Merge branch 'main' of ssh://forge.3gpp.org:29419/ivas-codec-pc/ivas-codec... (57f82fcb) · Commits · IVAS Codec Public Collaboration / IVAS Codec

.gitlab-ci.yml

+3 −0

Original line number	Diff line number	Diff line
		@@ -1251,6 +1251,7 @@ ivas-conformance:
		- cp -force IVAS_cod.exe IVAS_cod_ref.exe
		- cp -force IVAS_dec.exe IVAS_dec_ref.exe
		- cp -force IVAS_rend.exe IVAS_rend_ref.exe
		- cp -force ISAR_post_rend.exe ISAR_post_rend_ref.exe

		# Reference creation
		- python scripts/prepare_combined_format_inputs.py
		@@ -1316,6 +1317,7 @@ ivas-conformance:
		expire_in: 1 week
		when: always
		paths:
		- report_cmd.html
		- report-junit.xml
		- report.html
		- Readme_IVAS_dec.txt
		@@ -1344,6 +1346,7 @@ ivas-conformance-linux:
		- cp IVAS_cod IVAS_cod_ref
		- cp IVAS_dec IVAS_dec_ref
		- cp IVAS_rend IVAS_rend_ref
		- cp ISAR_post_rend ISAR_post_rend_ref

		# Reference creation
		- python3 scripts/prepare_combined_format_inputs.py

apps/decoder.c

+2 −3

Original line number	Diff line number	Diff line
		@@ -1858,13 +1858,13 @@ static ivas_error initOnFirstGoodFrame(
		return error;
		}

		int16_t zeroBuf = malloc( pcmFrameSize sizeof( int16_t ) );
		/* Write zeros to the output audio buffer */
		int16_t *zeroBuf = calloc( pcmFrameSize, sizeof( int16_t ) );
		if ( zeroBuf == NULL )
		{
		fprintf( stdout, "Error: Unable to allocate memory for output buffer.\n" );
		return IVAS_ERR_FAILED_ALLOC;
		}
		memset( zeroBuf, 0, pcmFrameSize * sizeof( int16_t ) );

		for ( int16_t i = 0; i < numInitialBadFrames; ++i )
		{
		@@ -1888,7 +1888,6 @@ static ivas_error initOnFirstGoodFrame(
		}
		else
		{

		if ( pRemainingDelayNumSamples < numOutSamples )
		{
		if ( ( error = AudioFileWriter_write( ppAfWriter, zeroBuf, numOutSamples * pNumOutChannels - ( pRemainingDelayNumSamples * *pNumOutChannels ) ) ) != IVAS_ERR_OK )

apps/renderer.c

+29 −0

Original line number	Diff line number	Diff line
		@@ -68,6 +68,10 @@
		#define IVAS_MAX16B_FLT 32767.0f
		#define IVAS_MIN16B_FLT ( -32768.0f )

		#ifdef NONBE_1359_FIX_IVASREND_OMASA_BINAURAL_LOUDNESS
		#define OMASA_TDREND_MATCHING_GAIN_DB ( -2.0f )
		#endif

		#if !defined( DEBUGGING ) && !defined( WMOPS )
		static
		#endif
		@@ -1153,20 +1157,39 @@ int main(
		}
		}

		#ifndef NONBE_1359_FIX_IVASREND_OMASA_BINAURAL_LOUDNESS
		/* Set the total number of objects */
		#endif
		if ( args.inConfig.numAudioObjects > 0 )
		{
		#ifdef NONBE_1359_FIX_IVASREND_OMASA_BINAURAL_LOUDNESS
		/* Set the total number of objects */
		#endif
		if ( ( error = IVAS_REND_SetTotalNumberOfObjects( hIvasRend, args.inConfig.numAudioObjects ) ) != IVAS_ERR_OK )
		{
		fprintf( stderr, "\nError in IVAS_REND_SetTotalNumberOfObjects(): %s\n", ivas_error_to_string( error ) );
		goto cleanup;
		}

		#ifdef NONBE_1359_FIX_IVASREND_OMASA_BINAURAL_LOUDNESS
		/* Set the metadata delay for objects */
		#endif
		if ( ( error = IVAS_REND_SetIsmMetadataDelay( hIvasRend, args.syncMdDelay ) ) != IVAS_ERR_OK )
		{
		fprintf( stderr, "\nError in IVAS_REND_SetIsmMetadataDelay(): %s\n", ivas_error_to_string( error ) );
		goto cleanup;
		}
		#ifdef NONBE_1359_FIX_IVASREND_OMASA_BINAURAL_LOUDNESS

		/* For OMASA input and BINAURAL output, apply a gain to objects to match the loudness with MASA part */
		if ( args.inConfig.numMasaBuses > 0 && args.outConfig.audioConfig == IVAS_AUDIO_CONFIG_BINAURAL )
		{
		for ( i = 0; i < args.inConfig.numAudioObjects; ++i )
		{
		args.inConfig.audioObjects[i].gain_dB += OMASA_TDREND_MATCHING_GAIN_DB;
		}
		}
		#endif
		}

		IVAS_REND_LfePanMtx lfePanMatrix;
		@@ -3491,14 +3514,20 @@ static void parseCombinedFormatInput(
		inConfig->numAmbisonicsBuses = 1;
		inConfig->ambisonicsBuses[0].audioConfig = audioConfig;
		inConfig->ambisonicsBuses[0].inputChannelIndex = inConfig->numAudioObjects;
		#ifdef NONBE_1352_HARMONIZE_OSBA_LOUDNESS
		inConfig->ambisonicsBuses[0].gain_dB = 0.f;
		#else
		inConfig->ambisonicsBuses[0].gain_dB = -6.f;
		#endif
		*configString += 4;

		#ifndef NONBE_1352_HARMONIZE_OSBA_LOUDNESS
		/* Modify input gain for objects too */
		for ( int16_t i = 0; i < inConfig->numAudioObjects; ++i )
		{
		inConfig->audioObjects[i].gain_dB = -6.f;
		}
		#endif
		}
		else if ( audioConfig == IVAS_AUDIO_CONFIG_MASA1 \|\| audioConfig == IVAS_AUDIO_CONFIG_MASA2 )
		{

ci/process_long_term_logs.py

0 → 100644

+180 −0

Original line number	Diff line number	Diff line
		import os
		import pandas as pd
		import argparse
		import plotly.express as px
		import plotly.graph_objects as go
		from plotly.subplots import make_subplots


		def read_csv_files(root_dir):
		"""Read csv files as dictionary of panda dataframes."""
		csv_data = {}
		for subdir, dirs, files in os.walk(root_dir):
		for file in files:
		if file.endswith(".csv"):
		file_path = os.path.join(subdir, file)
		try:
		df = pd.read_csv(file_path)
		csv_data[file_path] = df
		except Exception as e:
		print(f"Failed to read {file_path}: {e}")
		exit(-1)
		return csv_data


		def parse_csv_data(csv_data):
		"""keep 'testcase', 'format', 'MLD', 'MAX_ABS_DIFF' and add
		'date' column."""
		cols_to_keep = ["testcase", "format", "MLD", "MAX_ABS_DIFF"]
		parsed_data = {}
		for key, df in csv_data.items():
		cols = [col for col in cols_to_keep if col in df.columns]
		date = os.path.basename(os.path.dirname(key))
		new_df = df[cols].copy()
		new_df["date"] = date
		parsed_data[key] = new_df

		# concatenate all dataframe in the dictionary
		concat_df = pd.concat(parsed_data.values(), ignore_index=True)
		return concat_df


		def plot_data(df, output_filename):
		"""plot max values for 'MLD' and 'MAX_ABS_DIFF' data and save
		to html file."""
		# Convert 'date' to datetime
		df["date"] = pd.to_datetime(df["date"], errors="coerce")
		df["MLD"] = pd.to_numeric(df["MLD"], errors="coerce")
		df["MAX_ABS_DIFF"] = pd.to_numeric(df["MAX_ABS_DIFF"], errors="coerce")

		# Drop rows with NaT and NaN
		clean_df = df.dropna(subset=["date", "MLD", "MAX_ABS_DIFF"])

		# Group by 'format' and 'date' to get rows with max 'MLD' per group
		max_mld = (
		clean_df.groupby(["format", "date"])
		.apply(lambda x: x.loc[x["MLD"].idxmax()])
		.reset_index(drop=True)
		)

		# Group by 'format' and 'date' to get rows with max 'MAX_ABS_DIFF' per
		# group
		max_abs_diff = (
		clean_df.groupby(["format", "date"])
		.apply(lambda x: x.loc[x["MAX_ABS_DIFF"].idxmax()])
		.reset_index(drop=True)
		)

		formats = sorted(clean_df["format"].unique())

		fig = make_subplots(
		rows=5,
		cols=2,
		specs=[[{"secondary_y": True}] * 2] * 5,
		subplot_titles=[f"{i}" for i in formats],
		shared_xaxes="columns",
		)

		for i, fmt in enumerate(formats):
		row = i // 2 + 1
		col = i % 2 + 1

		data_mld = max_mld[max_mld["format"] == fmt].sort_values("date")
		data_diff = max_abs_diff[max_abs_diff["format"]
		== fmt].sort_values("date")

		# Add max 'MLD' to primary y-axis
		fig.add_trace(
		go.Scatter(
		x=data_mld["date"],
		y=data_mld["MLD"],
		mode="lines+markers",
		name=f" {fmt} - Max MLD",
		hovertext=[
		f"Testcase: {tc}<br>MLD: {mld:.4f}<br>MAX_ABS_DIFF:"
		f"{abs_diff}<br>Format:"
		f" {format}<br>Date: {date.date()}"
		for tc, mld, abs_diff, format, date in zip(
		data_mld["testcase"],
		data_mld["MLD"],
		data_mld["MAX_ABS_DIFF"],
		data_mld["format"],
		data_mld["date"],
		)
		],
		hoverinfo="text",
		),
		row=row,
		col=col,
		secondary_y=False,
		)

		# Add max 'MAX_ABS_DIFF' to secondary y-axis
		fig.add_trace(
		go.Scatter(
		x=data_diff["date"],
		y=data_diff["MAX_ABS_DIFF"],
		mode="lines+markers",
		name=f"{fmt} - Max MAX_ABS_DIFF",
		hovertext=[
		f"Testcase: {tc}<br>MLD: {mld:.4f}<br>MAX_ABS_DIFF:"
		f" {abs_diff:.4f}<br>Format:"
		f" {format}<br>Date: {date.date()}"
		for tc, mld, abs_diff, format, date in zip(
		data_diff["testcase"],
		data_diff["MLD"],
		data_diff["MAX_ABS_DIFF"],
		data_diff["format"],
		data_diff["date"],
		)
		],
		hoverinfo="text",
		),
		row=row,
		col=col,
		secondary_y=True,
		)

		fig.update_layout(
		title_text="Long-term regression: max MLD and max MAX_ABS_DIFF",
		legend=dict(x=1, y=1, orientation="v"),
		hovermode="x unified",
		)

		fig.update_xaxes(automargin=True)
		fig.update_yaxes(automargin=True)

		# Update y-axes titles per subplot
		for i in range(10):
		yaxis_num = i * 2 + 1
		yaxis2_num = yaxis_num + 1
		fig["layout"][f"yaxis{yaxis_num}"].update(
		title="Max MLD", titlefont=dict(color="blue"), tickfont=dict(color="blue")
		)
		fig["layout"][f"yaxis{yaxis2_num}"].update(
		title="Max MAX_ABS_DIFF",
		titlefont=dict(color="green"),
		tickfont=dict(color="green"),
		)

		# Save to html
		fig.write_html(output_filename)


		if __name__ == "__main__":
		parser = argparse.ArgumentParser(description="Plot long term logs")
		parser.add_argument(
		"root_dir",
		type=str,
		help="Root directory containing subdirectories" " with CSV log files",
		)
		parser.add_argument(
		"output_filename",
		type=str,
		help="Filename of the generated plot. e.g" ". long_term_regression.html",
		)
		args = parser.parse_args()

		csv_data = read_csv_files(args.root_dir)
		data = parse_csv_data(csv_data)
		plot_data(data, args.output_filename)

lib_com/options.h

+4 −0

Original line number	Diff line number	Diff line
		@@ -162,6 +162,7 @@

		/#define FIX_I4_OL_PITCH/ /* fix open-loop pitch used for EVS core switching */
		#define TMP_FIX_1119_SPLIT_RENDERING_VOIP /* FhG: Add error check for unsupported config: split rendering with VoIP mode */
		/#define CODE_IMPROVEMENTS/ /* FhG: Small code improvements that do not change the functionality */

		/* #################### End BE switches ################################## */

		@@ -174,6 +175,9 @@
		#define NONBE_1122_KEEP_EVS_MODE_UNCHANGED /* FhG: Disables fix for issue 1122 in EVS mode to keep BE tests green. This switch should be removed once the 1122 fix is added to EVS via a CR. */
		#define NONBE_FIX_TCX5_INTERLEAVING_FOR_FS_IN_UNEQUAL_FS_OUT /* FhG: apply correct TCX5 grouping/interleaving when input_fs != output_fs */
		#define NONBE_1339_FIXOSBA_EXT_LOUDNESS /* FhG: issue 1339: apply scaling with EXT output in OSBA high-BR mode */
		#define NONBE_1352_HARMONIZE_OSBA_LOUDNESS /* FhG: do not scale OSBA inputs by 0.5 any more */
		#define NONBE_1359_FIX_IVASREND_OMASA_BINAURAL_LOUDNESS /* Nokia: issue 1339: Apply scaling to the object-part of OMASA for binaural rendering in IVAS_rend. */
		#define NONBE_1362_FIX_OMASA_TO_MASA1_RENDERING /* Nokia: Fix OMASA to MASA1 rendering in IVAS_rend */

		/* ##################### End NON-BE switches ########################### */