remove SIMPLIFY_SCALING_OLD_INP (4c8bdbdf) · Commits · SA4 / Audio / IVAS BASOP

lib_com/ivas_prot_fx.h

+0 −15

Original line number	Diff line number	Diff line
		@@ -3077,24 +3077,15 @@ void InternalTCXDecoder_fx(

		void stereo_tcx_core_enc(
		Encoder_State st, / i/o: encoder state structure */
		#ifdef SIMPLIFY_SCALING_OLD_INP
		Word16 new_samples_12k8[], /* i : buffer of input signal @12.8 kHz Q_new-1*/
		Word16 new_samples_16k[], /* i : buffer of input signal @16 kHz Q_new-1*/
		#else
		const Word16 new_samples_12k8[], /* i : buffer of input signal @12.8 kHz */
		const Word16 new_samples_16k[], /* i : buffer of input signal @16 kHz */
		#endif
		const Word16 Aw_fx[], /* i : weighted A(z) unquant. for subframes,Q12*/
		Word16 lsp_new_fx[], /* i : LSPs at the end of the frame, Q15 */
		Word16 lsp_mid_fx[], /* i : LSPs in the middle of the frame, Q15 */
		Word16 pitch_buf_fx[NB_SUBFR16k], /* o : pitch for each subframe, Q6 */
		const Word16 last_element_mode, /* i : last element mode, Q0 */
		const Word16 vad_hover_flag, /* i : VAD hangover flag, Q0 */
		#ifdef SIMPLIFY_SCALING_OLD_INP
		Word16 Q_new_inp /* i : Q factor of input buffer */
		#else
		Word16 Q_new
		#endif
		);

		Word16 transient_analysis_ivas_fx(
		@@ -6404,9 +6395,6 @@ UWord16 get_indice_st(
		void stereo_mdct_core_enc_fx(
		CPE_ENC_HANDLE hCPE, /* i/o: CPE encoder structure */
		Word16 new_samples[CPE_CHANNELS][L_INP], /* i : new samples Qnew-1*/
		#ifdef SIMPLIFY_SCALING_OLD_INP
		Word16 Q_new_inp[CPE_CHANNELS], /* i : Q factor of input buffer */
		#endif
		Word16 old_wsp[CPE_CHANNELS][L_WSP], /* i : 12.8kHz weighted speech (for LTP Qx*/
		Word16 pitch_buf_fx[CPE_CHANNELS][NB_SUBFR16k] /* o : floating pitch for each subframe Q6*/
		);
		@@ -6454,9 +6442,6 @@ void stereo_switching_dec(
		void ivas_mdct_core_whitening_enc_fx(
		CPE_ENC_HANDLE hCPE, /* i/o: CPE encoder structure */
		Word16 new_samples_fx[CPE_CHANNELS][L_INP], /* i : new samples */
		#ifdef SIMPLIFY_SCALING_OLD_INP
		Word16 Q_new_inp[CPE_CHANNELS], /* i : Q factor of input buffer */
		#endif
		Word16 old_wsp_fx[CPE_CHANNELS][L_WSP], /* i : 12.8kHz weighted speech (for LTP */
		Word16 pitch_buf[CPE_CHANNELS][NB_SUBFR16k], /* o : floating pitch for each subframe */
		Word32 mdst_spectrum_long[CPE_CHANNELS], / o : buffer for MDST spectrum */

lib_com/options.h

+0 −1

Original line number	Diff line number	Diff line
		@@ -107,7 +107,6 @@
		#define FIX_2330_CLANG_18_WARNINGS_REND /* FhG: Fix renderer warnings */
		#define FIX_BASOP_2350_HARM_0B_BWE /* VA: basop issue 2350: harmonization of the 0b BWE */
		#define FIX_2349_HARM_FIND_UV /* VA: basop issue 2349: harmonization of find_uv() function */
		//#define SIMPLIFY_SCALING_OLD_INP
		//#define REMOVE_UNUSED_BUFFER_CORE_ENC
		#define REMOVE_SCALING_SHB_SPEECH_BE
		#define SCALING_SWB_SPEECH_32

lib_enc/ivas_core_enc_fx.c

+0 −23

Original line number	Diff line number	Diff line
		@@ -359,10 +359,8 @@ ivas_error ivas_core_enc_fx(
		TCX_ENC_HANDLE hTcxEnc = st->hTcxEnc;
		Word16 Q_spec_old, L_spec;

		#ifndef SIMPLIFY_SCALING_OLD_INP
		Scale_sig( old_inp_12k8_fx[n], L_INP_12k8, sub( Q1, Q_new[n] ) ); // Q0
		Scale_sig( old_inp_16k_fx[n], L_INP, sub( Q1, Q_new[n] ) ); // Q0
		#endif
		Scale_sig( st->hTcxEnc->Txnq, L_FRAME32k / 2 + 64, sub( negate( 1 ), st->hTcxEnc->q_Txnq ) ); /* Q(-1) */
		st->hTcxEnc->q_Txnq = -Q1;
		move16();
		@@ -370,14 +368,10 @@ ivas_error ivas_core_enc_fx(
		move16();

		/* TCX core encoder */
		#ifdef SIMPLIFY_SCALING_OLD_INP
		stereo_tcx_core_enc( st, old_inp_12k8_fx[n] + L_INP_MEM, old_inp_16k_fx[n] + L_INP_MEM, Aw_fx[n], lsp_new_fx[n], lsp_mid_fx[n], pitch_buf_fx[n], last_element_mode, vad_hover_flag[0], Q_new[n] );
		#else
		stereo_tcx_core_enc( st, old_inp_12k8_fx[n] + L_INP_MEM, old_inp_16k_fx[n] + L_INP_MEM, Aw_fx[n], lsp_new_fx[n], lsp_mid_fx[n], pitch_buf_fx[n], last_element_mode, vad_hover_flag[0], 0 );

		Scale_sig( old_inp_12k8_fx[n], L_INP_12k8, sub( Q_new[n], Q1 ) ); // Q_new[n] - 1
		Scale_sig( old_inp_16k_fx[n], L_INP, sub( Q_new[n], Q1 ) ); // Q_new[n] - 1
		#endif

		Scale_sig( st->hTcxEnc->old_out_fx, L_FRAME32k, negate( st->hTcxEnc->Q_old_out ) ); // scaling back to Q0
		st->hTcxEnc->Q_old_out = 0;
		@@ -458,9 +452,7 @@ ivas_error ivas_core_enc_fx(
		FOR( n = 0; n < n_CoreChannels; n++ )
		{
		st = sts[n];
		#ifndef SIMPLIFY_SCALING_OLD_INP
		Scale_sig( old_inp_16k_fx[n], L_INP, sub( Q1, Q_new[n] ) ); // Q0
		#endif
		IF( NE_16( st->element_mode, IVAS_CPE_DFT ) && NE_16( st->element_mode, IVAS_CPE_MDCT ) )
		{
		st->hTcxEnc->exp_buf_speech_ltp = st->exp_buf_speech_enc;
		@@ -473,22 +465,15 @@ ivas_error ivas_core_enc_fx(
		Q_spec_old[1] = hCPE->hCoreCoder[1]->hTcxEnc->spectrum_long_e;
		move16();

		#ifdef SIMPLIFY_SCALING_OLD_INP
		ivas_mdct_core_whitening_enc_fx( hCPE, old_inp_16k_fx, Q_new, old_wsp_fx, pitch_buf_fx_new, hMCT->p_mdst_spectrum_long_fx[cpe_id], hMCT->tnsBits[cpe_id], hMCT->p_orig_spectrum_long_fx[cpe_id],
		hMCT->tnsSize[cpe_id], hMCT->p_param[cpe_id], hMCT->hBstr, 1, hMCT->nchan_out_woLFE, mdst_spectrum_e, orig_spectrum_e );
		#else
		ivas_mdct_core_whitening_enc_fx( hCPE, old_inp_16k_fx, old_wsp_fx, pitch_buf_fx_new, hMCT->p_mdst_spectrum_long_fx[cpe_id], hMCT->tnsBits[cpe_id], hMCT->p_orig_spectrum_long_fx[cpe_id],
		hMCT->tnsSize[cpe_id], hMCT->p_param[cpe_id], hMCT->hBstr, 1, hMCT->nchan_out_woLFE, mdst_spectrum_e, orig_spectrum_e );
		#endif

		FOR( i = 0; i < CPE_CHANNELS; i++ )
		{
		st = sts[i];
		Word16 nSubframes = NB_DIV;
		move16();
		#ifndef SIMPLIFY_SCALING_OLD_INP
		Scale_sig( old_inp_16k_fx[i], L_INP, sub( Q_new[i], Q1 ) ); // Q_new[n] - 1
		#endif
		if ( EQ_16( st->hTcxEnc->tcxMode, TCX_20 ) )
		{
		nSubframes = 1;
		@@ -553,9 +538,7 @@ ivas_error ivas_core_enc_fx(
		FOR( i = 0; i < CPE_CHANNELS; i++ )
		{
		st = sts[i];
		#ifndef SIMPLIFY_SCALING_OLD_INP
		Scale_sig( old_inp_16k_fx[i], L_INP, sub( Q1, Q_new[i] ) ); // Q0
		#endif
		test();
		IF( NE_16( st->element_mode, IVAS_CPE_DFT ) && NE_16( st->element_mode, IVAS_CPE_MDCT ) )
		{
		@@ -581,17 +564,11 @@ ivas_error ivas_core_enc_fx(
		Q_spec_old[1] = hCPE->hCoreCoder[1]->hTcxEnc->spectrum_long_e;
		move16();

		#ifdef SIMPLIFY_SCALING_OLD_INP
		stereo_mdct_core_enc_fx( hCPE, old_inp_16k_fx, Q_new, old_wsp_fx, pitch_buf_fx );
		#else
		stereo_mdct_core_enc_fx( hCPE, old_inp_16k_fx, old_wsp_fx, pitch_buf_fx );
		#endif

		FOR( i = 0; i < CPE_CHANNELS; i++ )
		{
		#ifndef SIMPLIFY_SCALING_OLD_INP
		Scale_sig( old_inp_16k_fx[i], L_INP, sub( Q_new[i], Q1 ) ); // Q_new[n] - 1
		#endif
		st = sts[i];
		IF( EQ_16( st->hTcxEnc->tcxMode, TCX_20 ) )
		{

lib_enc/ivas_mdct_core_enc_fx.c

+0 −18

Original line number	Diff line number	Diff line
		@@ -1118,9 +1118,6 @@ void enc_prm_igf_mdct(
		void ivas_mdct_core_whitening_enc_fx(
		CPE_ENC_HANDLE hCPE, /* i/o: CPE encoder structure */
		Word16 new_samples_fx[CPE_CHANNELS][L_INP], /* i : new samples Qnew-1*/
		#ifdef SIMPLIFY_SCALING_OLD_INP
		Word16 Q_new_inp[CPE_CHANNELS], /* i : Q factor of input buffer */
		#endif
		Word16 old_wsp_fx[CPE_CHANNELS][L_WSP], /* i : 12.8kHz weighted speech (for LTP Qx*/
		Word16 pitch_buf_fx[CPE_CHANNELS][NB_SUBFR16k], /* o : floating pitch for each subframe Q6*/
		Word32 mdst_spectrum_long_fx[CPE_CHANNELS], / o : buffer for MDST spectrum mdst_spectrum_e*/
		@@ -1186,13 +1183,6 @@ void ivas_mdct_core_whitening_enc_fx(
		move16();
		move16();

		#ifdef SIMPLIFY_SCALING_OLD_INP
		FOR( ch = 0; ch < CPE_CHANNELS; ch++ )
		{
		Scale_sig( new_samples_fx[ch], L_INP, sub( Q1, Q_new_inp[ch] ) ); // Q0
		}
		#endif

		/--------------------------------------------------------------
		* Initialization
		---------------------------------------------------------------/
		@@ -2490,14 +2480,6 @@ void ivas_mdct_core_whitening_enc_fx(
		}
		}

		#ifdef SIMPLIFY_SCALING_OLD_INP
		// scaling the input buffer back to Q_new-1 - actually not needed because the buffer is not used in the subsequent processing
		FOR( ch = 0; ch < CPE_CHANNELS; ch++ )
		{
		Scale_sig( new_samples_fx[ch], L_INP, sub( Q_new_inp[ch], Q1 ) ); // Q0
		}
		#endif

		pop_wmops();
		return;
		}

lib_enc/ivas_stereo_mdct_core_enc_fx.c

+0 −8

Original line number	Diff line number	Diff line
		@@ -130,9 +130,6 @@ static void sync_tcx_mode_fx(
		void stereo_mdct_core_enc_fx(
		CPE_ENC_HANDLE hCPE, /* i/o: CPE encoder structure */
		Word16 new_samples[CPE_CHANNELS][L_INP], /* i : new samples Q_new-1*/
		#ifdef SIMPLIFY_SCALING_OLD_INP
		Word16 Q_new_inp[CPE_CHANNELS], /* i : Q factor of input buffer */
		#endif
		Word16 old_wsp[CPE_CHANNELS][L_WSP], /* i : 12.8kHz weighted speech (for LTP Qx*/
		Word16 pitch_buf_fx[CPE_CHANNELS][NB_SUBFR16k] /* o : floating pitch for each subframe Q6*/
		)
		@@ -288,13 +285,8 @@ void stereo_mdct_core_enc_fx(
		* - TNS
		---------------------------------------------------------------/

		#ifdef SIMPLIFY_SCALING_OLD_INP
		ivas_mdct_core_whitening_enc_fx( hCPE, new_samples, Q_new_inp, old_wsp, pitch_buf_fx, p_mdst_spectrum_long_fx,
		tnsBits, p_orig_spectrum_long_fx, tnsSize, p_param, hBstr, 0, CPE_CHANNELS, mdst_spectrum_e, orig_spectrum_e );
		#else
		ivas_mdct_core_whitening_enc_fx( hCPE, new_samples, old_wsp, pitch_buf_fx, p_mdst_spectrum_long_fx,
		tnsBits, p_orig_spectrum_long_fx, tnsSize, p_param, hBstr, 0, CPE_CHANNELS, mdst_spectrum_e, orig_spectrum_e );
		#endif

		FOR( i = 0; i < CPE_CHANNELS; i++ )
		{