Added TD renderer ITD update and cleanup under define FIX_ITD (7f79519e) · Commits · IVAS Codec Public Collaboration / IVAS Codec

lib_com/ivas_cnst.h

+9 −1

Original line number	Diff line number	Diff line
		@@ -1390,8 +1390,16 @@ typedef enum
		#define SFX_SPAT_BIN_MAX_FILTER_LENGTH 256

		#define SPAT_BIN_MAX_INPUT_CHANNELS 1 /* Max number of input channels per source/object. Mono for now, but stereo objects may be considered. */

		#ifdef FIX_ITD
		#define MAX_ITD 50
		#define SFX_SPAT_BIN_SINC_M 5
		#define SFX_SPAT_BIN_NUM_SUBSAMPLES 64
		#define ITD_MEM_LEN (MAX_ITD + SFX_SPAT_BIN_SINC_M)
		#define L_SUBFRAME5MS_48k (L_FRAME48k/4)
		#define BINAURAL_TD_LATENCY_S 0.0f /* ITD fix removes TD renderer delay -- should be cleaned out */
		#else
		#define BINAURAL_TD_LATENCY_S 0.00675f /* Binaural TD renderer latency in second == 324 samples in between 333 and 315 */
		#endif

		/* ----- Enums - TD Renderer ----- */

lib_com/ivas_prot.h

+46 −1

Original line number	Diff line number	Diff line
		@@ -4992,7 +4992,13 @@ void GetFilterFromAngle(
		TDREND_HRFILT_FiltSet_t HrFiltSet_p, / i/o: HR filter set structure */
		const float Elev, /* i : Elevation, degrees */
		float Azim, /* i : Azimuth, degrees */
		#ifdef FIX_ITD
		float LeftFilter_p, / o : Left HR filter */
		float RightFilter_p, / o : Right HR filter */
		int16_t itd / o : ITD value */
		#else
		SFX_SpatBin_Params_t SfxSpatBinParams_p / i/o: Currently used HR filter */
		#endif
		);

		void HRTF_model_precalc(
		@@ -5021,8 +5027,12 @@ ivas_error TDREND_REND_RenderSourceHRFilt(
		BINAURAL_TD_OBJECT_RENDERER_HANDLE hBinRendererTd, /* i/o: TD renderer handle */
		#endif
		float output_buf[][L_SPATIAL_SUBFR_48k], /* o : Output buffer */
		#ifdef FIX_ITD
		const int16_t subframe_length /* i : Subframe length in use */
		#else
		const int16_t subframe_length, /* i : Subframe length in use */
		const int32_t output_Fs /* i : Output sample rate */
		#endif
		);

		/* ----- Object renderer - sources ----- */
		@@ -5055,7 +5065,15 @@ void TDREND_SRC_REND_UpdateFiltersFromSpatialParams(
		BINAURAL_TD_OBJECT_RENDERER_HANDLE hBinRendererTd, /* i/o: TD renderer handle */
		TDREND_SRC_REND_t SrcRend_p, / i/o: Source object */
		TDREND_SRC_SPATIAL_t SrcSpatial_p, / i : Spatial aspects of source */
		#ifdef FIX_ITD
		float LeftFilter_p, / o: Left filter */
		float RightFilter_p, / o: Right filter */
		int16_t filterlength, / o: Length of filters */
		int16_t itd, / o: ITD value */
		float Gain / o: Gain value */
		#else
		const int32_t output_Fs /* i : Output sample rate */
		#endif
		);

		ivas_error TDREND_SRC_Alloc(
		@@ -5068,8 +5086,12 @@ void TDREND_SRC_Dealloc(

		void TDREND_SRC_Init(
		TDREND_SRC_t Src_p, / i/o: Source to initialize */
		#ifdef FIX_ITD
		const TDREND_PosType_t PosType /* i : Position type specifier */
		#else
		const TDREND_PosType_t PosType, /* i : Position type specifier */
		const int32_t output_Fs /* i : Output sampling rate */
		#endif
		);

		/* ----- Object renderer - vec ----- */
		@@ -5114,8 +5136,12 @@ int16_t TDREND_SPATIAL_EvalOrthonormOrient(
		ivas_error TDREND_MIX_AddSrc(
		BINAURAL_TD_OBJECT_RENDERER_HANDLE hBinRendererTd, /* i/o: TD renderer handle */
		int16_t SrcInd, / o : Source index */
		#ifdef FIX_ITD
		const TDREND_PosType_t PosType /* i : Position type (absolute/relative) */
		#else
		const TDREND_PosType_t PosType, /* i : Position type (absolute/relative) */
		const int32_t output_Fs /* i : Output sampling rate */
		#endif
		);

		ivas_error TDREND_MIX_SetDistAttenModel(
		@@ -5146,7 +5172,7 @@ ivas_error TDREND_MIX_Init(
		);

		/* ----- Object renderer - sfx ----- */

		#ifndef FIX_ITD
		ivas_error TDREND_SFX_SpatBin_Initialize(
		SFX_SpatBin_t SfxSpatBin_p, / i/o: Spatial parameters handle */
		const int32_t output_Fs /* i : Output sampling rate */
		@@ -5168,8 +5194,27 @@ void TDREND_SFX_SpatBin_Execute_Main(
		int16_t NoOfDeliveredOutputSamples_p, / o : Number of output samples actually rendered */
		const int32_t output_Fs /* i : Output sample rate */
		);
		#endif

		#ifdef FIX_ITD
		void TDREND_Apply_ITD(
		float input, / i: Input SCE subframe to be time adjusted */
		float out_left, / o: Output left channels with ITD applied */
		float out_right, / o: Output right channels with ITD applied */
		int16_t previtd, /i/o: Previous ITD value */
		const int16_t itd, /* i: Current subframe ITD value */
		float mem_itd, /i/o: ITD buffer memory */
		const int16_t length /* i: Subframe length */
		);

		void TDREND_firfilt(
		float signal, / i/o: Input signal / Filtered signal */
		const float filter, / i/o: FIR filter */
		float mem, / i/o: filter memory */
		const int16_t subframe_length, /* i : Length of signal */
		const int16_t filterlength /* i : Filter length */
		);
		#endif
		/----------------------------------------------------------------------------------
		* Filter-bank (FB) Mixer
		----------------------------------------------------------------------------------/

lib_com/options.h

+1 −1

Original line number	Diff line number	Diff line
		@@ -151,7 +151,7 @@
		#define FIX_MCT_PLC_RECOVERY /* Issue 184: scale the old synthesis part correctly in the first good frame after lost frames in MCT modes - to be activated after previous switch is merged */
		#define SBA_BR_SWITCHING /* Issue 114: Changes for sba bit rate switching*/
		#define FIX_AGC_WINFUNC_MEMORY /* Issue 62: lower agc_com.winFunc memory consumption */

		#define FIX_ITD

		/* ################## End DEVELOPMENT switches ######################### */
		/* clang-format on */

lib_dec/ivas_stat_dec.h

+16 −3

Original line number	Diff line number	Diff line
		@@ -1324,6 +1324,8 @@ typedef struct ivas_binaural_head_track_struct
		/----------------------------------------------------------------------------------
		* TD ISm Object Renderer structure
		----------------------------------------------------------------------------------/

		#ifndef FIX_ITD
		// VE2AT: move to ivas_rom_rend.h ?
		typedef struct
		{
		@@ -1387,6 +1389,7 @@ typedef struct TDREND_LIST_Item_s
		struct TDREND_LIST_Item_s Next_p; / Pointer to the next item */

		} TDREND_LIST_Item_t;
		#endif

		typedef struct
		{
		@@ -1579,10 +1582,10 @@ typedef struct TDREND_SRC_REND_s
		float SrcGainMax_p[SPAT_BIN_MAX_INPUT_CHANNELS];
		float DirGain_p[SPAT_BIN_MAX_INPUT_CHANNELS];
		float DistGain_p[SPAT_BIN_MAX_INPUT_CHANNELS];

		#ifndef FIX_ITD
		/* HR filtering parameters */
		SFX_SpatBin_t *SfxSpatBin_p;

		#endif
		} TDREND_SRC_REND_t;


		@@ -1605,7 +1608,17 @@ typedef struct
		float InputFrame_p; / Input frame pointer */
		TDREND_SRC_SPATIAL_t *SrcSpatial_p;
		TDREND_SRC_REND_t *SrcRend_p;

		#ifdef FIX_ITD
		int16_t itd;
		int16_t previtd;
		int16_t filterlength;
		float mem_itd[ITD_MEM_LEN];
		float hr_filt_left[SFX_SPAT_BIN_MAX_FILTER_LENGTH]; /* Todo: Should we allocate these buffers with count_malloc instead of the maximum length? */
		float hr_filt_right[SFX_SPAT_BIN_MAX_FILTER_LENGTH];
		float mem_hr_filt_left[SFX_SPAT_BIN_MAX_FILTER_LENGTH - 1];
		float mem_hr_filt_right[SFX_SPAT_BIN_MAX_FILTER_LENGTH - 1];
		float Gain;
		#endif
		} TDREND_SRC_t;

		/* Top level TD binaural renderer handle */

lib_rend/ivas_lib_rend_internal.h

+2 −0

Original line number	Diff line number	Diff line
		@@ -109,7 +109,9 @@ ivas_error ivas_rend_TDObjRenderFrame(
		const IVAS_REND_HeadRotData headRotData, / i : Input head positions */
		const IVAS_REND_AudioObjectPosition currentPos, / i : Object position */
		const int16_t output_frame, /* i : output frame length */
		#ifndef FIX_ITD
		const int32_t output_Fs, /* i : output sampling rate */
		#endif
		float output[][L_FRAME48k] /* i/o: SCE channels / Binaural synthesis */
		);