# ensure the length of the audio source signal is a multiple of 20ms

+27 −3

Original line number	Diff line number	Diff line
		@@ -244,6 +244,23 @@ def generate_OMASA_scene(
		if isinstance(scene["level"], list)
		else scene["level"]
		)

		# check if level is "-Inf" string or "N/A" or "n/a" or "NaN"
		if isinstance(level, str):
		if level.lower() in ["-inf"]:
		level = -np.inf
		elif level.lower() in ["none", "no", "", "null", "n/a", "na", "nan"]:
		level = None
		else:
		# convert to float
		try:
		level = float(level)
		except ValueError:
		logger.error(f"Error: Level {level} is not a valid entry!")
		sys.exit(-1)
		elif isinstance(level, (int, float)):
		# do nothing, it is already a number
		pass
		else:
		level = -26

		@@ -284,6 +301,13 @@ def generate_OMASA_scene(
		x.fs = cfg.fs

		# adjust the level of the source file
		if np.isinf(level):
		# set all channels to zero
		x.audio = np.zeros_like(x.audio)
		elif level is None:
		# do not change the level of the audio source signal
		logger.info("-- Level of the audio source signal is not changed")
		else:
		if fmt in ["FOA", "HOA2", "HOA3"]:
		x.audio, _ = loudness_norm(x, level, loudness_format="STEREO")
		else:

+27 −3

Original line number	Diff line number	Diff line
		@@ -235,6 +235,23 @@ def generate_OSBA_scene(
		if isinstance(scene["level"], list)
		else scene["level"]
		)

		# check if level is "-Inf" string or "N/A" or "n/a" or "NaN"
		if isinstance(level, str):
		if level.lower() in ["-inf"]:
		level = -np.inf
		elif level.lower() in ["none", "no", "", "null", "n/a", "na", "nan"]:
		level = None
		else:
		# convert to float
		try:
		level = float(level)
		except ValueError:
		logger.error(f"Error: Level {level} is not a valid entry!")
		sys.exit(-1)
		elif isinstance(level, (int, float)):
		# do nothing, it is already a number
		pass
		else:
		level = -26

		@@ -275,6 +292,13 @@ def generate_OSBA_scene(
		x.fs = cfg.fs

		# adjust the level of the source file
		if np.isinf(level):
		# set all channels to zero
		x.audio = np.zeros_like(x.audio)
		elif level is None:
		# do not change the level of the audio source signal
		logger.info("-- Level of the audio source signal is not changed")
		else:
		if fmt in ["FOA", "HOA2", "HOA3"]:
		x.audio, _ = loudness_norm(x, level, loudness_format="STEREO")
		else: