Merge branch '967-ism-mono-downmix-accumulating-energies' into 'main' (15499265) · Commits · IVAS Codec Public Collaboration / IVAS Codec

lib_com/ivas_prot.h

+4 −0

Original line number	Diff line number	Diff line
		@@ -5469,7 +5469,11 @@ void ivas_mono_dmx_renderer_close(
		MONO_DOWNMIX_RENDERER_HANDLE hMonoDmxRenderer / i/ i/o: Mono downmix structure */
		);

		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		void ivas_ism_mono_dmx(
		#else
		void ivas_mono_downmix_render_passive(
		#endif
		Decoder_Struct st_ivas, / i/o: IVAS decoder structure */
		float output_f[], / i/o: synthesized core-coder transport channels/mono output */
		const int16_t output_frame /* i : output frame length */

lib_com/options.h

+1 −0

Original line number	Diff line number	Diff line
		@@ -175,6 +175,7 @@
		#define NONBE_FIX_935_EARLY_REFLECTIONS_WRONG_ORDER /* Qualcomm: issue 953: fix order or ER channels in LC mode*/
		#define NONBE_FIX_949_MC_5MS_FRAMING /* Dlb: issue 949: fix for issue 949, distorted output in MC mode with 5ms framing*/
		#define NONBE_FIX_952_MC_PARAMUPMIX_5MS /* Dlb : issue 952 : Differences between 5ms and 20ms rendering for ParamUpmix*/
		#define NONBE_FIX_967_ISM_MONO_DMX /* FhG: issue 967: accumulating energies in ISM mono DMX */
		#define NONBE_FIX_968_ISM_BRIR_WITH_HEADROTATION_5MS_FIX /* FhG : issue #968: differences between 5ms and 20ms rendering for discrete ISM with BRIR and head rotation*/
		#define NONBE_FIX_977_OSBA_GAIN_MISMATCH /* Dlb : issue 977 : Output gain mismatch for different bit rates in OSBA */

lib_dec/ivas_jbm_dec.c

+12 −0

Original line number	Diff line number	Diff line
		@@ -196,7 +196,11 @@ ivas_error ivas_jbm_dec_tc(

		if ( st_ivas->renderer_type == RENDERER_MONO_DOWNMIX )
		{
		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		ivas_ism_mono_dmx( st_ivas, p_output, output_frame );
		#else
		ivas_mono_downmix_render_passive( st_ivas, p_output, output_frame );
		#endif
		}
		else if ( st_ivas->ism_mode == ISM_MODE_PARAM && ( st_ivas->renderer_type == RENDERER_BINAURAL_PARAMETRIC \|\| st_ivas->renderer_type == RENDERER_BINAURAL_PARAMETRIC_ROOM \|\| st_ivas->renderer_type == RENDERER_STEREO_PARAMETRIC ) )
		{
		@@ -388,7 +392,11 @@ ivas_error ivas_jbm_dec_tc(

		if ( st_ivas->renderer_type == RENDERER_MONO_DOWNMIX )
		{
		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		ivas_ism_mono_dmx( st_ivas, p_output, output_frame );
		#else
		ivas_mono_downmix_render_passive( st_ivas, p_output, output_frame );
		#endif
		}
		else if ( output_config == IVAS_AUDIO_CONFIG_EXTERNAL )
		{
		@@ -512,7 +520,11 @@ ivas_error ivas_jbm_dec_tc(
		}
		else if ( st_ivas->renderer_type == RENDERER_MONO_DOWNMIX && st_ivas->ism_mode == ISM_SBA_MODE_DISC )
		{
		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		ivas_ism_mono_dmx( st_ivas, p_output, output_frame );
		#else
		ivas_mono_downmix_render_passive( st_ivas, p_output, output_frame );
		#endif

		/* add W */
		for ( n = 0; n < nchan_out; n++ )

lib_dec/ivas_mono_dmx_renderer.c

+36 −4

Original line number	Diff line number	Diff line
		@@ -48,7 +48,9 @@

		#define DOWNMIX_ALPHA 0.95f /* Smoothing coefficient */
		#define DOWNMIX_MAX_GAIN 4.0f /* Maximum allowed gain */
		#define MONO_DOWNMIX_RENDERER_MAX_INPUT_CHANS 4
		#ifndef NONBE_FIX_967_ISM_MONO_DMX
		#define MONO_DOWNMIX_RENDERER_MAX_INPUT_CHANS 4 // FhG esr: this seems to be obsolete; not related to issue 967
		#endif


		/*-------------------------------------------------------------------------
		@@ -99,12 +101,16 @@ void ivas_mono_dmx_renderer_close(


		/*-------------------------------------------------------------------------
		* ivas_mono_downmix_render_passive()
		* ivas_ism_mono_dmx()
		*
		* Downmix process
		------------------------------------------------------------------------/

		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		void ivas_ism_mono_dmx(
		#else
		void ivas_mono_downmix_render_passive(
		#endif
		Decoder_Struct st_ivas, / i/o: IVAS decoder structure */
		float output_f[], / i/o: synthesized core-coder transport channels/mono output */
		const int16_t output_frame /* i : output frame length */
		@@ -112,6 +118,9 @@ void ivas_mono_downmix_render_passive(
		{
		int16_t i, j, numInputChannels;
		float proto_signal[L_FRAME48k], eq;
		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		float inputEnergy_local, protoEnergy_local;
		#endif
		MONO_DOWNMIX_RENDERER_HANDLE hDownmix;

		numInputChannels = st_ivas->nSCE;
		@@ -134,6 +143,10 @@ void ivas_mono_downmix_render_passive(

		hDownmix = st_ivas->hMonoDmxRenderer;
		set_zero( proto_signal, output_frame );
		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		inputEnergy_local = 0;
		protoEnergy_local = 0;
		#endif

		/* Compute the Proto Signal */
		for ( i = 0; i < numInputChannels; i++ )
		@@ -141,6 +154,18 @@ void ivas_mono_downmix_render_passive(
		v_add( output_f[i], proto_signal, proto_signal, output_frame );
		}

		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		/* compute the input energy, proto energy */
		for ( i = 0; i < output_frame; i++ )
		{
		protoEnergy_local += proto_signal[i] * proto_signal[i];

		for ( j = 0; j < numInputChannels; j++ )
		{
		inputEnergy_local += ( output_f[j][i] * output_f[j][i] );
		}
		}
		#else
		/* compute the input energy, proto energy after smoothing */
		hDownmix->inputEnergy *= DOWNMIX_ALPHA;
		hDownmix->protoEnergy *= DOWNMIX_ALPHA;
		@@ -153,6 +178,13 @@ void ivas_mono_downmix_render_passive(
		hDownmix->inputEnergy += ( output_f[j][i] * output_f[j][i] );
		}
		}
		#endif
		#ifdef NONBE_FIX_967_ISM_MONO_DMX
		hDownmix->inputEnergy *= DOWNMIX_ALPHA;
		hDownmix->protoEnergy *= DOWNMIX_ALPHA;
		hDownmix->inputEnergy += ( 1.0f - DOWNMIX_ALPHA ) * inputEnergy_local;
		hDownmix->protoEnergy += ( 1.0f - DOWNMIX_ALPHA ) * protoEnergy_local;
		#endif

		/* compute the eq factor */
		eq = min( DOWNMIX_MAX_GAIN, sqrtf( hDownmix->inputEnergy / ( EPSILON + hDownmix->protoEnergy ) ) );

scripts/config/self_test.prm

+1 −1

Original line number	Diff line number	Diff line
		@@ -402,7 +402,7 @@ eid-xor -fer -vbr -bs g192 -ep g192 bit ../scripts/dly_error_profiles/ep_5pct.g1

		// 3 ISM with metadata at 48 kbps, 48 kHz in, 48 kHz out, EXT out
		../IVAS_cod -ism 3 testv/stvISM1.csv testv/stvISM2.csv testv/stvISM3.csv 48000 48 testv/stv3ISM48s.wav bit
		../IVAS_dec MONO 48 bit testv/stv3ISM48s.48000_48-48_MONO.tst
		../IVAS_dec EXT 48 bit testv/stv3ISM48s.48000_48-48_MONO.tst

		// 3 ISM with metadata at 64 kbps, 48 kHz in, 48 kHz out, EXT out
		../IVAS_cod -ism 3 testv/stvISM1.csv testv/stvISM2.csv testv/stvISM3.csv 64000 48 testv/stv3ISM48s.wav bit