Line | Count | Source |
1 | | /*********************************************************************** |
2 | | Copyright (c) 2006-2011, Skype Limited. All rights reserved. |
3 | | Redistribution and use in source and binary forms, with or without |
4 | | modification, are permitted provided that the following conditions |
5 | | are met: |
6 | | - Redistributions of source code must retain the above copyright notice, |
7 | | this list of conditions and the following disclaimer. |
8 | | - Redistributions in binary form must reproduce the above copyright |
9 | | notice, this list of conditions and the following disclaimer in the |
10 | | documentation and/or other materials provided with the distribution. |
11 | | - Neither the name of Internet Society, IETF or IETF Trust, nor the |
12 | | names of specific contributors, may be used to endorse or promote |
13 | | products derived from this software without specific prior written |
14 | | permission. |
15 | | THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
16 | | AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
17 | | IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
18 | | ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE |
19 | | LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
20 | | CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
21 | | SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
22 | | INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
23 | | CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
24 | | ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
25 | | POSSIBILITY OF SUCH DAMAGE. |
26 | | ***********************************************************************/ |
27 | | |
28 | | #ifdef HAVE_CONFIG_H |
29 | | #include "config.h" |
30 | | #endif |
31 | | |
32 | | #include "main.h" |
33 | | #include "stack_alloc.h" |
34 | | #include "NSQ.h" |
35 | | |
36 | | |
37 | | static OPUS_INLINE void silk_nsq_scale_states( |
38 | | const silk_encoder_state *psEncC, /* I Encoder State */ |
39 | | silk_nsq_state *NSQ, /* I/O NSQ state */ |
40 | | const opus_int16 x16[], /* I input */ |
41 | | opus_int32 x_sc_Q10[], /* O input scaled with 1/Gain */ |
42 | | const opus_int16 sLTP[], /* I re-whitened LTP state in Q0 */ |
43 | | opus_int32 sLTP_Q15[], /* O LTP state matching scaled input */ |
44 | | opus_int subfr, /* I subframe number */ |
45 | | const opus_int LTP_scale_Q14, /* I */ |
46 | | const opus_int32 Gains_Q16[ MAX_NB_SUBFR ], /* I */ |
47 | | const opus_int pitchL[ MAX_NB_SUBFR ], /* I Pitch lag */ |
48 | | const opus_int signal_type /* I Signal type */ |
49 | | ); |
50 | | |
51 | | #if !defined(OPUS_X86_MAY_HAVE_SSE4_1) |
52 | | static OPUS_INLINE void silk_noise_shape_quantizer( |
53 | | silk_nsq_state *NSQ, /* I/O NSQ state */ |
54 | | opus_int signalType, /* I Signal type */ |
55 | | const opus_int32 x_sc_Q10[], /* I */ |
56 | | opus_int8 pulses[], /* O */ |
57 | | opus_int16 xq[], /* O */ |
58 | | opus_int32 sLTP_Q15[], /* I/O LTP state */ |
59 | | const opus_int16 a_Q12[], /* I Short term prediction coefs */ |
60 | | const opus_int16 b_Q14[], /* I Long term prediction coefs */ |
61 | | const opus_int16 AR_shp_Q13[], /* I Noise shaping AR coefs */ |
62 | | opus_int lag, /* I Pitch lag */ |
63 | | opus_int32 HarmShapeFIRPacked_Q14, /* I */ |
64 | | opus_int Tilt_Q14, /* I Spectral tilt */ |
65 | | opus_int32 LF_shp_Q14, /* I */ |
66 | | opus_int32 Gain_Q16, /* I */ |
67 | | opus_int Lambda_Q10, /* I */ |
68 | | opus_int offset_Q10, /* I */ |
69 | | opus_int length, /* I Input length */ |
70 | | opus_int shapingLPCOrder, /* I Noise shaping AR filter order */ |
71 | | opus_int predictLPCOrder, /* I Prediction filter order */ |
72 | | int arch /* I Architecture */ |
73 | | ); |
74 | | #endif |
75 | | |
76 | | void silk_NSQ_c |
77 | | ( |
78 | | const silk_encoder_state *psEncC, /* I Encoder State */ |
79 | | silk_nsq_state *NSQ, /* I/O NSQ state */ |
80 | | SideInfoIndices *psIndices, /* I/O Quantization Indices */ |
81 | | const opus_int16 x16[], /* I Input */ |
82 | | opus_int8 pulses[], /* O Quantized pulse signal */ |
83 | | const opus_int16 *PredCoef_Q12, /* I Short term prediction coefs */ |
84 | | const opus_int16 LTPCoef_Q14[ LTP_ORDER * MAX_NB_SUBFR ], /* I Long term prediction coefs */ |
85 | | const opus_int16 AR_Q13[ MAX_NB_SUBFR * MAX_SHAPE_LPC_ORDER ], /* I Noise shaping coefs */ |
86 | | const opus_int HarmShapeGain_Q14[ MAX_NB_SUBFR ], /* I Long term shaping coefs */ |
87 | | const opus_int Tilt_Q14[ MAX_NB_SUBFR ], /* I Spectral tilt */ |
88 | | const opus_int32 LF_shp_Q14[ MAX_NB_SUBFR ], /* I Low frequency shaping coefs */ |
89 | | const opus_int32 Gains_Q16[ MAX_NB_SUBFR ], /* I Quantization step sizes */ |
90 | | const opus_int pitchL[ MAX_NB_SUBFR ], /* I Pitch lags */ |
91 | | const opus_int Lambda_Q10, /* I Rate/distortion tradeoff */ |
92 | | const opus_int LTP_scale_Q14 /* I LTP state scaling */ |
93 | | ) |
94 | 184k | { |
95 | 184k | opus_int k, lag, start_idx, LSF_interpolation_flag; |
96 | 184k | const opus_int16 *A_Q12, *B_Q14, *AR_shp_Q13; |
97 | 184k | opus_int16 *pxq; |
98 | 184k | VARDECL( opus_int32, sLTP_Q15 ); |
99 | 184k | VARDECL( opus_int16, sLTP ); |
100 | 184k | opus_int32 HarmShapeFIRPacked_Q14; |
101 | 184k | opus_int offset_Q10; |
102 | 184k | VARDECL( opus_int32, x_sc_Q10 ); |
103 | 184k | SAVE_STACK; |
104 | | |
105 | 184k | NSQ->rand_seed = psIndices->Seed; |
106 | | |
107 | | /* Set unvoiced lag to the previous one, overwrite later for voiced */ |
108 | 184k | lag = NSQ->lagPrev; |
109 | | |
110 | 184k | silk_assert( NSQ->prev_gain_Q16 != 0 ); |
111 | | |
112 | 184k | offset_Q10 = silk_Quantization_Offsets_Q10[ psIndices->signalType >> 1 ][ psIndices->quantOffsetType ]; |
113 | | |
114 | 184k | if( psIndices->NLSFInterpCoef_Q2 == 4 ) { |
115 | 184k | LSF_interpolation_flag = 0; |
116 | 184k | } else { |
117 | 0 | LSF_interpolation_flag = 1; |
118 | 0 | } |
119 | | |
120 | 184k | ALLOC( sLTP_Q15, psEncC->ltp_mem_length + psEncC->frame_length, opus_int32 ); |
121 | 184k | ALLOC( sLTP, psEncC->ltp_mem_length + psEncC->frame_length, opus_int16 ); |
122 | 184k | ALLOC( x_sc_Q10, psEncC->subfr_length, opus_int32 ); |
123 | | /* Set up pointers to start of sub frame */ |
124 | 184k | NSQ->sLTP_shp_buf_idx = psEncC->ltp_mem_length; |
125 | 184k | NSQ->sLTP_buf_idx = psEncC->ltp_mem_length; |
126 | 184k | pxq = &NSQ->xq[ psEncC->ltp_mem_length ]; |
127 | 772k | for( k = 0; k < psEncC->nb_subfr; k++ ) { |
128 | 587k | A_Q12 = &PredCoef_Q12[ (( k >> 1 ) | ( 1 - LSF_interpolation_flag )) * MAX_LPC_ORDER ]; |
129 | 587k | B_Q14 = <PCoef_Q14[ k * LTP_ORDER ]; |
130 | 587k | AR_shp_Q13 = &AR_Q13[ k * MAX_SHAPE_LPC_ORDER ]; |
131 | | |
132 | | /* Noise shape parameters */ |
133 | 587k | silk_assert( HarmShapeGain_Q14[ k ] >= 0 ); |
134 | 587k | HarmShapeFIRPacked_Q14 = silk_RSHIFT( HarmShapeGain_Q14[ k ], 2 ); |
135 | 587k | HarmShapeFIRPacked_Q14 |= silk_LSHIFT( (opus_int32)silk_RSHIFT( HarmShapeGain_Q14[ k ], 1 ), 16 ); |
136 | | |
137 | 587k | NSQ->rewhite_flag = 0; |
138 | 587k | if( psIndices->signalType == TYPE_VOICED ) { |
139 | | /* Voiced */ |
140 | 127k | lag = pitchL[ k ]; |
141 | | |
142 | | /* Re-whitening */ |
143 | 127k | if( ( k & ( 3 - silk_LSHIFT( LSF_interpolation_flag, 1 ) ) ) == 0 ) { |
144 | | /* Rewhiten with new A coefs */ |
145 | 31.9k | start_idx = psEncC->ltp_mem_length - lag - psEncC->predictLPCOrder - LTP_ORDER / 2; |
146 | 31.9k | celt_assert( start_idx > 0 ); |
147 | | |
148 | 31.9k | silk_LPC_analysis_filter( &sLTP[ start_idx ], &NSQ->xq[ start_idx + k * psEncC->subfr_length ], |
149 | 31.9k | A_Q12, psEncC->ltp_mem_length - start_idx, psEncC->predictLPCOrder, psEncC->arch ); |
150 | | |
151 | 31.9k | NSQ->rewhite_flag = 1; |
152 | 31.9k | NSQ->sLTP_buf_idx = psEncC->ltp_mem_length; |
153 | 31.9k | } |
154 | 127k | } |
155 | | |
156 | 587k | silk_nsq_scale_states( psEncC, NSQ, x16, x_sc_Q10, sLTP, sLTP_Q15, k, LTP_scale_Q14, Gains_Q16, pitchL, psIndices->signalType ); |
157 | | |
158 | 587k | silk_noise_shape_quantizer( NSQ, psIndices->signalType, x_sc_Q10, pulses, pxq, sLTP_Q15, A_Q12, B_Q14, |
159 | 587k | AR_shp_Q13, lag, HarmShapeFIRPacked_Q14, Tilt_Q14[ k ], LF_shp_Q14[ k ], Gains_Q16[ k ], Lambda_Q10, |
160 | 587k | offset_Q10, psEncC->subfr_length, psEncC->shapingLPCOrder, psEncC->predictLPCOrder, psEncC->arch ); |
161 | | |
162 | 587k | x16 += psEncC->subfr_length; |
163 | 587k | pulses += psEncC->subfr_length; |
164 | 587k | pxq += psEncC->subfr_length; |
165 | 587k | } |
166 | | |
167 | | /* Update lagPrev for next frame */ |
168 | 184k | NSQ->lagPrev = pitchL[ psEncC->nb_subfr - 1 ]; |
169 | | |
170 | | /* Save quantized speech and noise shaping signals */ |
171 | 184k | silk_memmove( NSQ->xq, &NSQ->xq[ psEncC->frame_length ], psEncC->ltp_mem_length * sizeof( opus_int16 ) ); |
172 | 184k | silk_memmove( NSQ->sLTP_shp_Q14, &NSQ->sLTP_shp_Q14[ psEncC->frame_length ], psEncC->ltp_mem_length * sizeof( opus_int32 ) ); |
173 | 184k | RESTORE_STACK; |
174 | 184k | } |
175 | | |
176 | | /******************************/ |
177 | | /* silk_noise_shape_quantizer */ |
178 | | /******************************/ |
179 | | |
180 | | #if !defined(OPUS_X86_MAY_HAVE_SSE4_1) |
181 | | static OPUS_INLINE |
182 | | #endif |
183 | | void silk_noise_shape_quantizer( |
184 | | silk_nsq_state *NSQ, /* I/O NSQ state */ |
185 | | opus_int signalType, /* I Signal type */ |
186 | | const opus_int32 x_sc_Q10[], /* I */ |
187 | | opus_int8 pulses[], /* O */ |
188 | | opus_int16 xq[], /* O */ |
189 | | opus_int32 sLTP_Q15[], /* I/O LTP state */ |
190 | | const opus_int16 a_Q12[], /* I Short term prediction coefs */ |
191 | | const opus_int16 b_Q14[], /* I Long term prediction coefs */ |
192 | | const opus_int16 AR_shp_Q13[], /* I Noise shaping AR coefs */ |
193 | | opus_int lag, /* I Pitch lag */ |
194 | | opus_int32 HarmShapeFIRPacked_Q14, /* I */ |
195 | | opus_int Tilt_Q14, /* I Spectral tilt */ |
196 | | opus_int32 LF_shp_Q14, /* I */ |
197 | | opus_int32 Gain_Q16, /* I */ |
198 | | opus_int Lambda_Q10, /* I */ |
199 | | opus_int offset_Q10, /* I */ |
200 | | opus_int length, /* I Input length */ |
201 | | opus_int shapingLPCOrder, /* I Noise shaping AR filter order */ |
202 | | opus_int predictLPCOrder, /* I Prediction filter order */ |
203 | | int arch /* I Architecture */ |
204 | | ) |
205 | 1.76M | { |
206 | 1.76M | opus_int i; |
207 | 1.76M | opus_int32 LTP_pred_Q13, LPC_pred_Q10, n_AR_Q12, n_LTP_Q13; |
208 | 1.76M | opus_int32 n_LF_Q12, r_Q10, rr_Q10, q1_Q0, q1_Q10, q2_Q10, rd1_Q20, rd2_Q20; |
209 | 1.76M | opus_int32 exc_Q14, LPC_exc_Q14, xq_Q14, Gain_Q10; |
210 | 1.76M | opus_int32 tmp1, tmp2, sLF_AR_shp_Q14; |
211 | 1.76M | opus_int32 *psLPC_Q14, *shp_lag_ptr, *pred_lag_ptr; |
212 | | #ifdef silk_short_prediction_create_arch_coef |
213 | | opus_int32 a_Q12_arch[MAX_LPC_ORDER]; |
214 | | #endif |
215 | | |
216 | 1.76M | shp_lag_ptr = &NSQ->sLTP_shp_Q14[ NSQ->sLTP_shp_buf_idx - lag + HARM_SHAPE_FIR_TAPS / 2 ]; |
217 | 1.76M | pred_lag_ptr = &sLTP_Q15[ NSQ->sLTP_buf_idx - lag + LTP_ORDER / 2 ]; |
218 | 1.76M | Gain_Q10 = silk_RSHIFT( Gain_Q16, 6 ); |
219 | | |
220 | | /* Set up short term AR state */ |
221 | 1.76M | psLPC_Q14 = &NSQ->sLPC_Q14[ NSQ_LPC_BUF_LENGTH - 1 ]; |
222 | | |
223 | | #ifdef silk_short_prediction_create_arch_coef |
224 | | silk_short_prediction_create_arch_coef(a_Q12_arch, a_Q12, predictLPCOrder); |
225 | | #endif |
226 | | |
227 | 79.9M | for( i = 0; i < length; i++ ) { |
228 | | /* Generate dither */ |
229 | 78.1M | NSQ->rand_seed = silk_RAND( NSQ->rand_seed ); |
230 | | |
231 | | /* Short-term prediction */ |
232 | 78.1M | LPC_pred_Q10 = silk_noise_shape_quantizer_short_prediction(psLPC_Q14, a_Q12, a_Q12_arch, predictLPCOrder, arch); |
233 | | |
234 | | /* Long-term prediction */ |
235 | 78.1M | if( signalType == TYPE_VOICED ) { |
236 | | /* Unrolled loop */ |
237 | | /* Avoids introducing a bias because silk_SMLAWB() always rounds to -inf */ |
238 | 16.9M | LTP_pred_Q13 = 2; |
239 | 16.9M | LTP_pred_Q13 = silk_SMLAWB( LTP_pred_Q13, pred_lag_ptr[ 0 ], b_Q14[ 0 ] ); |
240 | 16.9M | LTP_pred_Q13 = silk_SMLAWB( LTP_pred_Q13, pred_lag_ptr[ -1 ], b_Q14[ 1 ] ); |
241 | 16.9M | LTP_pred_Q13 = silk_SMLAWB( LTP_pred_Q13, pred_lag_ptr[ -2 ], b_Q14[ 2 ] ); |
242 | 16.9M | LTP_pred_Q13 = silk_SMLAWB( LTP_pred_Q13, pred_lag_ptr[ -3 ], b_Q14[ 3 ] ); |
243 | 16.9M | LTP_pred_Q13 = silk_SMLAWB( LTP_pred_Q13, pred_lag_ptr[ -4 ], b_Q14[ 4 ] ); |
244 | 16.9M | pred_lag_ptr++; |
245 | 61.2M | } else { |
246 | 61.2M | LTP_pred_Q13 = 0; |
247 | 61.2M | } |
248 | | |
249 | | /* Noise shape feedback */ |
250 | 78.1M | celt_assert( ( shapingLPCOrder & 1 ) == 0 ); /* check that order is even */ |
251 | 78.1M | n_AR_Q12 = silk_NSQ_noise_shape_feedback_loop(&NSQ->sDiff_shp_Q14, NSQ->sAR2_Q14, AR_shp_Q13, shapingLPCOrder, arch); |
252 | | |
253 | 78.1M | n_AR_Q12 = silk_SMLAWB( n_AR_Q12, NSQ->sLF_AR_shp_Q14, Tilt_Q14 ); |
254 | | |
255 | 78.1M | n_LF_Q12 = silk_SMULWB( NSQ->sLTP_shp_Q14[ NSQ->sLTP_shp_buf_idx - 1 ], LF_shp_Q14 ); |
256 | 78.1M | n_LF_Q12 = silk_SMLAWT( n_LF_Q12, NSQ->sLF_AR_shp_Q14, LF_shp_Q14 ); |
257 | | |
258 | 78.1M | celt_assert( lag > 0 || signalType != TYPE_VOICED ); |
259 | | |
260 | | /* Combine prediction and noise shaping signals */ |
261 | 78.1M | tmp1 = silk_SUB32_ovflw( silk_LSHIFT32( LPC_pred_Q10, 2 ), n_AR_Q12 ); /* Q12 */ |
262 | 78.1M | tmp1 = silk_SUB32_ovflw( tmp1, n_LF_Q12 ); /* Q12 */ |
263 | 78.1M | if( lag > 0 ) { |
264 | | /* Symmetric, packed FIR coefficients */ |
265 | 54.4M | n_LTP_Q13 = silk_SMULWB( silk_ADD_SAT32( shp_lag_ptr[ 0 ], shp_lag_ptr[ -2 ] ), HarmShapeFIRPacked_Q14 ); |
266 | 54.4M | n_LTP_Q13 = silk_SMLAWT( n_LTP_Q13, shp_lag_ptr[ -1 ], HarmShapeFIRPacked_Q14 ); |
267 | 54.4M | n_LTP_Q13 = silk_LSHIFT( n_LTP_Q13, 1 ); |
268 | 54.4M | shp_lag_ptr++; |
269 | | |
270 | 54.4M | tmp2 = silk_SUB32( LTP_pred_Q13, n_LTP_Q13 ); /* Q13 */ |
271 | 54.4M | tmp1 = silk_ADD32_ovflw( tmp2, silk_LSHIFT32( tmp1, 1 ) ); /* Q13 */ |
272 | 54.4M | tmp1 = silk_RSHIFT_ROUND( tmp1, 3 ); /* Q10 */ |
273 | 54.4M | } else { |
274 | 23.7M | tmp1 = silk_RSHIFT_ROUND( tmp1, 2 ); /* Q10 */ |
275 | 23.7M | } |
276 | | |
277 | 78.1M | r_Q10 = silk_SUB32( x_sc_Q10[ i ], tmp1 ); /* residual error Q10 */ |
278 | | |
279 | | /* Flip sign depending on dither */ |
280 | 78.1M | if( NSQ->rand_seed < 0 ) { |
281 | 37.7M | r_Q10 = -r_Q10; |
282 | 37.7M | } |
283 | 78.1M | r_Q10 = silk_LIMIT_32( r_Q10, -(31 << 10), 30 << 10 ); |
284 | | |
285 | | /* Find two quantization level candidates and measure their rate-distortion */ |
286 | 78.1M | q1_Q10 = silk_SUB32( r_Q10, offset_Q10 ); |
287 | 78.1M | q1_Q0 = silk_RSHIFT( q1_Q10, 10 ); |
288 | 78.1M | if (Lambda_Q10 > 2048) { |
289 | | /* For aggressive RDO, the bias becomes more than one pulse. */ |
290 | 6.08M | int rdo_offset = Lambda_Q10/2 - 512; |
291 | 6.08M | if (q1_Q10 > rdo_offset) { |
292 | 776k | q1_Q0 = silk_RSHIFT( q1_Q10 - rdo_offset, 10 ); |
293 | 5.30M | } else if (q1_Q10 < -rdo_offset) { |
294 | 888k | q1_Q0 = silk_RSHIFT( q1_Q10 + rdo_offset, 10 ); |
295 | 4.42M | } else if (q1_Q10 < 0) { |
296 | 2.60M | q1_Q0 = -1; |
297 | 2.60M | } else { |
298 | 1.81M | q1_Q0 = 0; |
299 | 1.81M | } |
300 | 6.08M | } |
301 | 78.1M | if( q1_Q0 > 0 ) { |
302 | 5.66M | q1_Q10 = silk_SUB32( silk_LSHIFT( q1_Q0, 10 ), QUANT_LEVEL_ADJUST_Q10 ); |
303 | 5.66M | q1_Q10 = silk_ADD32( q1_Q10, offset_Q10 ); |
304 | 5.66M | q2_Q10 = silk_ADD32( q1_Q10, 1024 ); |
305 | 5.66M | rd1_Q20 = silk_SMULBB( q1_Q10, Lambda_Q10 ); |
306 | 5.66M | rd2_Q20 = silk_SMULBB( q2_Q10, Lambda_Q10 ); |
307 | 72.5M | } else if( q1_Q0 == 0 ) { |
308 | 19.5M | q1_Q10 = offset_Q10; |
309 | 19.5M | q2_Q10 = silk_ADD32( q1_Q10, 1024 - QUANT_LEVEL_ADJUST_Q10 ); |
310 | 19.5M | rd1_Q20 = silk_SMULBB( q1_Q10, Lambda_Q10 ); |
311 | 19.5M | rd2_Q20 = silk_SMULBB( q2_Q10, Lambda_Q10 ); |
312 | 52.9M | } else if( q1_Q0 == -1 ) { |
313 | 46.3M | q2_Q10 = offset_Q10; |
314 | 46.3M | q1_Q10 = silk_SUB32( q2_Q10, 1024 - QUANT_LEVEL_ADJUST_Q10 ); |
315 | 46.3M | rd1_Q20 = silk_SMULBB( -q1_Q10, Lambda_Q10 ); |
316 | 46.3M | rd2_Q20 = silk_SMULBB( q2_Q10, Lambda_Q10 ); |
317 | 46.3M | } else { /* Q1_Q0 < -1 */ |
318 | 6.55M | q1_Q10 = silk_ADD32( silk_LSHIFT( q1_Q0, 10 ), QUANT_LEVEL_ADJUST_Q10 ); |
319 | 6.55M | q1_Q10 = silk_ADD32( q1_Q10, offset_Q10 ); |
320 | 6.55M | q2_Q10 = silk_ADD32( q1_Q10, 1024 ); |
321 | 6.55M | rd1_Q20 = silk_SMULBB( -q1_Q10, Lambda_Q10 ); |
322 | 6.55M | rd2_Q20 = silk_SMULBB( -q2_Q10, Lambda_Q10 ); |
323 | 6.55M | } |
324 | 78.1M | rr_Q10 = silk_SUB32( r_Q10, q1_Q10 ); |
325 | 78.1M | rd1_Q20 = silk_SMLABB( rd1_Q20, rr_Q10, rr_Q10 ); |
326 | 78.1M | rr_Q10 = silk_SUB32( r_Q10, q2_Q10 ); |
327 | 78.1M | rd2_Q20 = silk_SMLABB( rd2_Q20, rr_Q10, rr_Q10 ); |
328 | | |
329 | 78.1M | if( rd2_Q20 < rd1_Q20 ) { |
330 | 51.9M | q1_Q10 = q2_Q10; |
331 | 51.9M | } |
332 | | |
333 | 78.1M | pulses[ i ] = (opus_int8)silk_RSHIFT_ROUND( q1_Q10, 10 ); |
334 | | |
335 | | /* Excitation */ |
336 | 78.1M | exc_Q14 = silk_LSHIFT( q1_Q10, 4 ); |
337 | 78.1M | if ( NSQ->rand_seed < 0 ) { |
338 | 37.7M | exc_Q14 = -exc_Q14; |
339 | 37.7M | } |
340 | | |
341 | | /* Add predictions */ |
342 | 78.1M | LPC_exc_Q14 = silk_ADD_LSHIFT32( exc_Q14, LTP_pred_Q13, 1 ); |
343 | 78.1M | xq_Q14 = silk_ADD32_ovflw( LPC_exc_Q14, silk_LSHIFT32( LPC_pred_Q10, 4 ) ); |
344 | | |
345 | | /* Scale XQ back to normal level before saving */ |
346 | 78.1M | xq[ i ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( silk_SMULWW( xq_Q14, Gain_Q10 ), 8 ) ); |
347 | | |
348 | | /* Update states */ |
349 | 78.1M | psLPC_Q14++; |
350 | 78.1M | *psLPC_Q14 = xq_Q14; |
351 | 78.1M | NSQ->sDiff_shp_Q14 = silk_SUB32_ovflw( xq_Q14, silk_LSHIFT32( x_sc_Q10[ i ], 4 ) ); |
352 | 78.1M | sLF_AR_shp_Q14 = silk_SUB32_ovflw( NSQ->sDiff_shp_Q14, silk_LSHIFT32( n_AR_Q12, 2 ) ); |
353 | 78.1M | NSQ->sLF_AR_shp_Q14 = sLF_AR_shp_Q14; |
354 | | |
355 | 78.1M | NSQ->sLTP_shp_Q14[ NSQ->sLTP_shp_buf_idx ] = silk_SUB32_ovflw(sLF_AR_shp_Q14, silk_LSHIFT32(n_LF_Q12, 2)); |
356 | 78.1M | sLTP_Q15[ NSQ->sLTP_buf_idx ] = silk_LSHIFT( LPC_exc_Q14, 1 ); |
357 | 78.1M | NSQ->sLTP_shp_buf_idx++; |
358 | 78.1M | NSQ->sLTP_buf_idx++; |
359 | | |
360 | | /* Make dither dependent on quantized signal */ |
361 | 78.1M | NSQ->rand_seed = silk_ADD32_ovflw( NSQ->rand_seed, pulses[ i ] ); |
362 | 78.1M | } |
363 | | |
364 | | /* Update LPC synth buffer */ |
365 | 1.76M | silk_memcpy( NSQ->sLPC_Q14, &NSQ->sLPC_Q14[ length ], NSQ_LPC_BUF_LENGTH * sizeof( opus_int32 ) ); |
366 | 1.76M | } |
367 | | |
368 | | static OPUS_INLINE void silk_nsq_scale_states( |
369 | | const silk_encoder_state *psEncC, /* I Encoder State */ |
370 | | silk_nsq_state *NSQ, /* I/O NSQ state */ |
371 | | const opus_int16 x16[], /* I input */ |
372 | | opus_int32 x_sc_Q10[], /* O input scaled with 1/Gain */ |
373 | | const opus_int16 sLTP[], /* I re-whitened LTP state in Q0 */ |
374 | | opus_int32 sLTP_Q15[], /* O LTP state matching scaled input */ |
375 | | opus_int subfr, /* I subframe number */ |
376 | | const opus_int LTP_scale_Q14, /* I */ |
377 | | const opus_int32 Gains_Q16[ MAX_NB_SUBFR ], /* I */ |
378 | | const opus_int pitchL[ MAX_NB_SUBFR ], /* I Pitch lag */ |
379 | | const opus_int signal_type /* I Signal type */ |
380 | | ) |
381 | 587k | { |
382 | 587k | opus_int i, lag; |
383 | 587k | opus_int32 gain_adj_Q16, inv_gain_Q31, inv_gain_Q26; |
384 | | |
385 | 587k | lag = pitchL[ subfr ]; |
386 | 587k | inv_gain_Q31 = silk_INVERSE32_varQ( silk_max( Gains_Q16[ subfr ], 1 ), 47 ); |
387 | 587k | silk_assert( inv_gain_Q31 != 0 ); |
388 | | |
389 | | /* Scale input */ |
390 | 587k | inv_gain_Q26 = silk_RSHIFT_ROUND( inv_gain_Q31, 5 ); |
391 | 26.2M | for( i = 0; i < psEncC->subfr_length; i++ ) { |
392 | 25.6M | x_sc_Q10[ i ] = silk_SMULWW( x16[ i ], inv_gain_Q26 ); |
393 | 25.6M | } |
394 | | |
395 | | /* After rewhitening the LTP state is un-scaled, so scale with inv_gain_Q16 */ |
396 | 587k | if( NSQ->rewhite_flag ) { |
397 | 31.9k | if( subfr == 0 ) { |
398 | | /* Do LTP downscaling */ |
399 | 31.9k | inv_gain_Q31 = silk_LSHIFT( silk_SMULWB( inv_gain_Q31, LTP_scale_Q14 ), 2 ); |
400 | 31.9k | } |
401 | 1.34M | for( i = NSQ->sLTP_buf_idx - lag - LTP_ORDER / 2; i < NSQ->sLTP_buf_idx; i++ ) { |
402 | 1.31M | silk_assert( i < MAX_FRAME_LENGTH ); |
403 | 1.31M | sLTP_Q15[ i ] = silk_SMULWB( inv_gain_Q31, sLTP[ i ] ); |
404 | 1.31M | } |
405 | 31.9k | } |
406 | | |
407 | | /* Adjust for changing gain */ |
408 | 587k | if( Gains_Q16[ subfr ] != NSQ->prev_gain_Q16 ) { |
409 | 509k | gain_adj_Q16 = silk_DIV32_varQ( NSQ->prev_gain_Q16, Gains_Q16[ subfr ], 16 ); |
410 | | |
411 | | /* Scale long-term shaping state */ |
412 | 89.8M | for( i = NSQ->sLTP_shp_buf_idx - psEncC->ltp_mem_length; i < NSQ->sLTP_shp_buf_idx; i++ ) { |
413 | 89.3M | NSQ->sLTP_shp_Q14[ i ] = silk_SMULWW( gain_adj_Q16, NSQ->sLTP_shp_Q14[ i ] ); |
414 | 89.3M | } |
415 | | |
416 | | /* Scale long-term prediction state */ |
417 | 509k | if( signal_type == TYPE_VOICED && NSQ->rewhite_flag == 0 ) { |
418 | 3.38M | for( i = NSQ->sLTP_buf_idx - lag - LTP_ORDER / 2; i < NSQ->sLTP_buf_idx; i++ ) { |
419 | 3.30M | sLTP_Q15[ i ] = silk_SMULWW( gain_adj_Q16, sLTP_Q15[ i ] ); |
420 | 3.30M | } |
421 | 81.0k | } |
422 | | |
423 | 509k | NSQ->sLF_AR_shp_Q14 = silk_SMULWW( gain_adj_Q16, NSQ->sLF_AR_shp_Q14 ); |
424 | 509k | NSQ->sDiff_shp_Q14 = silk_SMULWW( gain_adj_Q16, NSQ->sDiff_shp_Q14 ); |
425 | | |
426 | | /* Scale short-term prediction and shaping states */ |
427 | 8.66M | for( i = 0; i < NSQ_LPC_BUF_LENGTH; i++ ) { |
428 | 8.15M | NSQ->sLPC_Q14[ i ] = silk_SMULWW( gain_adj_Q16, NSQ->sLPC_Q14[ i ] ); |
429 | 8.15M | } |
430 | 12.7M | for( i = 0; i < MAX_SHAPE_LPC_ORDER; i++ ) { |
431 | 12.2M | NSQ->sAR2_Q14[ i ] = silk_SMULWW( gain_adj_Q16, NSQ->sAR2_Q14[ i ] ); |
432 | 12.2M | } |
433 | | |
434 | | /* Save inverse gain */ |
435 | 509k | NSQ->prev_gain_Q16 = Gains_Q16[ subfr ]; |
436 | 509k | } |
437 | 587k | } |