updt (764c841d) · Commits · SA4 / Audio / IVAS BASOP

lib_enc/cng_enc_fx.c

+13 −3

Original line number	Diff line number	Diff line
		@@ -2973,7 +2973,12 @@ void swb_CNG_enc_fx(
		void swb_CNG_enc_ivas_fx(
		#endif
		Encoder_State st, / i/o: State structure */
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		const Word16 shb_speech_fx_orig, / i : SHB target signal (6-14kHz) at 16kHz Q_shb_spch */
		const Word16 Q_shb_spch,
		#else
		const Word16 shb_speech_fx, / i : SHB target signal (6-14kHz) at 16kHz Q0 */
		#endif
		const Word16 syn_12k8_16k_fx / i : ACELP core synthesis at 12.8kHz or 16kHz Q0 */
		)
		{
		@@ -2987,6 +2992,11 @@ void swb_CNG_enc_ivas_fx(
		test();
		IF( GE_32( st->input_Fs, L_FRAME32k * FRAMES_PER_SEC ) )
		{
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		Word16 shb_speech_fx[L_FRAME16k];
		Copy_Scale_sig( shb_speech_fx_orig, shb_speech_fx, L_FRAME16k, negate( Q_shb_spch ) ); // Q0
		#endif

		/* decide if SHB SID encoding or not */
		#ifdef HARM_NON_LINEARITY
		shb_SID_updt = shb_DTX_fx( st, shb_speech_fx, syn_12k8_16k_fx );

lib_enc/evs_enc_fx.c

+4 −0

Original line number	Diff line number	Diff line
		@@ -528,7 +528,11 @@ ivas_error evs_enc_fx(
		test();
		IF( st->Opt_DTX_ON && GE_16( input_frame, L_FRAME32k ) )
		{
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		swb_CNG_enc_fx( st, shb_speech, 0, old_syn_12k8_16k );
		#else
		swb_CNG_enc_fx( st, shb_speech, old_syn_12k8_16k );
		#endif
		}

		/---------------------------------------------------------------------

lib_enc/ivas_core_enc_fx.c

+24 −36

Original line number	Diff line number	Diff line
		@@ -776,12 +776,6 @@ ivas_error ivas_core_enc_fx(
		st->cldfbSynTd->Q_cldfb_state = sub( q_re_im_buf[n], 1 );
		}

		#ifdef SCALING_SWB_SPEECH_32
		Word32 *new_swb_speech_fx;
		Word32 new_swb_speech_buffer_fx[L_FRAME48k + STEREO_DFT_OVL_MAX];
		set32_fx( new_swb_speech_buffer_fx, 0, L_FRAME48k + STEREO_DFT_OVL_MAX );
		new_swb_speech_fx = new_swb_speech_buffer_fx + STEREO_DFT_OVL_MAX;
		#endif
		#ifdef SCALING_SWB_SPEECH_32
		swb_pre_proc_ivas_fx( st, new_swb_speech_fx_16, shb_speech_fx, &Q_shb_spch, realBuffer_fx[n], imagBuffer_fx[n], q_re_im_buf[n], hCPE );
		#else
		@@ -798,9 +792,10 @@ ivas_error ivas_core_enc_fx(
		}
		}

		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		Word16 flag_shb_scale = 0; // just temporary to keep BE, it will disappear
		#ifdef REMOVE_SCALING_SHB_SPEECH_NONBE
		Word16 shift_shb_spch = 0;
		#endif
		Q_shb_spch = 0;
		Word16 Q_shb_spch_16 = Q_shb_spch;
		move16();
		#ifdef FIX_2344_ALIGN_PREPROC
		@@ -808,11 +803,15 @@ ivas_error ivas_core_enc_fx(
		// IF( st->tcxonly == 0 )
		#endif
		{
		#ifndef REMOVE_SCALING_SHB_SPEECH_BE
		shift = add( getScaleFactor16( shb_speech_fx, L_FRAME16k ), Q16 );
		#ifdef REMOVE_SCALING_SHB_SPEECH_NONBE
		shift_shb_spch = shift;
		#endif
		#ifndef REMOVE_SCALING_SHB_SPEECH_NONBE
		Copy_Scale_sig_16_32_no_sat( shb_speech_fx, shb_speech_fx32, L_FRAME16k, shift ); // Q_shb_spch
		Q_shb_spch = add( Q_shb_spch, shift );
		#endif
		Q_shb_spch = add( Q_shb_spch, shift );

		IF( st->hBWE_FD )
		{
		shift = getScaleFactor16( st->hBWE_FD->L_old_wtda_swb_fx, L_FRAME48k );
		@@ -844,13 +843,9 @@ ivas_error ivas_core_enc_fx(
		}
		ELSE IF( EQ_16( st->extl, SWB_BWE ) \|\| EQ_16( st->extl, FB_BWE ) )
		{
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		flag_shb_scale = 1;
		shift = add( getScaleFactor16( shb_speech_fx, L_FRAME16k ), Q16 );
		Q_shb_spch = add( Q_shb_spch_16, shift );
		Copy_Scale_sig_16_32_no_sat( shb_speech_fx, shb_speech_fx32, L_FRAME16k, shift ); // Q_shb_spch
		#endif
		#ifndef REMOVE_SCALING_SHB_SPEECH_BE
		Copy_Scale_sig_32_16( shb_speech_fx32, shb_speech_fx, L_FRAME16k, -Q16 ); // Q_shb_spch - 16
		#endif
		/* SWB(FB) BWE encoder */
		swb_bwe_enc_ivas_fx( st, last_element_mode, old_inp_12k8_fx[n], old_inp_16k_fx[n], old_syn_12k8_16k_fx[n], new_swb_speech_fx_16, st->q_inp, shb_speech_fx, sub( Q_shb_spch, Q16 ), sub( Q_new[n], 1 ) );
		}
		@@ -867,18 +862,15 @@ ivas_error ivas_core_enc_fx(
		IF( st->hTdCngEnc != NULL && st->Opt_DTX_ON && ( GE_16( input_frame, L_FRAME32k ) \|\| EQ_16( st->element_mode, IVAS_CPE_DFT ) ) )
		{
		/* SHB DTX/CNG encoder */
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		if( flag_shb_scale == 0 )
		{
		flag_shb_scale = 1;
		shift = add( getScaleFactor16( shb_speech_fx, L_FRAME16k ), Q16 );
		Q_shb_spch = add( Q_shb_spch_16, shift );
		Copy_Scale_sig_16_32_no_sat( shb_speech_fx, shb_speech_fx32, L_FRAME16k, shift ); // Q_shb_spch
		}
		#endif
		#ifndef REMOVE_SCALING_SHB_SPEECH_BE
		Copy_Scale_sig_32_16( shb_speech_fx32, shb_speech_fx, L_FRAME16k, negate( Q_shb_spch ) ); // Q0
		#endif
		#ifdef HARM_NON_LINEARITY
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		swb_CNG_enc_fx( st, shb_speech_fx, Q_shb_spch_16, old_syn_12k8_16k_fx[n] );
		#else
		swb_CNG_enc_fx( st, shb_speech_fx /* Unmodified */, old_syn_12k8_16k_fx[n] );
		#endif
		#else
		swb_CNG_enc_ivas_fx( st, shb_speech_fx /* Unmodified */, old_syn_12k8_16k_fx[n] );
		#endif
		@@ -901,20 +893,16 @@ ivas_error ivas_core_enc_fx(
		#endif
		stereo_icBWE_preproc_fx( hCPE, input_frame, new_swb_speech_buffer_fx_16 /tmp buffer/, q_new_swb_speech_buffer );

		q_new_swb_speech_buffer = add( q_new_swb_speech_buffer, 16 );
		q_new_swb_speech_buffer = add( q_new_swb_speech_buffer, Q16 );
		#ifndef SCALING_SWB_SPEECH_32
		Copy_Scale_sig_16_32_no_sat( new_swb_speech_buffer_fx_16, new_swb_speech_buffer_fx, L_FRAME48k + STEREO_DFT_OVL_MAX, Q16 ); // q_new_swb_speech_buffer+st->q_inp - 16 - > q_new_swb_speech_buffer+st->q_inp
		Copy_Scale_sig_16_32_no_sat( voice_factors_fx[0], voice_factors_fx32[0], NB_SUBFR16k, Q16 ); // Q31
		#endif

		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		if ( flag_shb_scale == 0 )
		{
		shift = add( getScaleFactor16( shb_speech_fx, L_FRAME16k ), Q16 );
		Q_shb_spch = add( Q_shb_spch_16, shift );
		Copy_Scale_sig_16_32_no_sat( shb_speech_fx, shb_speech_fx32, L_FRAME16k, shift ); // Q_shb_spch
		}
		#ifdef REMOVE_SCALING_SHB_SPEECH_NONBE
		Copy_Scale_sig_16_32_no_sat( shb_speech_fx, shb_speech_fx32, L_FRAME16k, shift_shb_spch ); // Q_shb_spch
		#endif

		#ifdef SCALING_SWB_SPEECH_32
		stereo_icBWE_enc_fx( hCPE, shb_speech_fx32, sub( Q31, Q_shb_spch ), new_swb_speech_buffer_fx_16, sub( Q31, q_new_swb_speech_buffer ), voice_factors_fx[0] );
		#else

lib_enc/prot_fx_enc.h

+8 −3

Original line number	Diff line number	Diff line
		@@ -641,7 +641,12 @@ void swb_bwe_enc_ivas_fx(

		void swb_CNG_enc_fx(
		Encoder_State st_fx, / i/o: State structure */
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		const Word16 shb_speech_fx_orig, / i : SHB target signal (6-14kHz) at 16kHz Q_shb_spch */
		const Word16 Q_shb_spch,
		#else
		const Word16 shb_speech_fx, / i : SHB target signal (6-14kHz) at 16kHz */
		#endif
		const Word16 syn_12k8_16k_fx / i : ACELP core synthesis at 12.8kHz or 16kHz */
		);

lib_enc/swb_bwe_enc_fx.c

+11 −1

Original line number	Diff line number	Diff line
		@@ -262,6 +262,7 @@ void wb_bwe_enc_ivas_fx(
		*
		* SWB BWE encoder (only for 32kHz signals)
		-------------------------------------------------------------------/

		void swb_bwe_enc_ivas_fx(
		Encoder_State st_fx, / i/o: encoder state structure */
		const Word16 last_element_mode, /* i : last element mode */
		@@ -270,7 +271,11 @@ void swb_bwe_enc_ivas_fx(
		const Word16 old_syn_12k8_16k_fx, / i : ACELP core synthesis at 12.8kHz or 16kHz */
		const Word16 new_swb_speech_fx, / i : original input signal at 32kHz */
		const Word16 Q_new_swb_speech, /* i : Q for new_swb_speech_fx */
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		Word16 shb_speech_fx_orig, / i : SHB target signal (6-14kHz) at 16kHz */
		#else
		Word16 shb_speech_fx, / i : SHB target signal (6-14kHz) at 16kHz */
		#endif
		Word16 Q_shb_speech,
		Word16 Q_slb_speech )
		{
		@@ -304,6 +309,11 @@ void swb_bwe_enc_ivas_fx(
		move16();
		Word16 fb_band_begin;
		Word16 q_new_input_hp;
		#ifdef REMOVE_SCALING_SHB_SPEECH_BE
		Word16 shb_speech_fx[L_FRAME16k];

		Copy_Scale_sig( shb_speech_fx_orig, shb_speech_fx, L_FRAME16k, Q_shb_speech ); // Q_shb_spch
		#endif

		FD_BWE_ENC_HANDLE hBWE_FD = st_fx->hBWE_FD;
		TD_BWE_ENC_HANDLE hBWE_TD = st_fx->hBWE_TD;