Add float interface to lib_enc [untested] (441d8b14) · Commits · IVAS Codec Public Collaboration / IVAS Codec

apps/encoder.c

+39 −1

Original line number	Diff line number	Diff line
		@@ -695,13 +695,51 @@ int main(
		}
		}

		#ifdef FLOAT_INTERFACE
		float *tmpFloatBuf = NULL;
		bool useFloat = true; /* TODO(sgi): get from input file type or command line flag */
		if (useFloat)
		{
		/* TODO(sgi): Don't allocate on every frame */
		tmpFloatBuf = malloc(pcmBufSize * sizeof(float));
		for (int32_t i = 0; i < pcmBufSize; ++i)
		{
		tmpFloatBuf[i] = pcmBuf[i];
		}
		/* Feed input audio */
		if ( ( error = IVAS_ENC_FeedInputAudioFloat( hIvasEnc, tmpFloatBuf, pcmBufSize ) ) != IVAS_ERR_OK )
		{
		fprintf( stderr, "\nIVAS_ENC_FeedInputAudioInt failed: %s\n\n", IVAS_ENC_GetErrorMessage( error ) );
		return error;
		}
		} else {
		/* Feed input audio */
		if ( ( error = IVAS_ENC_FeedInputAudioInt( hIvasEnc, pcmBuf, pcmBufSize ) ) != IVAS_ERR_OK )
		{
		fprintf( stderr, "\nIVAS_ENC_FeedInputAudioInt failed: %s\n\n", IVAS_ENC_GetErrorMessage( error ) );
		return error;
		}
		}
		#endif

		/* * Encode one frame * */
		if ( ( error = IVAS_ENC_EncodeFrameToSerial( hIvasEnc, pcmBuf, pcmBufSize, bitStream, &numBits ) ) != IVAS_ERR_OK )
		if ( ( error = IVAS_ENC_EncodeFrameToSerial( hIvasEnc,
		#ifndef FLOAT_INTERFACE
		pcmBuf, pcmBufSize,
		#endif
		bitStream, &numBits ) ) != IVAS_ERR_OK )
		{
		fprintf( stderr, "\nencodeFrame failed: %s\n\n", IVAS_ENC_GetErrorMessage( error ) );
		goto cleanup;
		}

		#ifdef FLOAT_INTERFACE
		if (tmpFloatBuf != NULL)
		{
		free(tmpFloatBuf);
		}
		#endif

		/* write bitstream */
		if ( ( error = BS_Writer_WriteFrame_short( hBsWriter, bitStream, numBits, totalBitrate ) ) != IVAS_ERR_OK )
		{

lib_com/ivas_fb_mixer.c

+4 −0

Original line number	Diff line number	Diff line
		@@ -462,7 +462,11 @@ void ivas_FB_mixer_close(

		void ivas_fb_mixer_pcm_ingest(
		IVAS_FB_MIXER_HANDLE hFbMixer, /* i/o: FB mixer handle */
		#ifdef FLOAT_INTERFACE
		float* pcm_in[], /* i : input audio channels */
		#else
		float pcm_in[][L_FRAME48k], /* i : input audio channels */
		#endif
		float *ppOut_pcm, / o : output audio channels */
		const int16_t frame_len /* i : frame length */
		)

lib_com/ivas_prot.h

+105 −0

Original line number	Diff line number	Diff line
		@@ -33,6 +33,7 @@
		#ifndef IVAS_PROT_H
		#define IVAS_PROT_H

		#include "ivas_cnst.h"
		#include <stdint.h>
		#include "options.h"
		#include <stdio.h>
		@@ -56,14 +57,22 @@

		ivas_error ivas_enc(
		Encoder_Struct st_ivas, / i : IVAS encoder structure */
		#ifdef FLOAT_INTERFACE
		float data, / i : input signal */
		#else
		const int16_t data, / i : input signal */
		#endif
		const int16_t n_samples /* i : number of input samples */
		);

		void stereo_dmx_evs_enc(
		STEREO_DMX_EVS_ENC_HANDLE hStereoDmxEVS, /* i/o: Stereo downmix for EVS encoder handle */
		const int32_t input_Fs, /* i : input sampling rate */
		#ifdef FLOAT_INTERFACE
		float data[CPE_CHANNELS], /* i/o: input signal */
		#else
		int16_t data[CPE_CHANNELS * L_FRAME48k], /* i/o: input signal */
		#endif
		const int16_t n_samples /* i : number of input samples */
		);

		@@ -177,7 +186,11 @@ ivas_error ivas_cpe_enc(

		ivas_error ivas_mct_enc(
		Encoder_Struct st_ivas, / i/o: IVAS encoder structure */
		#ifdef FLOAT_INTERFACE
		float* data[MCT_MAX_CHANNELS], /* i : input signals */
		#else
		float data[MCT_MAX_CHANNELS][L_FRAME48k], /* i : input signals */
		#endif
		const int16_t input_frame, /* i : input frame length per channel */
		const int16_t nb_bits_metadata /* i : number of metadata bits */
		);
		@@ -844,7 +857,11 @@ ivas_error create_ism_metadata_dec(

		ivas_error ivas_ism_enc(
		Encoder_Struct st_ivas, / i/o: IVAS encoder structure */
		#ifdef FLOAT_INTERFACE
		float* data[MAX_NUM_OBJECTS], /* i : input signal */
		#else
		float data[MAX_NUM_OBJECTS][L_FRAME48k], /* i : input signal */
		#endif
		const int16_t input_frame, /* i : input frame length per channel */
		int16_t nb_bits_metadata / i : number of metadata bits */
		);
		@@ -889,7 +906,11 @@ ivas_error ivas_param_ism_enc_open(

		void ivas_param_ism_enc(
		Encoder_Struct st_ivas, / i/o: IVAS encoder structure */
		#ifdef FLOAT_INTERFACE
		float* data[MAX_NUM_OBJECTS], /* i : input signal */
		#else
		float data[MAX_NUM_OBJECTS][L_FRAME48k], /* i : input signal */
		#endif
		const int16_t input_frame /* i : input frame length per channel */
		);

		@@ -900,7 +921,11 @@ void ivas_param_ism_enc_close(

		void ivas_param_ism_stereo_dmx(
		Encoder_Struct st_ivas, / i : IVAS encoder structure */
		#ifdef FLOAT_INTERFACE
		float* data[MAX_NUM_OBJECTS], /* i/o: input signal/stereo dmx */
		#else
		float data[MAX_NUM_OBJECTS][L_FRAME48k], /* i/o: input signal/stereo dmx */
		#endif
		const int16_t input_frame /* i : Length of input frame */
		);

		@@ -3090,7 +3115,11 @@ void ivas_dirac_param_est_enc(
		DIRAC_ENC_HANDLE hDirAC,
		IVAS_QDIRECTION *q_direction,
		const uint8_t useLowerRes,
		#ifdef FLOAT_INTERFACE
		float* data_f[],
		#else
		float data_f[][L_FRAME48k],
		#endif
		float **pp_fr_real,
		float **pp_fr_imag,
		const int16_t input_frame,
		@@ -3170,14 +3199,32 @@ int16_t ivas_sba_get_spar_hoa_md_flag(
		);

		void ivas_sba_zero_vert_comp(
		#ifdef FLOAT_INTERFACE
		float* sba_data[], /* i/o: SBA data frame */
		#else
		float sba_data[][L_FRAME48k], /* i/o: SBA data frame */
		#endif
		const int16_t sba_order, /* i : Ambisonic (SBA) order */
		const int16_t sba_planar, /* i : SBA planar flag */
		const int16_t input_frame /* i : input frame length */
		);

		#ifdef FLOAT_INTERFACE
		/* TODO(sgi): tmp function, remove when FLOAT_INTERFACE is done */
		void ivas_sba_zero_vert_comp_old(
		float sba_data[][L_FRAME48k], /* i/o: SBA data frame */
		const int16_t sba_order, /* i : Ambisonic (SBA) order */
		const int16_t sba_planar, /* i : SBA planar flag */
		const int16_t input_frame /* i : input frame length */
		);
		#endif

		void ivas_sba_getTCs(
		#ifdef FLOAT_INTERFACE
		float* sba_data[], /* i : SBA signals */
		#else
		float sba_data[][L_FRAME48k], /* i : SBA signals */
		#endif
		Encoder_Struct st_ivas, / i/o: Encoder struct */
		const int16_t input_frame /* i : frame length */
		);
		@@ -3192,10 +3239,23 @@ ivas_error ivas_sba_linear_renderer(
		);

		int16_t ivas_sba_remapTCs(
		#ifdef FLOAT_INTERFACE
		float* sba_data[], /* i/o: SBA signals */
		#else
		float sba_data[][L_FRAME48k], /* i/o: SBA signals */
		#endif
		Decoder_Struct st_ivas, / i/o: decoder struct */
		const int16_t output_frame /* i : frame length */
		);

		#ifdef FLOAT_INTERFACE
		/* TODO(sgi): tmp function, remove when FLOAT_INTERFACE is done */
		int16_t ivas_sba_remapTCs_old(
		float sba_data[][L_FRAME48k], /* i/o: SBA signals */
		Decoder_Struct st_ivas, / i/o: decoder struct */
		const int16_t output_frame /* i : frame length */
		);
		#endif

		void ivas_sba_dirac_stereo_dec(
		Decoder_Struct st_ivas, / i/o: IVAS decoder structure */
		@@ -3248,7 +3308,11 @@ void ivas_dirac_enc(
		BSTR_ENC_HANDLE hMetaData, /* i/o: Metadata bitstream handle */
		int16_t nb_bits_metadata, / o : number of metadata bits written */
		const int16_t Opt_DTX_ON, /* i : flag signaling DTX on */
		#ifdef FLOAT_INTERFACE
		float* data_f[], /* i/o: SBA channels */
		#else
		float data_f[][L_FRAME48k], /* i/o: SBA channels */
		#endif
		const int16_t input_frame, /* i : input frame length */
		const int16_t sba_planar /* i : SBA planar flag */
		);
		@@ -3588,7 +3652,11 @@ void ivas_param_mc_enc_close(
		void ivas_param_mc_enc(
		Encoder_Struct st_ivas, / i/o: IVAS Encoder handle */
		BSTR_ENC_HANDLE hMetaData, /* i/o: IVAS Metadata bitstream handle */
		#ifdef FLOAT_INTERFACE
		float* data_f[], /* i/o: input: CICP6, CICP12, CICP14, CICP16 or CICP19 MC data */
		#else
		float data_f[][L_FRAME48k], /* i/o: input: CICP6, CICP12, CICP14, CICP16 or CICP19 MC data */
		#endif
		const int16_t input_frame /* i : input frame length */
		);

		@@ -3821,7 +3889,11 @@ void ivas_spar_enc_close(

		ivas_error ivas_spar_enc(
		Encoder_Struct st_ivas, / i/o: IVAS encoder structure */
		#ifdef FLOAT_INTERFACE
		float* data_f[], /* i/o: input/transport audio channels */
		#else
		float data_f[][L_FRAME48k], /* i/o: input/transport audio channels */
		#endif
		const int16_t input_frame, /* i : input frame length */
		int16_t nb_bits_metadata, / i : number of MD bits written */
		BSTR_ENC_HANDLE hMetaData /* o : MetaData handle */
		@@ -4493,7 +4565,11 @@ void ivas_masa_encode(

		void ivas_masa_estimate_energy(
		MASA_ENCODER_HANDLE hMasa, /* i/o: MASA encoder structure */
		#ifdef FLOAT_INTERFACE
		float* data_f[], /* i : Input audio channels */
		#else
		float data_f[][L_FRAME48k], /* i : Input audio channels */
		#endif
		const int16_t input_frame, /* i : frame length */
		const int16_t nchan_transport /* i : number of MASA input/transport channels */
		);
		@@ -4923,7 +4999,11 @@ void ivas_mcmasa_enc(
		MCMASA_ENC_HANDLE hMcMasa, /* i/o: Encoder McMASA handle */
		IVAS_QMETADATA_HANDLE hQMeta, /* o : Qmetadata handle */
		MASA_ENCODER_HANDLE hMasa, /* i/o: Encoder MASA handle */
		#ifdef FLOAT_INTERFACE
		float* data_f[], /* i : Input frame of audio */
		#else
		float data_f[][L_FRAME48k], /* i : Input frame of audio */
		#endif
		const int16_t input_frame, /* i : Input frame size */
		const int16_t nchan_transport, /* i : Number of transport channels */
		const int16_t nchan_inp /* i : Number of input channels */
		@@ -4932,7 +5012,11 @@ void ivas_mcmasa_enc(
		void ivas_mcmasa_param_est_enc(
		MCMASA_ENC_HANDLE hMcMasa, /* i/o: Encoder McMASA handle */
		MASA_ENCODER_HANDLE hMasa, /* i/o: Encoder MASA handle */
		#ifdef FLOAT_INTERFACE
		float* data_f[], /* i : Input frame of audio */
		#else
		float data_f[][L_FRAME48k], /* i : Input frame of audio */
		#endif
		float elevation_m_values[MAX_PARAM_SPATIAL_SUBFRAMES][MASA_FREQUENCY_BANDS], /* o : Estimated elevation */
		float azimuth_m_values[MAX_PARAM_SPATIAL_SUBFRAMES][MASA_FREQUENCY_BANDS], /* o : Estimated azimuth */
		float energyRatio[MAX_PARAM_SPATIAL_SUBFRAMES][MASA_FREQUENCY_BANDS], /* o : Estimated direct-to-total ratio*/
		@@ -5317,7 +5401,11 @@ void ivas_FB_mixer_close(

		void ivas_fb_mixer_pcm_ingest(
		IVAS_FB_MIXER_HANDLE hFbMixer, /* i/o: FB mixer handle */
		#ifdef FLOAT_INTERFACE
		float* pcm_in[], /* i : input audio channels */
		#else
		float pcm_in[][L_FRAME48k], /* i : input audio channels */
		#endif
		float *ppOut_pcm, / o : output audio channels */
		const int16_t frame_length /* i : frame length */
		);
		@@ -5325,7 +5413,11 @@ void ivas_fb_mixer_pcm_ingest(
		void ivas_dirac_enc_spar_delay_synchro(
		Encoder_Struct st_ivas, / i/o: IVAS encoder structure */
		const int16_t input_frame, /* i : input frame length */
		#ifdef FLOAT_INTERFACE
		float* data_f[] /* i/o: SBA channels (ACN / SN3D) */
		#else
		float data_f[][L_FRAME48k] /* i/o: SBA channels (ACN / SN3D) */
		#endif
		);

		void ivas_fb_mixer_update_prior_input(
		@@ -5717,6 +5809,19 @@ ivas_error ivas_orient_trk_GetTrackedOrientation(
		float *roll
		);

		#ifdef FLOAT_INTERFACE
		/----------------------------------------------------------------------------------
		* Audio buffer handling
		----------------------------------------------------------------------------------/

		void save_channel_pointers(
		float* linearBuffer,
		int16_t numChannels,
		int16_t numSamplesPerChannel,
		float** channelArray
		);

		#endif

		/* clang-format on */

lib_com/ivas_sba_config.c

+33 −0

Original line number	Diff line number	Diff line
		@@ -321,6 +321,38 @@ int16_t ivas_sba_get_spar_hoa_md_flag(
		-------------------------------------------------------------------/

		void ivas_sba_zero_vert_comp(
		#ifdef FLOAT_INTERFACE
		float* sba_data[], /* i : SBA signals */
		#else
		float sba_data[][L_FRAME48k], /* i : SBA signals */
		#endif
		const int16_t sba_order, /* i : SBA order */
		const int16_t sba_planar, /* i : SBA planar flag */
		const int16_t input_frame /* i : frame length */
		)
		{
		int16_t i, j;

		/* Channels in the range i^2+1 to (i+1)^2 -1 are zeroed (retain only first and last channel for that order) */
		for ( i = 1; i <= sba_order; i++ )
		{
		/* Keep Z if not planar */
		if ( !sba_planar && i == 1 )
		{
		continue;
		}

		for ( j = ( i * i + 1 ); j < ( ( i + 1 ) * ( i + 1 ) - 1 ); j++ )
		{
		set_zero( sba_data[j], input_frame );
		}
		}

		return;
		}

		#ifdef FLOAT_INTERFACE
		void ivas_sba_zero_vert_comp_old(
		float sba_data[][L_FRAME48k], /* i : SBA signals */
		const int16_t sba_order, /* i : SBA order */
		const int16_t sba_planar, /* i : SBA planar flag */
		@@ -346,3 +378,4 @@ void ivas_sba_zero_vert_comp(

		return;
		}
		#endif

lib_com/ivas_tools.c

+32 −0

Original line number	Diff line number	Diff line
		@@ -1218,3 +1218,35 @@ float rand_triangular_signed(
		return 0.5f - 0.5f * sqrtf( 1.0f - rand_val );
		}
		}

		#ifdef FLOAT_INTERFACE

		void save_channel_pointers(
		float* linearBuffer,
		int16_t numChannels,
		int16_t numSamplesPerChannel,
		float** channelArray
		)
		{
		/* Save a pointer to the beginning of each channel in a linear buffer, so that 2D indexing into a 1D array is possible.
		*
		* With linearBuffer: 111111111111111122222222222222223333333333333333
		* Pointers: ^ ^ ^
		* channelArray[0] \| \|
		* channelArray[1] \|
		* channelArray[2]
		*
		* Example 2D indexing: channelArray[1][3] - returns 4th sample of the second channel
		*/

		int16_t i;

		for (i = 0; i < numChannels; ++i)
		{
		channelArray[i] = linearBuffer + i * numSamplesPerChannel;
		}

		return;
		}

		#endif
		No newline at end of file