Commit 7a875c41 authored by vaclav's avatar vaclav
Browse files

- Merge remote-tracking branch 'remotes/origin/main' into... 

- Merge remote-tracking branch 'remotes/origin/main' into 880-functions-renaming-after-nonbe_unified_decoding_paths
parents c8df4f07 8009f2af

lib_com/options.h

+1 −1
Original line number Diff line number Diff line
@@ -174,12 +174,12 @@ 


#define NONBE_FIX_1176_OSBA_REVERB_JBM_ASAN_ERROR       /* Ericsson: Issue 1176, fix in TDREND_firfilt for subframes shorter than the filter length */

#define NONBE_FIX_1212_TONAL_MDCT_CONCEALMENT           /* FhG: Fix issue 1212, zero IGF spec also in 1st concealed frame */

#define NONBE_FIX_1204_MDCT_STEREO_NOISE_EST_SCALING    /* FhG: fixes for decoder-side noise level estimation in MDCT-Stereo to prevent noise bursts in stereo switching */

#define NONBE_1200_ISM_JBM_BRATE_SW_FLUSH               /* VA: issue 1200: fix bug in renderer flush in ISM JBM bitrate switching */

#define NONBE_FIX_1205_TD_STEREO_MOD_CT                 /* VA: fix mismatch of coder_type (mod_ct) btw. TD stereo encoder and decoder */

#define NONBE_1203_MDCT2DFT_SWITCHING                   /* VA: issue 1203: fix severe artifacts during MDCT to DFT stereo switching when MDCT ITD is not used */







/* ##################### End NON-BE switches ########################### */



/* ################## End DEVELOPMENT switches ######################### */

lib_dec/ivas_stereo_mdct_core_dec.c

+24 −0
Original line number Diff line number Diff line
@@ -620,24 +620,48 @@ static void run_min_stats( 
               computed only once (for ch == 0) and not again in the second run sive the outcome will be the same anyway */

            if ( ( will_estimate_noise_on_channel[0] == will_estimate_noise_on_channel[1] ) || ch == 0 )

            {

#ifdef NONBE_FIX_1204_MDCT_STEREO_NOISE_EST_SCALING

                float power_spec_scale_fac;



                /* calculate power spectrum from MDCT coefficients and estimated MDST coeffs */

                power_spec_scale_fac = 1.f / (float) ( L_FRAME16k * L_FRAME16k );

                power_spec[0] = spec_in[0] * spec_in[0] * power_spec_scale_fac;

                power_spec[L_FRAME16k - 1] = spec_in[L_FRAME16k - 1] * spec_in[L_FRAME16k - 1] * power_spec_scale_fac;

#else

                /* calculate power spectrum from MDCT coefficients and estimated MDST coeffs */

                power_spec[0] = spec_in[0] * spec_in[0];

                power_spec[L_FRAME16k - 1] = spec_in[L_FRAME16k - 1] * spec_in[L_FRAME16k - 1];

#endif

                for ( int16_t i = 1; i < L_FRAME16k - 1; i++ )

                {

                    float mdst;

                    mdst = spec_in[i + 1] - spec_in[i - 1];

#ifdef NONBE_FIX_1204_MDCT_STEREO_NOISE_EST_SCALING

                    power_spec[i] = power_spec_scale_fac * ( spec_in[i] * spec_in[i] + mdst * mdst );

#else

                    power_spec[i] = spec_in[i] * spec_in[i] + mdst * mdst;

#endif

                }

            }

#ifndef NONBE_FIX_1204_MDCT_STEREO_NOISE_EST_SCALING



            noisy_speech_detection( st->hFdCngDec, st->VAD && st->m_frame_type == ACTIVE_FRAME, power_spec );



            st->hFdCngDec->hFdCngCom->likelihood_noisy_speech = 0.99f * st->hFdCngDec->hFdCngCom->likelihood_noisy_speech + 0.01f * (float) st->hFdCngDec->hFdCngCom->flag_noisy_speech;



            st->lp_noise = st->hFdCngDec->lp_noise;

#endif

        }



#ifdef NONBE_FIX_1204_MDCT_STEREO_NOISE_EST_SCALING

        if ( st->core == TCX_20_CORE )

        {

            noisy_speech_detection( st->hFdCngDec, save_VAD[ch] && st->m_frame_type == ACTIVE_FRAME, x[ch][0] );

            st->hFdCngDec->hFdCngCom->likelihood_noisy_speech = 0.99f * st->hFdCngDec->hFdCngCom->likelihood_noisy_speech + 0.01f * (float) st->hFdCngDec->hFdCngCom->flag_noisy_speech;

            st->lp_noise = st->hFdCngDec->lp_noise;

        }

#endif



        if ( will_estimate_noise_on_channel[0] || will_estimate_noise_on_channel[1] || st->bfi )

        {

            ApplyFdCng( NULL, st->bfi ? NULL : power_spec, NULL, NULL, st, st->bfi, 0 );

lib_enc/ivas_decision_matrix_enc.c

+1 −1
Original line number Diff line number Diff line
@@ -219,7 +219,7 @@ void ivas_decision_matrix_enc( 
#ifdef NONBE_FIX_1205_TD_STEREO_MOD_CT

        /* In TD stereo below 24.4 kbps we cannot overwrite the `coder_type` when it is set to TRANSITION, */

        /* as it is used for TD stereo bit allocation. To ensure consistent bit allocation, it must remain unchanged on the decoder side. */

        if ( st->idchan == 0 && !( element_brate <= IVAS_16k4 && st->coder_type == TRANSITION && st->element_mode == IVAS_CPE_TD ) )

        if ( st->idchan == 0 && !( element_brate < IVAS_24k4 && st->coder_type == TRANSITION && st->element_mode == IVAS_CPE_TD ) )

        {

            st->coder_type = AUDIO;

        }

scripts/parse_xml_report.py

+1 −1
Original line number Diff line number Diff line
@@ -44,7 +44,7 @@ IVAS_CATEGORIES = { 
EVS_CATEGORIES = {

    "Normal operation": r".*",

    "DTX": r"DTX",

    "PLC": r"b10|f06",

    "PLC": r"b10|f06|EPF",

    "Bitrate switching": r"sw",

    "JBM": r"JBM",

}

tests/test_26444.py

+4 −4
Original line number Diff line number Diff line
@@ -144,8 +144,8 @@ def test_evs_26444( 
            if enc_opts:

                for file in [ref, test]:

                    output_config = ""  # Empty for EVS operation

                    in_file = file

                    out_file = file + ".wav"

                    in_file = os.path.abspath(file)

                    out_file = os.path.abspath(file + ".wav")

                    add_option_list = []

                    ref_decoder_frontend.run(

                        output_config,
@@ -155,8 +155,8 @@ def test_evs_26444( 
                        add_option_list=add_option_list,

                    )

                fs = int(sampling_rate) * 1000

                reffile = ref + ".wav"

                testfile = test + ".wav"

                reffile = os.path.abspath(ref + ".wav")

                testfile = os.path.abspath(test + ".wav")

            else:

                fs = int(re.search(r"(\d+)kHz", ref).group(1)) * 1000

                # pyaudio3dtools.audiofile.readfile only handles .wav, .pcm and .raw suffixes.