hash stringlengths 32 32 | doc_id stringlengths 7 13 | section stringlengths 3 121 | content stringlengths 0 2.2M |
|---|---|---|---|
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6 Advanced spectral extension tool - A-SPX | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.1 Data and control interfaces | The A-SPX tool is used to parametrically code the higher frequencies of the audio signal. Interleaved waveform coding can be employed in addition, if the parametric model cannot re-instate the correct frequency components, or to achieve a more accurate temporal reproduction of the signal. Input: • QinASPX,a QMF subsamp... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.2 A-SPX specific variables | • sbg_master Array of master QMF subband groups, of length num_sbg_master + 1. • sbg_sig_highres Array of QMF subband group borders for high frequency resolution signal envelopes; the array is of length num_sbg_sig_highres + 1. • sbg_sig_lowres Array of QMF subband group borders for low frequency resolution signal enve... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3 Decoding A-SPX control data | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1 Subband groups | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1.0 Introduction | QMF subband samples are combined to form subband groups (sbg). These are described by subband group tables of the following types: • Template subband group tables, sbg_template_[high|low]res Static high-resolution and low-resolution subband group tables serving as a template for deriving the master table. • Master subb... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1.1 Master subband group table | Two static subband group tables are defined: one for low-frequency resolutions (sbg_template_lowres) and one for high-frequency resolutions (sbg_template_highres). The master table is derived from these statically pre-defined tables. The template subband group tables are defined as: sbg_template_lowres = [10,11,12,13,1... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1.2 Signal envelope subband group tables | The subband group table for high-frequency resolution signal envelopes, sbg_sig_highres, covers the upper subband groups from sbg_master. For that, sbg_master gets split into two parts using the A-SPX cross over subband offset, aspx_xover_subband_offset. The aspx_xover_subband_offset element is an index into the master... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1.3 Noise subband group table | The noise envelope subband group table, sbg_sig_noise, is obtained from sbg_sig_lowres as shown in the following pseudocode. Pseudocode 70 num_sbg_noise = max(1, floor(aspx_noise_sbg * log2(sbz/sbx) + 0.5)); idx[0] = 0; sbg_noise[0] = sbg_sig_lowres[0]; for (sbg = 1; sbg <= num_sbg_noise; sbg++) { idx[sbg] = idx[sbg-1]... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1.4 Patch subband group table | The patch subband group tables, sbg_patch_num_sb, sbg_patch_start_sb, and the number of patches, num_sbg_patches, as well as the derived table containing the patch borders in the A-SPX range, sbg_patches, are created using the following pseudocode. Pseudocode 71 msb = sba;usb = sbx; num_sbg_patches = 0; if (base_samp_f... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.1.5 Limiter subband group table | The limiter subband group table, sbg_lim, and the number of limiter subband groups, num_sbg_lim, is derived from the low frequency resolution table, sbg_sig_lowres, and the patch table, sbg_patches, as shown in the following pseudocode. It has two limiter subband groups per octave. Pseudocode 72 /* Copy sbg_sig_lowres ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.2 Low band filter and QMF delay line | The QMF time-frequency samples from the analysis filterbank, stored in QinASPX, are truncated at the crossover subband, sbx, and delayed by ts_offset_hfgen before being fed into the HF generator according to the following pseudocode. Pseudocode 75 for (ts = 0; ts < ts_offset_hfgen; ts++) { for (sb = 0; sb < num_qmf_sub... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.3 Time/frequency matrix | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.3.0 Introduction | The A-SPX time/frequency matrix is derived from the QMF matrix Q_low. Similarly, it has time on the horizontal axis and frequency on the vertical axis. The A-SPX matrix is generated using two methods: • framing, which determines the time slot borders of the signal and noise envelopes; and • tiling, which determines the... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.3.1 Framing | The temporal extent of the time/frequency grouping for A-SPX in each frame is defined as the A-SPX interval. Four possible A-SPX interval classes exist, represented by the bitstream element aspx_int_class: FIXFIX, FIXVAR, VARFIX, and VARVAR. The class name has two parts: the first denotes the nature of the start (left)... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.3.2 Tiling | On the frequency axis, two subband group tables are available for signal envelopes, sbg_sig_highres and sbg_sig_lowres, and one for the noise envelopes, sbg_noise. They are derived in clause 5.7.6.3.1.2 and clause 5.7.6.3.1.3, respectively. The resolution of the subband grouping for signal envelopes depends on atsg_fre... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.4 Decoding A-SPX signal and noise envelopes | For each A-SPX signal envelope and noise envelope, the signal and noise scale factors are delta coded either along time or frequency directions. Upon crossing an A-SPX interval boundary (along the time axis), the first envelope in the current A-SPX interval may be delta coded using Huffman decoding with respect to the ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.3.5 Dequantization and stereo decoding | Two quantization steps are possible for the quantization of the signal scale factors: • aspx_qmode_env[ch] = 0 corresponds to a quantization step of 1,5 dB; and • aspx_qmode_env[ch] = 1 corresponds to a quantization step of 3,0 dB. For a 1-channel element and for a 2-channel element with aspx_balance = 0, i.e. when ste... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4 HF signal construction | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.1 HF generator tool | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.1.1 Introduction | The HF generator patches subband signals from consecutive subbands of the matrix Q_low to consecutive subbands of the matrix Q_high based on parameters transmitted in the bitstream. The tonal to noise ratio of the subband signals Q_high are adjusted to the levels signalled in the bitstream by the encoder. Before the su... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.1.2 Pre-flattening control data calculation | The calculation of the pre-flattened control data involves the fitting of a third-order polynomial to the spectral envelope of the low band, Q_low. The fitted polynomial is a smoothed representation of the overall spectral slope of Q_low. The overall slope obtained is translated into a gain vector. The pseudocode for t... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.1.3 Subband tonal to noise ratio adjustment data calculation | The purpose of the per subband tonal to noise ratio adjustment is to adjust the tonal to noise ratio within the subbands of the patched signal and this is a two-step process. The first step is to perform linear prediction within each of the QMF subband signals of Q Low. The second step is the actual tonal to noise rati... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.1.4 HF signal creation | The high-frequency signal, Q_high, is created using the following pseudocode, which implements both the tonal to noise ratio adjustment of the lower spectrum as well as the pre-flattening. Pseudocode 89 /* Loop over QMF time slots */ for (ts = atsg_sig[0]*num_ts_in_ats; ts < atsg_sig[num_atsg_sig]*num_ts_in_ats; ts++) ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.2 HF envelope adjustment tool | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.2.0 Introduction | The envelope adjustment process takes Q_high as input from the HF generator and produces as output a QMF matrix Y as output. The adjustment is performed upon the entire A-SPX range spanning the time slots of the current A-SPX interval (given by atsg_sig) and num_sb_aspx QMF subbands, starting at subband sbx. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.2.1 Estimation of transmitted and actual envelopes in the current interval | The spectral envelopes within the current A-SPX interval are estimated depending on the bitstream element aspx_interpolation. The estimation is performed by taking the average of the squared complex subband samples over the time and frequency regions of the time-frequency matrix Q_high. The estimation can be calculated... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.2.2 Calculation of compensatory gains | In order for the QMF subband samples signals to retain the correct envelope, compensatory gains are calculated. The level of additional sinusoids as well as the level of the additional added noise are taken into account. The initial values for the gain, sig_gain_sb, are derived from the dequantized scale factors for no... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.3 Noise generator tool | Overview The noise generator tool is used to inject noise into the recreated high-frequency spectrum in order to decrease its tonality and to match the original signal more closely. The tool outputs a noise envelope that is subsequently added to the generated HF signal. Data and control interfaces Input: • noise_lev_sb... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.4 Tone generator tool | Overview The tone generator tool is used to match the tonality of the original signal by adding the appropriate missing sinusoids to the recreated high band. The tool outputs an envelope worth of values, the dimensions of which are specified in the bitstream by the encoder. Data and control interfaces Input: • sine_lev... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.4.5 HF signal assembling tool | The compensated gain values are applied to the input subband matrix, Q_high, for all signal envelopes of the current A-SPX interval as follows. NOTE: For variable A-SPX interval borders, the previously assembled HF signals, Y_prev, are pushed into the current output buffer Y. Pseudocode 106 atsg = 0; /* Get delayed QMF... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.5 Interleaved waveform coding | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.5.1 Introduction | A-SPX involves patching the spectral content of suitable low-frequency regions to higher frequencies and adjusting them so that they perceptually match the high-frequency content that was present in the original signal. The entire signal spectrum above the A-SPX crossover frequency is derived from the spectrum below th... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.5.2 Signalling interleaved waveform coding | The use of frequency-interleaved waveform coding in an A-SPX interval is indicated by the bitstream element aspx_fic_present. The presence of a frequency-interleaved waveform component in a particular subband group is indicated by the bitstream element aspx_fic_used_in_sfb. ETSI ETSI TS 103 190-1 V1.4.1 (2025-07) 231 F... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.6.5.3 Interleaving WCC and SEC | The waveform-coded components are contained in the matrix QinASPX. The spectral extension components are contained in the matrix Y. For frequency-interleaved waveform coding, the waveform-coded components are added to the spectral extension components as follows: ( Re{QoutASPXm,i)) =Re{QinASPXm,i-δASPX)) + Re*Ym, I... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7 Advanced coupling tool - A-CPL | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.1 Introduction | The A-CPL tool is a coding tool for improved coding of signals with more than one channel. The tool operates in one of the following configurations: • Advanced coupling in the channel pair element For improved stereo coding. This configuration is described in clause 5.7.7.5. • Advanced coupling in the multichannel elem... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.2 Parameter band to QMF subband mapping | Advanced Coupling parameters are transmitted per parameter band. Parameter bands are groupings of QMF subbands and have lower frequency resolution than the QMF subbands. The mapping of parameter bands to QMF subbands is shown below in table 197. The number of parameter bands, 7, 9, 12, or 15, is given by the variable a... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.3 Interpolation | Decoded and dequantized advanced coupling parameters or products of these (following advanced coupling parameters) carried in the bitstream are time-interpolated to calculate values that are applied to the input of the decorrelator and to the ducked output of the decorrelator. Parameter sets are transmitted either once... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.4 Decorrelator and transient ducker | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.4.1 Introduction | The decorrelation filters consist of a number of all-pass Infinite Impulse Response (IIR) filter sections preceded by a constant frequency-dependent delay. The frequency axis of the QMF time/frequency matrix is divided into three different regions as shown in table 198, which correspond to the QMF split frequencies giv... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.4.2 Decorrelator IIR filtering | The decorrelator filters for the different frequency regions given in table 198 are implemented by a delay and a subsequent IIR filter. Pseudocode 111 inputSignalModification(x) { for (sb = 0; sb < num_qmf_subbands; sb++) { for (ts = 0; ts < num_qmf_timeslots; ts++) { b[0] = a[filterLength(sb)]; y[ts][sb] = b[0]*x[ts-d... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.4.3 Transient ducker | To be able to handle transients and other fast time-envelopes, the output of the decorrelator all-pass filter has to be attenuated and this is performed according to the following pseudocode. Pseudocode 112 ALPHA = 0.76592833836465; ALPHA_SMOOTH = 0.25; GAMMA = 1.5; EPSILON = 1.0e-9; acpl_max_num_param_bands = 15; for ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.5 Advanced coupling in the channel pair element | For advanced coupling in the channel pair element, two input channels are present. Their elements are addressed as follows: • Left input channel x 0(ts,sb) ∈ Qin ACPL,L • Right input channel x 1(ts,sb) ∈ Qin ACPL,R The output channels are addressed as: • Left output channel z 0(ts,sb) ∈ Qout ACPL,L • Right output chann... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.6 Advanced coupling in the multichannel element | 5.7.7.6.1 5_X_codec_mode {ASPX_ACPL_1, ASPX_ACPL_2} For the configuration used when 5_X_codec_mode = ASPX_ACPL_1 or ASPX_ACPL_2, two parallel Advanced Coupling modules are used. Five input channels are present. Their elements are addressed as: • Left input channel x0(ts,sb) ∈ QinACPL,L • Right input channel x1(ts,sb)... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.6.2 5_X_codec_mode = ASPX_ACPL_3 | For the configuration used when 5_X_codec_mode = ASPX_ACPL_3, three parallel advanced coupling modules are used. Five input channels are present. Their elements are addressed as: • Left input channel x0(ts,sb) ∈ QinACPL,L • Right input channel x1(ts,sb) ∈ QinACPL,R • Centre input channel x2(ts,sb) ∈ QinACPL,C • Left su... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.7.7 Differential decoding and dequantization | To get the quantized values acpl_<SET>_q from the Huffman decoded values acpl_<SET>, where <SET> is an identifier for the A-CPL parameter set type and <SET> {alpha1, alpha2, beta1, beta2, beta3, gamma1, gamma2, gamma3, gamma4, gamma5, gamma6}, differential decoding shall be done as outlined in the following pseudocod... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8 Dialogue enhancement | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.1 Introduction | The dialogue enhancement tool is a tool to increase intelligibility of the dialogue in an audio scene encoded in AC-4. The underlying algorithm uses metadata encoded in the bitstream to boost the dialogue in the scene. This technology is especially beneficial for the hearing impaired. Dialogue enhancement supports enha... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.2 Processed channels | Each of the three decoding methods described in this clause operates, by default, on the QMF domain time-frequency samples of a subset of the three front channels in one AC-4 frame: mchannelk,n ∈QinDE,channel for 0 ≤ channel < 3, 0 ≤ k < num_qmf_subbands and 0 ≤ n < num_qmf_timeslots. As the dialogue may be limited t... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.3 Dequantization | The parameter index data shall be dequantized using the quantization vectors in table 209 and table 210. pi 'b,n = Q de_par/0, 0 ≤0 < de_nr_bands ETSI ETSI TS 103 190-1 V1.4.1 (2025-07) 249 Table 209: Dequantization vector for channel-independent enhancement mode Parameter index Parameter value 0 0 1 0,1 2 0,2... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.4 Parameter bands | The parametric enhancement methods rely on segmenting the num_qmf_subbands QMF subbands into de_nr_bands = 8 dialog enhancement parameter bands. The mapping of QMF subbands to parameter bands is described in clause 4.3.14.5.1. pi'(b, n) (as defined in clause 5.7.8.3) can be mapped to the corresponding QMF band by utili... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.5 Rendering | In modes with cross-channel dialogue enhancement the parametrically generated dialogue-enhancing signal shall be mixed into the audio scene according to the mixing parameters transmitted in the bitstream. When de_nr_channels indicates one involved channel, the dialogue enhancement signal is added to this channel only. ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.6 Interpolation | For a smooth transition between parameter sets, interpolation is applied between successive frames. Clause 5.7.8.7 through clause 5.7.8.9 describe the parameter-based processing for each frame f in terms of a matrix operation. Y = H<DE= ∙ m dc where m is the subset of channels that dialogue enhancement is applied to ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.7 Parametric channel independent enhancement | In this enhancement method, each channel i that has been indicated for processing is processed separately. A set of parameters pi is transmitted for each channel. Additionally, a maximum overall gain Gmax, defined in clause 4.3.14.3.2, limits the user-defined gain g to a range determined at the encoding stage. ETSI ETS... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.8 Parametric cross-channel enhancement | This method focuses on enhancing the dialogue using a combination of up to three input channels. The formula for reconstructing the dialogue is noted as follows: • One processed channel: Y=m+g×r×p×m • Two processed channels: Y y = m m + ∙r r ∙p p ∙m m • Three processed channels: Y y y = m m m + ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.8.9 Waveform-parametric hybrid | The hybrid dialogue enhancement modes complement the parametric enhancement parameters with transmission of up to three waveforms. The encoder also transmits a trade-off parameter, αc, derived from de_signal_contribution in clause 4.3.14.4.6, that indicates how the enhancement contributions should be divided over the t... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9 Dynamic range control tool | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.1 Introduction | This clause describes a DRC processing unit that is configured by the AC-4 bitstream. For an explanation of how DRC can be used in the scope of an audio codec, refer to ETSI TS 102 366 [i.1], clause 6.7.1.1. DRC is operated in the QMF domain and is executed before QMF synthesis. Data and control interfaces Input: • Qin... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.2 DRC Modes | Compressor operation is controlled by DRC modes. Modes are configured either through a compression curve or by direct gains to be applied, similarly to ETSI TS 102 366 [i.1], clause 6.7.1. AC-4 supports four default modes: • home theatre mode, • flat-panel TV mode, • portable mode - speakers, and • portable mode - head... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3 Decoding process | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3.1 Compression curves | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3.1.0 Introduction | When the DRC decoder mode utilizes a compression curve (drc_default_profile_flag == 1 or drc_compression_curve_flag == 1), the DRC processor, as depicted in figure 8a, comprises the following functions: • Level computation This function calculates the current signal level. • Gain computation This function maps the curr... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3.1.1 Level computation | There are several methods of level computation; an implementation may, for example, employ a level calculation as described in Recommendation ITU-R BS.1770-3 [i.4], or as a frequency weighted Root Mean Square (RMS) computation. The level shall be expressed relative to the dialnorm. The result of the level calculation i... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3.1.2 Gain computation | The gain computation stage maps the relative level Lk,n into a gain gk,n through the compression curve. The compression curve consists of seven sections. The total number of sections Ntotal can be derived from the number of boost sections Nboost and the number of cut sections Ncut as Ntotal = Nboost + Ncut + 3 The gain... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3.2 Directly transmitted DRC gains | For profiles where DRC gains are transmitted (drc_compression_curve_flag == 0), there are four configurations supported for transmitting gain values in the AC-4 bitstream: • single wideband gain for all channels, • wideband gain for each channel group, • multiband for each channel group, 2 bands, and • multiband for ea... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.3.3 Application of gain values | As a last step, the DRC tool shall map the dialnorm level to the output level Lout, and apply the gains that have either been calculated per clause 5.7.9.3.1.2 or transmitted as per clause 5.7.9.3.2. Qoutch,, = 2 LoutLin × gk,n,ch × Qin1ch,, |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 5.7.9.4 Transcoding to a AC-3 or E-AC-3 format | When an AC-4 bitstream is transcoded into a format such as AC-3 or E-AC-3, no DRC processing should be applied. Instead, the encoder or transcoder should be configured with the compression curve indicated by the field drc_eac3_transcode_curve in table 160. ETSI ETSI TS 103 190-1 V1.4.1 (2025-07) 259 |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6 Decoding the AC-4 bitstream | 6.1 Introduction Clause 4 specifies the details of the AC-4 bitstream syntax and clause 5 specifies the algorithmic details of the tools used within AC-4. This clause describes the complete AC-4 decoding process and by this makes a connection between the bitstream elements and the AC-4 tools. Figure 9 shows a flow diag... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.1 Input bitstream | The input bitstream typically comes from a transmission or storage system. To decode an AC-4 bitstream, all raw_ac4_frame elements shall be extracted according to the used transport format and presented to the AC-4 decoder. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.2 Structure of the bitstream | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.2.0 Overview | An AC-4 bitstream is roughly structured in several containers, namely: • Raw AC-4 frame The actual codec frame, consisting of a table of contents plus several byte aligned substreams. • Table of contents Signalling information on how to treat the content of the substream container of the AC-4 frame. • Substream Encoded... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.2.1 Raw AC-4 frame | Each raw AC-4 frame contains a Table Of Contents (TOC) and at least one byte-aligned substream. The TOC can be considered as the bitstream inventory where all information that is important for the overlaying system resides. Substreams are decodable units that represent a specific channel configuration (mono, stereo, 3.... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.2.2 Table of contents | The table of contents contains the TOC data and at least one presentation as well as a substream index table. As the bitstream inventory the table of contents provides information about and access to individual substreams. The TOC data contains stream specific metadata, such as: • bitstream version, which signals if th... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.2.3 Presentation information | Typically, a presentation in AC-4 describes a set of substreams that is to be decoded and presented simultaneously. Mixing of decoded content only occurs within one presentation. That means only one presentation has to be decoded at a time. On the other hand, one substream can be part of several presentations. Examples... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.2.4 Substream | A substream contains the actual audio data, as well as the corresponding metadata. Additionally, the size of the audio data is given in the header of the substream, so that the metadata can be accessed without the need of parsing the audio data. The ac4_substream() element is specified in clause 4.3.4. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.3 Selecting and decoding a presentation | To successfully select and decode an appropriate presentation, a decoder system has to execute the following steps: 1) Create a list of presentations available in the bitstream. Initially, the number of presentations, n_presentations, has to be derived from the bitstream elements b_single_presentation and b_more_presen... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.4 Buffer model | Decoders implemented in accordance with the present document shall provide a minimum input buffer of size N bytes where N is set according to: N=v× Rstream Rframe Rstream and Rframe are the total bit rate and the frame_rate, respectively, and v= 6, RFrame ≤60 12, else . See table 211 for typical values of N. NOTE 1: VB... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.5 Decoding of a substream | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.5.0 Substream types | As described in clause 4.2.4, three different types of AC-4 substreams exist: • AC-4 substream; • AC-4 high sampling frequency extension substream; and • EMDF payloads substream. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.5.1 Decoding of an AC-4 substream | A decoder shall decode audio_data() and metadata() in the AC-4 substream. Both bitstream elements can be accessed independently of each other by using the variable audio_size. Decoding of the channel elements within audio_data() is described in clause 6.2.6.0 and the following clauses. The bitstream elements contained ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.5.2 Decoding of an AC-4 HSF extension substream | Decoding of the HSF substream can be achieved using the information derived from clause 5.4. Afterwards, the SAP tool and the IMDCT described in clause 5.3 and clause 5.5 need to be employed. No QMF domain processing is required. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.5.3 Decoding of an EMDF payloads substream | A decoder shall be able to skip an EMDF substream, and or may ignore its contents. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6 Spectral frontend decoding | |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6.0 audio_data element types | The channel element contained in the audio_data element of an AC-4 substream can be one of four types, signalled by the channel_mode, as indicated in clause 4.2.6. Within the channel elements, audio tracks are encoded as sf_data elements. These elements are associated with an sf_info element, as well as with a spectral... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6.1 Mono decoding | If channel_mode is mono, the bitstream contains only one track stored as an sf_data element within a mono_data element. The decoding of the sf_data element depends on the spectral frontend type and on the sf_info element and is done in the following steps. 1) Parsing of sf_info which includes transform and psy informat... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6.2 Stereo decoding | If channel_mode is stereo, the bitstream contains up to two audio tracks stored as sf_data elements within a channel_pair_element. Similarly to the mono decoding, the decoding of the sf_data element is done in steps described in clause 6.2.6.1. If the b_enable_mdct_stereo_proc is true, a common sf_info element is prese... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6.3 Multichannel audio decoding | Similarly to stereo decoding, first, all the sf_data elements are decoded, followed by routing the decoded tracks into the stereo and multichannel processing tool. This tool returns dedicated audio channels. The spectral lines of these channels shall subsequently be processed by the IMDCT tool. If channel mode is 5.1 o... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6.4 Audio spectral front end | The asf_section_data and asf_scalefac_data elements of the sf_data element are parsed and helper elements are set up. See clause 4.3.6.3 and clause 4.3.6.5. The asf_spectral_data is Huffman decoded using the helper elements. See clause 5.1.2 for details of the Huffman decoding process for the quantized spectral lines. ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.6.5 Speech spectral frontend (SSF) | The sf_data element contains just the ssf_data element if the spec_frontend value is '1'. Either one or two ssf_granule elements are present in the ssf_data element. Each ssf_granule element is decoded into a vector of spectral lines as specified in clause 5.2. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.7 (IMDCT) and window overlap/add | The decoding steps described in the preceding clauses result in a set of frame_length (reordered) spectral lines for each spectral frontend channel. The inverse MDCT transform converts the blocks of spectral lines into blocks of time samples. ETSI ETSI TS 103 190-1 V1.4.1 (2025-07) 266 The individual blocks of time sam... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.8 QMF analysis | The time domain samples of each channel are transformed individually into the QMF domain by the use of a QMF analysis filter as described in clause 5.7.3. Depending on the frame_length, 32, 30, 24, 15, 12 or 6 QMF slots are the result of the QMF analysis. Each QMF slot is the result of one QMF filtering step with num_q... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.9 Companding | The companding tool mitigates pre- and post-echo artefacts. The encoder applies attenuation to signal parts with higher energy, and gain to signal parts with lower energy. The decoder applies the inverse process, which effectively shapes the coding noise by the signal energy. Whether or not a channel is processed by th... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.10 Advanced spectral extension - A-SPX | Spectral extension includes translating the spectral content of suitable low-frequency regions of the signal to higher frequencies and adjusting them such that they perceptually match the high-frequency content that is present in the original signal. Waveform coding may be employed in addition, either when critical sig... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.11 Advanced coupling - A-CPL | Advanced coupling reconstructs multichannel content out of a lower number of input channels, using encoder controlled side information. As a special case, a stereo signal can be generated from a mono downmix. An AC-4 decoder shall process the QMF domain channels with the A-CPL after the A-SPX processing has been applie... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.12 Dialogue enhancement | Dialogue enhancement control metadata can be specified in the bitstream. When the decoder is configured to enable dialogue enhancement, the dialogue enhancement tool uses this metadata to apply a gain on the dialogue in the signal according to a system-defined gain. Table 215 lists the input channels for the dialogue e... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.13 Dynamic range control | Dynamic range control metadata can be specified in the bitstream. The DRC tool uses this metadata to apply gains to the channels in the QMF domain in order to adjust the dynamic range as specified in clause 5.7.9. The DRC tool takes two inputs per channel: the audio signal which is output by a preceding tool (usually d... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.14 QMF synthesis | A QMF synthesis filter shall transform each audio channel from the QMF domain back to the time domain as specified in clause 5.7.4. |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.15 Sampling rate converter | For all values of the frame_rate_index, the decoder may be operated at external sampling frequencies of 48 kHz, 96 kHz, or 192 kHz. For frame_rate_index = 13, 44,1 kHz is also a permissible external sampling frequency. To adjust the sampling frequency to the external sampling frequency, a sampling rate converter shall ... |
79462c15429e4b02e874d57a1faa88a5 | 103 190-1 | 6.2.16 Mixing substream outputs |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.