Issue 118: fix the DTX usage: default DTX up to 64 kbps, otherwise only in... (6ff997cc) · Commits · IVAS Codec Public Collaboration / IVAS Codec

lib_com/ivas_prot.h

+3 −0

Original line number	Diff line number	Diff line
		@@ -185,6 +185,9 @@ ivas_error pre_proc_front_ivas(
		const int16_t front_vad_flag, /* i : front-VAD flag to overwrite VAD decision */
		const int16_t force_front_vad, /* i : flag to force VAD decision */
		const int16_t front_vad_dtx_flag /* i : front-VAD DTX flag to overwrite VAD decision*/
		#ifdef FIX_DTX_RANGE
		,const int32_t ivas_total_brate /* i : IVAS total bitrate */
		#endif
		);

		ivas_error pre_proc_ivas(

lib_com/options.h

+1 −1

Original line number	Diff line number	Diff line
		@@ -149,7 +149,7 @@
		#define FIX_I106_TDREND_5MS /* Issue 106: 5 ms update rate in TD object renderer */
		#define ALIGN_SID_SIZE /* Issue 111: make all DTX modes use one SID frame bitrate (5.2 kbps) */
		#define FIX_135_MDCT_STEREO_MODE_UNINITIALIZED /* Issue 135: fix uninitialized value usage in SBA MDCT-Stereo core with PLC */

		#define FIX_DTX_RANGE /* Issue 118: fix the DTX usage: default DTX up to 64 kbps, otherwise only in silence */

lib_com/prot.h

+4 −1

Original line number	Diff line number	Diff line
		@@ -3890,6 +3890,9 @@ void td_cng_enc_init(

		void dtx(
		Encoder_State st, / i/o: encoder state structure */
		#ifdef FIX_DTX_RANGE
		const int32_t ivas_total_brate, /* i : IVAS total bitrate */
		#endif
		const int16_t vad, /* i : VAD flag for DTX */
		const float speech[] /* i : Pointer to the speech frame */
		);

lib_enc/amr_wb_enc.c

+4 −0

Original line number	Diff line number	Diff line
		@@ -311,7 +311,11 @@ void amr_wb_enc(
		st->fd_cng_reset_flag = 0;
		}

		#ifdef FIX_DTX_RANGE
		dtx( st, -1, vad_flag_dtx, inp );
		#else
		dtx( st, vad_flag_dtx, inp );
		#endif

		/----------------------------------------------------------------
		* Noise energy down-ward update and total noise energy estimation

lib_enc/dtx.c

+24 −1

Original line number	Diff line number	Diff line
		@@ -63,6 +63,10 @@

		#define LTE_VAR -4.0f

		#ifdef FIX_DTX_RANGE
		#define MAX_BRATE_DTX_EVS ACELP_24k40 /* maximum bitrate to which the default DTX is applied in EVS; otherwise DTX is applied only in silence */
		#define MAX_BRATE_DTX_IVAS IVAS_64k /* maximum bitrate to which the default DTX is applied in IVAS; otherwise DTX is applied only in silence */
		#endif

		/-------------------------------------------------------------------
		* Local function prototypes
		@@ -78,6 +82,9 @@ static void update_SID_cnt( DTX_ENC_HANDLE hDtxEnc, const int32_t core_brate, co

		void dtx(
		Encoder_State st, / i/o: encoder state structure */
		#ifdef FIX_DTX_RANGE
		const int32_t ivas_total_brate, /* i : IVAS total bitrate */
		#endif
		const int16_t vad, /* i : VAD flag for DTX */
		const float speech[] /* i : Pointer to the speech frame */
		)
		@@ -96,9 +103,15 @@ void dtx(
		}
		else
		{
		#ifdef FIX_DTX_RANGE
		last_br_cng_flag = st->last_total_brate_cng <= MAX_BRATE_DTX_EVS \|\| st->lp_noise < 15 \|\| ( st->element_mode == IVAS_SCE && st->last_total_brate_cng <= MAX_BRATE_DTX_IVAS );

		last_br_flag = st->last_total_brate <= MAX_BRATE_DTX_EVS \|\| st->lp_noise < 15 \|\| ( st->element_mode == IVAS_SCE && st->last_total_brate <= MAX_BRATE_DTX_IVAS );
		#else
		last_br_cng_flag = st->last_total_brate_cng <= ACELP_24k40 \|\| st->lp_noise < 15 \|\| ( ( st->element_mode == IVAS_SCE ) && st->last_total_brate_cng <= ACELP_32k );

		last_br_flag = st->last_total_brate <= ACELP_24k40 \|\| st->lp_noise < 15 \|\| ( ( st->element_mode == IVAS_SCE ) && st->last_total_brate <= ACELP_32k );
		#endif
		br_dtx_flag = 0;
		}

		@@ -174,8 +187,14 @@ void dtx(

		if ( st->dtx_sce_sba == 0 )
		{
		#ifdef FIX_DTX_RANGE
		br_dtx_flag = ( st->element_mode == EVS_MONO && st->total_brate <= MAX_BRATE_DTX_EVS ) \|\|
		( st->element_mode != EVS_MONO && ivas_total_brate <= MAX_BRATE_DTX_IVAS ) \|\|
		st->lp_noise < 15;
		#else
		br_dtx_flag = st->total_brate <= ACELP_24k40 \|\| st->lp_noise < 15 \|\| ( st->element_mode == IVAS_SCE && st->total_brate <= ACELP_32k ) \|\|
		st->element_mode == IVAS_CPE_DFT \|\| ( st->element_mode == IVAS_CPE_MDCT && ( st->element_brate <= IVAS_64k \|\| st->lp_noise < 15 ) );
		#endif
		}

		if ( st->Opt_DTX_ON && vad == 0 &&
		@@ -235,7 +254,11 @@ void dtx(
		}
		else
		{
		#ifdef FIX_DTX_RANGE
		if ( ( st->cng_type == FD_CNG && ( st->total_brate <= MAX_BRATE_DTX_EVS \|\| ( st->element_mode == IVAS_SCE && st->element_brate <= MAX_BRATE_DTX_IVAS ) ) ) \|\| ( st->element_mode == IVAS_CPE_MDCT ) ) /* at highest bitrates, use exclusively LP_CNG */
		#else
		if ( ( st->cng_type == FD_CNG && ( st->total_brate <= ACELP_24k40 \|\| ( st->element_mode == IVAS_SCE && st->total_brate <= ACELP_32k ) ) ) \|\| ( st->element_mode == IVAS_CPE_MDCT ) ) /* at highest bitrates, use exclusively LP_CNG */
		#endif
		{
		if ( st->element_mode == EVS_MONO && ( st->total_brate == ACELP_9k60 \|\| st->total_brate == ACELP_16k40 \|\| st->total_brate == ACELP_24k40 ) )
		{