implement delay harmonization and complexity reduction for SBA-to-stereo processing in the decoder (b3a78e75) · Commits · IVAS Codec Public Collaboration / IVAS Codec

lib_com/delay_comp.c

+11 −1

Original line number	Diff line number	Diff line
		@@ -57,6 +57,10 @@ int32_t get_delay(
		HANDLE_CLDFB_FILTER_BANK hCldfb, /* i : Handle of Cldfb analysis */
		RENDERER_TYPE renderer_type, /* i : IVAS rendering type */
		const int32_t binaural_latency_ns /* i : binaural renderer HRTF delay in ns */
		#ifdef DFT_STEREO_SPAR_MIXING
		,
		const int16_t sba_dirac_stereo_flag
		#endif
		)
		{
		int32_t delay = 0;
		@@ -99,6 +103,12 @@ int32_t get_delay(
		{
		delay += IVAS_FB_DEC_DELAY_NS;
		}
		#ifdef DFT_STEREO_SPAR_MIXING
		else if ( sba_dirac_stereo_flag )
		{
		delay += 0;
		}
		#endif

		/* compensate for Binaural renderer HRTF delay */
		{

lib_com/ivas_cnst.h

+6 −0

Original line number	Diff line number	Diff line
		@@ -574,7 +574,13 @@ typedef enum
		#define NO_SYMB_GR_PRED_G 8

		#define STEREO_DFT_RES_BW_MAX 66 /Maximum number of bin for residual signal in each frame (res_cod_band_max == 6 in 48kHz)/

		#ifdef DFT_STEREO_SPAR_MIXING
		#define SBA_DIRAC_STEREO_NUM_BANDS 12
		#else
		#define SBA_DIRAC_STEREO_NUM_BANDS 5
		#endif

		#define SBA_DIRAC_NRG_SMOOTH_LONG 10
		#define SBA_DIRAC_NRG_SMOOTH_SHORT 3

lib_com/ivas_prot.h

+26 −1

Original line number	Diff line number	Diff line
		@@ -1016,6 +1016,10 @@ ivas_error stereo_dft_dec_create(
		const int32_t element_brate, /* i : element bitrate */
		const int32_t output_Fs, /* i : output sampling rate */
		const int16_t sba_dirac_stereo_flag /* i : signal stereo output for SBA DirAC */
		#ifdef DFT_STEREO_SPAR_MIXING
		,
		const int16_t nchan_transport
		#endif
		);

		void stereo_dft_dec_reset(
		@@ -1055,10 +1059,21 @@ void stereo_dft_dec_synthesize(
		void stereo_dft_dec(
		STEREO_DFT_DEC_DATA_HANDLE hStereoDft, /* i/o: decoder DFT stereo handle */
		Decoder_State st0, / i/o: decoder state structure */
		#ifdef DFT_STEREO_SPAR_MIXING
		float DFT[CPE_CHANNELS + 1][STEREO_DFT_BUF_MAX], /* i/o: DFT buffers */
		#else
		float DFT[CPE_CHANNELS][STEREO_DFT_BUF_MAX], /* i/o: DFT buffers */
		#endif
		float input_mem, / i/o: mem of buffer DFT analysis */
		STEREO_CNG_DEC_HANDLE hStereoCng, /* i/o: Stereo CNG data structure */
		const int16_t sba_dirac_stereo_flag /* i : signal stereo output for SBA DirAC */
		#ifdef DFT_STEREO_SPAR_MIXING
		,
		ivas_spar_md_dec_state_t hMdDec, / SPAR MD handle for upmixing */
		int16_t cross_fade_start_offset, /* i: SPAR mixer delay compensation */
		int32_t output_Fs, /* i: Fs for delay calculation */
		int16_t nchan_transport /* i: number of transpor channels */
		#endif
		);

		void stereo_dft_res_ecu(
		@@ -3174,6 +3189,10 @@ void ivas_sba_dirac_stereo_dec(
		Decoder_Struct st_ivas, / i/o: IVAS decoder structure */
		float output[CPE_CHANNELS][L_FRAME48k], /* o : output synthesis signal */
		const int16_t output_frame /* i : output frame length per channel */
		#ifdef DFT_STEREO_SPAR_MIXING
		,
		const int16_t mcmasa
		#endif
		);

		void ivas_sba_dirac_stereo_config(
		@@ -3182,6 +3201,12 @@ void ivas_sba_dirac_stereo_config(

		void ivas_sba_dirac_stereo_smooth_parameters(
		STEREO_DFT_DEC_DATA_HANDLE hStereoDft /* i/o: encoder DFT stereo handle */
		#ifdef DFT_STEREO_SPAR_MIXING
		,
		ivas_spar_md_dec_state_t hMdDec, / i/o: SPAR MD handle for upmixing */
		int16_t cross_fade_start_offset, /* i: SPAR mixer delay compensation */
		int32_t output_Fs /* i: Fs for delay calculation */
		#endif
		);

		ivas_error ivas_sba_get_hoa_dec_matrix(

lib_com/options.h

+6 −0

Original line number	Diff line number	Diff line
		@@ -144,6 +144,7 @@
		#define DISABLE_ADAP_RES_COD_TMP /* temporary fix for IVAS-403, disables adaptive residual coding */
		/#define ITD_WINNER_GAIN_MODIFY / /* ITD optimization - WORK IN PROGRESS */
		/#define FIX_I4_OL_PITCH/ /* fix open-loop pitch used for EVS core switching */

		#define FIX_I1_113 /* under review : MCT bit distribution optimization for SBA high bitrates*/
		#define FIX_124_DONT_ALLOC_PLCINFO_IN_IVAS /* Issue 124: do not allocate unused plc struct in IVAS modes which is only used in EVS mono */
		#define SBA_BR_SWITCHING_2 /* Issue 114: Changes for sba bit rate switching with reconfigurations*/
		@@ -163,7 +164,12 @@
		#define IMPROVE_CMDLINE_ROBUSTNESS /* Issue 233: Improve robustness of command-line parameters */
		#define FIX_MDCT_AND_MC_MONO_ISSUES /* Issue 242: Fix some issues with TCX-LTP and delay alignement for mono output */

		#define DFT_STEREO_SPAR_MIXING /* For SBA to stereo output, perform SPAR upmix in DFT domain */

		#ifdef DFT_STEREO_SPAR_MIXING
		/#define DFT_STEREO_SPAR_MIXING_DEBUG/ /* more debugging output for DFT_STEREO_SPAR_MIXING_DEBUG */
		#define DISABLE_RES_CHANNELS_MCT /* decode only W and residual for Y when outputting to stereo */
		#endif
		/* ################## End DEVELOPMENT switches ######################### */
		/* clang-format on */
		#endif

lib_com/prot.h

+4 −0

Original line number	Diff line number	Diff line
		@@ -719,6 +719,10 @@ int32_t get_delay(
		HANDLE_CLDFB_FILTER_BANK hCldfb, /* i : Handle of Cldfb analysis */
		RENDERER_TYPE renderer_type, /* i : IVAS rendering type */
		const int32_t binaural_latency_ns /* i : binaural renderer HRTF delay in ns */
		#ifdef DFT_STEREO_SPAR_MIXING
		,
		const int16_t sba_dirac_stereo_flag
		#endif
		);

		void decision_matrix_enc(