correction of ranges of azimuth and elevation maintaining speed of talker movement (25c01c51) · Commits · IVAS Codec Public Collaboration / IVAS Processing Scripts

ivas_processing_scripts/generation/process_ism1_items.py

+23 −19

Original line number	Diff line number	Diff line
		@@ -267,9 +267,10 @@ def generate_ism1_scene(

		N_frames = int(len(y.audio) / y.fs * 50)

		# read azimuth information and create array
		# read azimuth information and convert to an array
		if isinstance(source_azi, str):
		if ":" in source_azi:
		# start with the initial azimuth value and apply step N_frames times
		source_azi = source_azi.split(":")
		azi = np.linspace(
		float(eval(source_azi[0])),
		@@ -277,46 +278,49 @@ def generate_ism1_scene(
		N_frames
		)
		else:
		azi = np.array(float(eval(source_azi)), ndmin=1)
		# replicate static azimuth value N_frames times
		azi = np.repeat(float(eval(source_azi)), N_frames)
		else:
		azi = np.array(source_azi, ndmin=1)[:N_frames]

		if len(azi) < N_frames:
		# replicate the last elevation
		azi = np.append(azi, np.full(N_frames - len(azi), azi[-1]))
		# replicate static azimuth value N_frames times
		azi = np.repeat(float(source_azi), N_frames)

		# convert azimuth from 0 .. 360 to -180 .. +180
		azi = (azi + 180) % 360 - 180

		# check if azimuth is from -180 .. +180
		# check, if azimuth is from -180 .. +180
		if any(azi > 180) or any(azi < -180):
		logger.error(
		f"Incorrect value(s) of azimuth: {azi[(azi > 180) \| (azi < -180)]}"
		)

		# read elevation information and create array
		# read elevation information and convert to an array
		if isinstance(source_ele, str):
		if ":" in source_ele:
		# convert into array (initial_value:step:stop_value)
		# note: the stop_value value is +-90 degrees depending on the sign of the step
		source_ele = source_ele.split(":")
		ele = np.linspace(
		ele = np.arange(
		float(eval(source_ele[0])),
		float(eval(source_ele[2])),
		N_frames
		)
		else:
		ele = np.array(float(eval(source_ele)), ndmin=1)
		else:
		ele = np.array(source_ele, ndmin=1)[:N_frames]
		np.sign(float(eval(source_ele[1]))) * 90,
		float(eval(source_ele[1]))
		)[:N_frames]

		# repeat the last elevation value, if array is shorter than N_frames
		if len(ele) < N_frames:
		# replicate the last elevation
		ele = np.append(ele, np.full(N_frames - len(ele), ele[-1]))
		else:
		# replicate static elevation value N_frames times
		ele = np.repeat(float(eval(source_ele)), N_frames)
		else:
		# replicate static elevation value N_frames times
		ele = np.repeat(float(source_ele), N_frames)

		# check if elevation is from -90 .. +90
		if any(ele > 90) or any(ele < -90):
		logger.error(
		f"Incorrect value(s) of elevation: {ele[(ele > 90) \| (ele < -90)]}"
		)

		# arrange all metadata fields column-wise into a matrix
		y_meta = np.column_stack((azi, ele))

ivas_processing_scripts/generation/process_ism2_items.py

+21 −18

Original line number	Diff line number	Diff line
		@@ -302,9 +302,10 @@ def generate_ism2_scene(

		N_frames = int(len(y.audio) / y.fs * 50)

		# read azimuth information and create array
		# read azimuth information and convert to an array
		if isinstance(source_azi, str):
		if ":" in source_azi:
		# start with the initial azimuth value and apply step N_frames times
		source_azi = source_azi.split(":")
		azi = np.linspace(
		float(eval(source_azi[0])),
		@@ -312,13 +313,11 @@ def generate_ism2_scene(
		N_frames
		)
		else:
		azi = np.array(float(eval(source_azi)), ndmin=1)
		# replicate static azimuth value N_frames times
		azi = np.repeat(float(eval(source_azi)), N_frames)
		else:
		azi = np.array(source_azi, ndmin=1)[:N_frames]

		if len(azi) < N_frames:
		# replicate the last elevation
		azi = np.append(azi, np.full(N_frames - len(azi), azi[-1]))
		# replicate static azimuth value N_frames times
		azi = np.repeat(float(source_azi), N_frames)

		# convert azimuth from 0 .. 360 to -180 .. +180
		azi = (azi + 180) % 360 - 180
		@@ -329,23 +328,27 @@ def generate_ism2_scene(
		f"Incorrect value(s) of azimuth: {azi[(azi > 180) \| (azi < -180)]}"
		)

		# read elevation information and create array
		# read elevation information and convert to an array
		if isinstance(source_ele, str):
		if ":" in source_ele:
		# convert into array (initial_value:step:stop_value)
		# note: the stop_value value is +-90 degrees depending on the sign of the step
		source_ele = source_ele.split(":")
		ele = np.linspace(
		ele = np.arange(
		float(eval(source_ele[0])),
		float(eval(source_ele[2])),
		N_frames
		)
		else:
		ele = np.array(float(eval(source_ele)), ndmin=1)
		else:
		ele = np.array(source_ele, ndmin=1)[:N_frames]
		np.sign(float(eval(source_ele[1]))) * 90,
		float(eval(source_ele[1]))
		)[:N_frames]

		# repeat the last elevation value, if array is shorter than N_frames
		if len(ele) < N_frames:
		# replicate the last elevation
		ele = np.append(ele, np.full(N_frames - len(ele), ele[-1]))
		else:
		# replicate static elevation value N_frames times
		ele = np.repeat(float(eval(source_ele)), N_frames)
		else:
		# replicate static elevation value N_frames times
		ele = np.repeat(float(source_ele), N_frames)

		# check if elevation is from -90 .. +90
		if any(ele > 90) or any(ele < -90):