Continue fixing 0 DOF (WIP) (3c5bad06) · Commits · IVAS Codec Public Collaboration / IVAS Codec

lib_com/common_api_types.h

+6 −0

Original line number	Diff line number	Diff line
		@@ -132,6 +132,9 @@ typedef struct ivas_split_rend_bits_t
		int32_t buf_len; /size of bits_buf in bytes. This field should be set by allocator of bits_buf/
		int32_t bits_written;
		int32_t bits_read;
		#ifdef API_5MS
		int16_t codec_frame_size_ms; /* TODO complete implementation */
		#endif
		IVAS_SPLIT_REND_CODEC codec;
		IVAS_SPLIT_REND_POSE_CORRECTION_MODE pose_correction;
		} ivas_split_rend_bits_t, IVAS_SPLIT_REND_BITS, *IVAS_SPLIT_REND_BITS_HANDLE;
		@@ -202,6 +205,9 @@ typedef struct _IVAS_SPLIT_REND_CONFIG
		3 - (3dof correction. By default YAW, PITCH and ROLL correction)
		*/
		int16_t codec_delay_ms; /PLACEHOLDER (currently being ignored) : look ahead delay of the codec that is used to code BIN signal output of pre-renderer/
		#ifdef API_5MS
		int16_t codec_frame_size_ms; /Codec frame size in milliseconds, only relevant with LC3plus /
		#endif
		IVAS_SPLIT_REND_POSE_CORRECTION_MODE poseCorrectionMode;
		IVAS_SPLIT_REND_CODEC codec;
		IVAS_SPLIT_REND_RENDERER_SELECTION rendererSelection;

lib_com/options.h

+1 −0

Original line number	Diff line number	Diff line
		@@ -164,6 +164,7 @@
		#define FIX_XXX_JBM_USAN /* FhG: fix usan in acessing transport channel buffers */
		#define FIX_XXX_PARAMISM_JBM_ENER_CORRECTION /* FhG: fix energy correction in ParamISM rendering */
		#define NONBE_FIX_589_JBM_TC_OFFSETS
		#define FIX_JBM_MC2SBA /* FhG: fix check for transport vs. internal channel count in JBM prior to ivas_mc2sba() */
		#define API_5MS /* FhG: 5ms rendering capability */
		#ifdef API_5MS
		#define API_5MS_BASELINE /* FhG: baseline with 20ms rendering and split rendering through 20ms branch */

lib_dec/ivas_jbm_dec.c

+35 −4

Original line number	Diff line number	Diff line
		@@ -370,7 +370,14 @@ ivas_error ivas_jbm_dec_tc(

		if ( st_ivas->transport_config != st_ivas->intern_config && ( st_ivas->intern_config == AUDIO_CONFIG_FOA \|\| st_ivas->intern_config == AUDIO_CONFIG_HOA2 \|\| st_ivas->intern_config == AUDIO_CONFIG_HOA3 ) )
		{
		if ( ( st_ivas->hTransSetup.nchan_out_woLFE + st_ivas->hTransSetup.num_lfe ) <= ( st_ivas->hIntSetup.nchan_out_woLFE + st_ivas->hIntSetup.num_lfe ) )
		if ( ( st_ivas->hTransSetup.nchan_out_woLFE + st_ivas->hTransSetup.num_lfe )
		#ifdef FIX_JBM_MC2SBA
		>=
		#else
		<=
		#endif
		( st_ivas->hIntSetup.nchan_out_woLFE + st_ivas->hIntSetup.num_lfe ) )

		{
		ivas_mc2sba( st_ivas->hTransSetup, p_output, p_output, output_frame, st_ivas->hIntSetup.ambisonics_order, GAIN_LFE );
		}
		@@ -936,10 +943,21 @@ ivas_error ivas_jbm_dec_render(
		{
		if ( st_ivas->mc_mode == MC_MODE_MCT )
		{
		#ifdef FIX_JBM_MC2SBA
		int16_t crendInPlaceRotation = FALSE;
		#endif
		*nSamplesRendered = min( st_ivas->hTcBuffer->n_samples_available, nSamplesAskedLocal );
		if ( st_ivas->transport_config != st_ivas->intern_config && ( st_ivas->intern_config == AUDIO_CONFIG_FOA \|\| st_ivas->intern_config == AUDIO_CONFIG_HOA2 \|\| st_ivas->intern_config == AUDIO_CONFIG_HOA3 ) )
		{
		#ifdef FIX_JBM_MC2SBA
		if ( ( st_ivas->hTransSetup.nchan_out_woLFE + st_ivas->hTransSetup.num_lfe ) < ( st_ivas->hIntSetup.nchan_out_woLFE + st_ivas->hIntSetup.num_lfe ) )
		{
		crendInPlaceRotation = TRUE;
		#endif
		ivas_mc2sba( st_ivas->hTransSetup, p_tc, p_output, *nSamplesRendered, st_ivas->hIntSetup.ambisonics_order, GAIN_LFE );
		#ifdef FIX_JBM_MC2SBA
		}
		#endif
		}

		/* Rendering */
		@@ -947,7 +965,14 @@ ivas_error ivas_jbm_dec_render(
		{
		if ( ( error = ivas_rend_crendProcessSubframe( st_ivas->hCrendWrapper, st_ivas->intern_config, st_ivas->hOutSetup.output_config, st_ivas->hDecoderConfig,
		st_ivas->hCombinedOrientationData,
		&st_ivas->hIntSetup, st_ivas->hEFAPdata, st_ivas->hTcBuffer, p_tc, p_output, *nSamplesRendered, output_Fs ) ) != IVAS_ERR_OK )
		&st_ivas->hIntSetup, st_ivas->hEFAPdata, st_ivas->hTcBuffer,
		#ifdef FIX_JBM_MC2SBA
		crendInPlaceRotation ? p_output : p_tc,
		#else
		p_tc,
		#endif
		p_output,
		*nSamplesRendered, output_Fs ) ) != IVAS_ERR_OK )
		{
		return error;
		}
		@@ -1580,7 +1605,13 @@ int16_t ivas_jbm_dec_get_num_tc_channels(
		/* do all static dmx already in the TC decoder if less channels than transported... */
		if ( st_ivas->transport_config != st_ivas->intern_config && ( st_ivas->intern_config == AUDIO_CONFIG_FOA \|\| st_ivas->intern_config == AUDIO_CONFIG_HOA2 \|\| st_ivas->intern_config == AUDIO_CONFIG_HOA3 ) )
		{
		if ( ( st_ivas->hTransSetup.nchan_out_woLFE + st_ivas->hTransSetup.num_lfe ) > ( st_ivas->hIntSetup.nchan_out_woLFE + st_ivas->hIntSetup.num_lfe ) )
		if ( ( st_ivas->hTransSetup.nchan_out_woLFE + st_ivas->hTransSetup.num_lfe )
		#ifdef FIX_JBM_MC2SBA
		>=
		#else
		>
		#endif
		( st_ivas->hIntSetup.nchan_out_woLFE + st_ivas->hIntSetup.num_lfe ) )
		{
		num_tc = st_ivas->hIntSetup.nchan_out_woLFE + st_ivas->hIntSetup.num_lfe;
		}

lib_dec/lib_dec.c

+18 −8

Original line number	Diff line number	Diff line
		@@ -1128,6 +1128,7 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(
		int32_t output_Fs;
		float output[BINAURAL_CHANNELS * MAX_HEAD_ROT_POSES][L_FRAME48k];
		int16_t output_int[BINAURAL_CHANNELS * MAX_HEAD_ROT_POSES * L_FRAME48k]; /* TODO(sgi): Need conversion */
		int16_t numSamplesPerChannelCacheSize;
		int16_t numSamplesPerChannelToDecode;
		int16_t numSamplesPerChannelToSplitEncode;
		int16_t i, j;
		@@ -1153,11 +1154,12 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(
		hIvasDec->st_ivas->hRenderConfig->split_rend_config.dof == 0 )
		{
		numSamplesPerChannelToSplitEncode = (int16_t) ( output_Fs / FRAMES_PER_SEC / MAX_PARAM_SPATIAL_SUBFRAMES );
		numSamplesPerChannelCacheSize = numSamplesPerChannelToDecode - numSamplesPerChannelToSplitEncode;

		if ( hSplitBinRend->tdDataOut == NULL )
		{
		/* Allocate enough space to save all decoded samples that will not be split encoded directly after decoding */
		hSplitBinRend->tdDataOut = malloc( ( numSamplesPerChannelToDecode - numSamplesPerChannelToSplitEncode ) * BINAURAL_CHANNELS * numPoses * sizeof( float ) );
		hSplitBinRend->tdDataOut = malloc( numSamplesPerChannelCacheSize * BINAURAL_CHANNELS * numPoses * sizeof( float ) );
		if ( hSplitBinRend->tdDataOut == NULL )
		{
		return IVAS_ERR_FAILED_ALLOC;
		@@ -1167,6 +1169,7 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(
		else
		{
		numSamplesPerChannelToSplitEncode = (int16_t) ( output_Fs / FRAMES_PER_SEC );
		numSamplesPerChannelCacheSize = 0;
		}

		if ( output_config != AUDIO_CONFIG_BINAURAL_SPLIT_CODED && output_config != AUDIO_CONFIG_BINAURAL_SPLIT_PCM )
		@@ -1189,7 +1192,7 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(
		}
		assert( numSamplesPerChannelToDecode == *nOutSamples );
		#ifdef DEBUGGING
		dbgwrite( output_int, sizeof( int16_t ), 960 * 2, 1, "res/output_int.raw" );
		dbgwrite( output_int, sizeof( int16_t ), numSamplesPerChannelToDecode * 2, 1, "res/output_int.pcm" );
		#endif

		/* copy to cache if cache is in use */
		@@ -1203,7 +1206,7 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(

		while ( readPtr != readEnd )
		{
		writePtr++ = readPtr++; /* TODO: test */
		writePtr++ = readPtr++;
		}
		hSplitBinRend->numTdSamplesPerChannelCached = *nOutSamples - numSamplesPerChannelToSplitEncode;
		}
		@@ -1215,19 +1218,19 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(

		int16_t *writePtr;
		float readPtr, readEnd;
		readPtr = hSplitBinRend->tdDataOut + ( numSamplesPerChannelToDecode - hSplitBinRend->numTdSamplesPerChannelCached ) * BINAURAL_CHANNELS * numPoses;
		readPtr = hSplitBinRend->tdDataOut + ( numSamplesPerChannelCacheSize - hSplitBinRend->numTdSamplesPerChannelCached ) * BINAURAL_CHANNELS * numPoses;
		readEnd = readPtr + numSamplesPerChannelToSplitEncode * BINAURAL_CHANNELS * numPoses;
		writePtr = output_int;

		while ( readPtr != readEnd )
		{
		writePtr++ = readPtr++; /* TODO: test */
		writePtr++ = readPtr++;
		}
		hSplitBinRend->numTdSamplesPerChannelCached -= numSamplesPerChannelToSplitEncode;
		}

		/* [tmp] convert int back to float and change buffer layout */
		for ( i = 0; i < *nOutSamples; ++i )
		for ( i = 0; i < numSamplesPerChannelToSplitEncode; ++i )
		{
		#ifdef DEBUGGING
		for ( j = 0; j < BINAURAL_CHANNELS * numPoses; ++j )
		@@ -1241,7 +1244,7 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(
		}
		}
		#ifdef DEBUGGING
		dbgwrite( output[0], sizeof( float ), 960, 1, "res/output.raw" );
		dbgwrite( output[0], sizeof( float ), 240, 1, "res/output.pcm" );
		#endif
		max_band = (int16_t) ( ( BINAURAL_MAXBANDS * output_Fs ) / 48000 );
		pcm_out = ( output_config == AUDIO_CONFIG_BINAURAL_SPLIT_PCM ) ? 1 : 0;
		@@ -1251,6 +1254,9 @@ ivas_error IVAS_DEC_GetSplitBinauralBitstream(
		st_ivas->hHeadTrackData->Quaternion,
		st_ivas->hRenderConfig->split_rend_config.splitRendBitRate,
		st_ivas->hRenderConfig->split_rend_config.codec,
		#ifdef API_5MS
		st_ivas->hRenderConfig->split_rend_config.codec_frame_size_ms,
		#endif
		hSplitBinRend->hSplitRendBits,
		hSplitBinRend->hMultiBinCldfbData->Cldfb_RealBuffer_Binaural,
		hSplitBinRend->hMultiBinCldfbData->Cldfb_ImagBuffer_Binaural,
		@@ -2063,6 +2069,9 @@ static ivas_error copyRendererConfigStruct( RENDER_CONFIG_HANDLE hRCin, IVAS_REN
		hRCout->split_rend_config.dof = 3;
		hRCout->split_rend_config.hq_mode = 0;
		hRCout->split_rend_config.codec_delay_ms = 0;
		#ifdef API_5MS
		hRCout->split_rend_config.codec_frame_size_ms = 5;
		#endif
		hRCout->split_rend_config.codec = IVAS_SPLIT_REND_CODEC_DEFAULT;
		hRCout->split_rend_config.poseCorrectionMode = IVAS_SPLIT_REND_POSE_CORRECTION_MODE_CLDFB;
		hRCout->split_rend_config.rendererSelection = hRCin->split_rend_config.rendererSelection;
		@@ -2140,7 +2149,8 @@ ivas_error IVAS_DEC_GetDefaultRenderConfig(
		IVAS_RENDER_CONFIG_HANDLE hRCout /* o : Render config handle */
		)
		{
		RENDER_CONFIG_HANDLE hRCin;
		RENDER_CONFIG_DATA RCin;
		RENDER_CONFIG_HANDLE hRCin = &RCin;
		ivas_error error;

		if ( ( error = ivas_render_config_init_from_rom( &hRCin ) ) != IVAS_ERR_OK )

lib_rend/ivas_crend.c

+4 −0

Original line number	Diff line number	Diff line
		@@ -1624,7 +1624,11 @@ ivas_error ivas_rend_crendProcessSubframe(
		{
		int16_t subframe_idx, subframe_len;
		int16_t nchan_out, nchan_in, ch, first_sf, last_sf, slot_size, slots_to_render;
		#ifdef FIX_JBM_MC2SBA
		float *tc_local[MAX_OUTPUT_CHANNELS];
		#else
		float *tc_local[MAX_TRANSPORT_CHANNELS];
		#endif
		float pcm_tmp[BINAURAL_CHANNELS][L_FRAME48k];
		float *p_pcm_tmp[BINAURAL_CHANNELS];
		AUDIO_CONFIG in_config;