Gitiles
Code Review Sign In
review.jami.net / jami-client-android / 8a1bab80c221136fb71c397be05a813932ca9fd4 / . / jni / libopus / silk / stereo_LR_to_MS.c
blob: 6a680e09ff3b696c31b94f993d78aa8599e3e856 [file] [log] [blame]
Alexandre Lision744f7422013-09-25 11:39:37 -0400[diff] [blame]1/***********************************************************************
2Copyright (c) 2006-2011, Skype Limited. All rights reserved.
3Redistribution and use in source and binary forms, with or without
4modification, are permitted provided that the following conditions
5are met:
6- Redistributions of source code must retain the above copyright notice,
7this list of conditions and the following disclaimer.
8- Redistributions in binary form must reproduce the above copyright
9notice, this list of conditions and the following disclaimer in the
10documentation and/or other materials provided with the distribution.
11- Neither the name of Internet Society, IETF or IETF Trust, nor the
12names of specific contributors, may be used to endorse or promote
13products derived from this software without specific prior written
14permission.
15THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS”
16AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
19LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
20CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
21SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
22INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
23CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
24ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
25POSSIBILITY OF SUCH DAMAGE.
26***********************************************************************/
27
28#ifdef HAVE_CONFIG_H
29#include "config.h"
30#endif
31
32#include "main.h"
33
34/* Convert Left/Right stereo signal to adaptive Mid/Side representation */
35void silk_stereo_LR_to_MS(
36 stereo_enc_state *state, /* I/O State */
37 opus_int16 x1[], /* I/O Left input signal, becomes mid signal */
38 opus_int16 x2[], /* I/O Right input signal, becomes side signal */
39 opus_int8 ix[ 2 ][ 3 ], /* O Quantization indices */
40 opus_int8 *mid_only_flag, /* O Flag: only mid signal coded */
41 opus_int32 mid_side_rates_bps[], /* O Bitrates for mid and side signals */
42 opus_int32 total_rate_bps, /* I Total bitrate */
43 opus_int prev_speech_act_Q8, /* I Speech activity level in previous frame */
44 opus_int toMono, /* I Last frame before a stereo->mono transition */
45 opus_int fs_kHz, /* I Sample rate (kHz) */
46 opus_int frame_length /* I Number of samples */
47)
48{
49 opus_int n, is10msFrame, denom_Q16, delta0_Q13, delta1_Q13;
50 opus_int32 sum, diff, smooth_coef_Q16, pred_Q13[ 2 ], pred0_Q13, pred1_Q13;
51 opus_int32 LP_ratio_Q14, HP_ratio_Q14, frac_Q16, frac_3_Q16, min_mid_rate_bps, width_Q14, w_Q24, deltaw_Q24;
52 opus_int16 side[ MAX_FRAME_LENGTH + 2 ];
53 opus_int16 LP_mid[ MAX_FRAME_LENGTH ], HP_mid[ MAX_FRAME_LENGTH ];
54 opus_int16 LP_side[ MAX_FRAME_LENGTH ], HP_side[ MAX_FRAME_LENGTH ];
55 opus_int16 *mid = &x1[ -2 ];
56
57 /* Convert to basic mid/side signals */
58 for( n = 0; n < frame_length + 2; n++ ) {
59 sum = x1[ n - 2 ] + (opus_int32)x2[ n - 2 ];
60 diff = x1[ n - 2 ] - (opus_int32)x2[ n - 2 ];
61 mid[ n ] = (opus_int16)silk_RSHIFT_ROUND( sum, 1 );
62 side[ n ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( diff, 1 ) );
63 }
64
65 /* Buffering */
66 silk_memcpy( mid, state->sMid, 2 * sizeof( opus_int16 ) );
67 silk_memcpy( side, state->sSide, 2 * sizeof( opus_int16 ) );
68 silk_memcpy( state->sMid, &mid[ frame_length ], 2 * sizeof( opus_int16 ) );
69 silk_memcpy( state->sSide, &side[ frame_length ], 2 * sizeof( opus_int16 ) );
70
71 /* LP and HP filter mid signal */
72 for( n = 0; n < frame_length; n++ ) {
73 sum = silk_RSHIFT_ROUND( silk_ADD_LSHIFT( mid[ n ] + mid[ n + 2 ], mid[ n + 1 ], 1 ), 2 );
74 LP_mid[ n ] = sum;
75 HP_mid[ n ] = mid[ n + 1 ] - sum;
76 }
77
78 /* LP and HP filter side signal */
79 for( n = 0; n < frame_length; n++ ) {
80 sum = silk_RSHIFT_ROUND( silk_ADD_LSHIFT( side[ n ] + side[ n + 2 ], side[ n + 1 ], 1 ), 2 );
81 LP_side[ n ] = sum;
82 HP_side[ n ] = side[ n + 1 ] - sum;
83 }
84
85 /* Find energies and predictors */
86 is10msFrame = frame_length == 10 * fs_kHz;
87 smooth_coef_Q16 = is10msFrame ?
88 SILK_FIX_CONST( STEREO_RATIO_SMOOTH_COEF / 2, 16 ) :
89 SILK_FIX_CONST( STEREO_RATIO_SMOOTH_COEF, 16 );
90 smooth_coef_Q16 = silk_SMULWB( silk_SMULBB( prev_speech_act_Q8, prev_speech_act_Q8 ), smooth_coef_Q16 );
91
92 pred_Q13[ 0 ] = silk_stereo_find_predictor( &LP_ratio_Q14, LP_mid, LP_side, &state->mid_side_amp_Q0[ 0 ], frame_length, smooth_coef_Q16 );
93 pred_Q13[ 1 ] = silk_stereo_find_predictor( &HP_ratio_Q14, HP_mid, HP_side, &state->mid_side_amp_Q0[ 2 ], frame_length, smooth_coef_Q16 );
94 /* Ratio of the norms of residual and mid signals */
95 frac_Q16 = silk_SMLABB( HP_ratio_Q14, LP_ratio_Q14, 3 );
96 frac_Q16 = silk_min( frac_Q16, SILK_FIX_CONST( 1, 16 ) );
97
98 /* Determine bitrate distribution between mid and side, and possibly reduce stereo width */
99 total_rate_bps -= is10msFrame ? 1200 : 600; /* Subtract approximate bitrate for coding stereo parameters */
100 if( total_rate_bps < 1 ) {
101 total_rate_bps = 1;
102 }
103 min_mid_rate_bps = silk_SMLABB( 2000, fs_kHz, 900 );
104 silk_assert( min_mid_rate_bps < 32767 );
105 /* Default bitrate distribution: 8 parts for Mid and (5+3*frac) parts for Side. so: mid_rate = ( 8 / ( 13 + 3 * frac ) ) * total_ rate */
106 frac_3_Q16 = silk_MUL( 3, frac_Q16 );
107 mid_side_rates_bps[ 0 ] = silk_DIV32_varQ( total_rate_bps, SILK_FIX_CONST( 8 + 5, 16 ) + frac_3_Q16, 16+3 );
108 /* If Mid bitrate below minimum, reduce stereo width */
109 if( mid_side_rates_bps[ 0 ] < min_mid_rate_bps ) {
110 mid_side_rates_bps[ 0 ] = min_mid_rate_bps;
111 mid_side_rates_bps[ 1 ] = total_rate_bps - mid_side_rates_bps[ 0 ];
112 /* width = 4 * ( 2 * side_rate - min_rate ) / ( ( 1 + 3 * frac ) * min_rate ) */
113 width_Q14 = silk_DIV32_varQ( silk_LSHIFT( mid_side_rates_bps[ 1 ], 1 ) - min_mid_rate_bps,
114 silk_SMULWB( SILK_FIX_CONST( 1, 16 ) + frac_3_Q16, min_mid_rate_bps ), 14+2 );
115 width_Q14 = silk_LIMIT( width_Q14, 0, SILK_FIX_CONST( 1, 14 ) );
116 } else {
117 mid_side_rates_bps[ 1 ] = total_rate_bps - mid_side_rates_bps[ 0 ];
118 width_Q14 = SILK_FIX_CONST( 1, 14 );
119 }
120
121 /* Smoother */
122 state->smth_width_Q14 = (opus_int16)silk_SMLAWB( state->smth_width_Q14, width_Q14 - state->smth_width_Q14, smooth_coef_Q16 );
123
124 /* At very low bitrates or for inputs that are nearly amplitude panned, switch to panned-mono coding */
125 *mid_only_flag = 0;
126 if( toMono ) {
127 /* Last frame before stereo->mono transition; collapse stereo width */
128 width_Q14 = 0;
129 pred_Q13[ 0 ] = 0;
130 pred_Q13[ 1 ] = 0;
131 silk_stereo_quant_pred( pred_Q13, ix );
132 } else if( state->width_prev_Q14 == 0 &&
133 ( 8 * total_rate_bps < 13 * min_mid_rate_bps || silk_SMULWB( frac_Q16, state->smth_width_Q14 ) < SILK_FIX_CONST( 0.05, 14 ) ) )
134 {
135 /* Code as panned-mono; previous frame already had zero width */
136 /* Scale down and quantize predictors */
137 pred_Q13[ 0 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 0 ] ), 14 );
138 pred_Q13[ 1 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 1 ] ), 14 );
139 silk_stereo_quant_pred( pred_Q13, ix );
140 /* Collapse stereo width */
141 width_Q14 = 0;
142 pred_Q13[ 0 ] = 0;
143 pred_Q13[ 1 ] = 0;
144 mid_side_rates_bps[ 0 ] = total_rate_bps;
145 mid_side_rates_bps[ 1 ] = 0;
146 *mid_only_flag = 1;
147 } else if( state->width_prev_Q14 != 0 &&
148 ( 8 * total_rate_bps < 11 * min_mid_rate_bps || silk_SMULWB( frac_Q16, state->smth_width_Q14 ) < SILK_FIX_CONST( 0.02, 14 ) ) )
149 {
150 /* Transition to zero-width stereo */
151 /* Scale down and quantize predictors */
152 pred_Q13[ 0 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 0 ] ), 14 );
153 pred_Q13[ 1 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 1 ] ), 14 );
154 silk_stereo_quant_pred( pred_Q13, ix );
155 /* Collapse stereo width */
156 width_Q14 = 0;
157 pred_Q13[ 0 ] = 0;
158 pred_Q13[ 1 ] = 0;
159 } else if( state->smth_width_Q14 > SILK_FIX_CONST( 0.95, 14 ) ) {
160 /* Full-width stereo coding */
161 silk_stereo_quant_pred( pred_Q13, ix );
162 width_Q14 = SILK_FIX_CONST( 1, 14 );
163 } else {
164 /* Reduced-width stereo coding; scale down and quantize predictors */
165 pred_Q13[ 0 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 0 ] ), 14 );
166 pred_Q13[ 1 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 1 ] ), 14 );
167 silk_stereo_quant_pred( pred_Q13, ix );
168 width_Q14 = state->smth_width_Q14;
169 }
170
171 /* Make sure to keep on encoding until the tapered output has been transmitted */
172 if( *mid_only_flag == 1 ) {
173 state->silent_side_len += frame_length - STEREO_INTERP_LEN_MS * fs_kHz;
174 if( state->silent_side_len < LA_SHAPE_MS * fs_kHz ) {
175 *mid_only_flag = 0;
176 } else {
177 /* Limit to avoid wrapping around */
178 state->silent_side_len = 10000;
179 }
180 } else {
181 state->silent_side_len = 0;
182 }
183
184 if( *mid_only_flag == 0 && mid_side_rates_bps[ 1 ] < 1 ) {
185 mid_side_rates_bps[ 1 ] = 1;
186 mid_side_rates_bps[ 0 ] = silk_max_int( 1, total_rate_bps - mid_side_rates_bps[ 1 ]);
187 }
188
189 /* Interpolate predictors and subtract prediction from side channel */
190 pred0_Q13 = -state->pred_prev_Q13[ 0 ];
191 pred1_Q13 = -state->pred_prev_Q13[ 1 ];
192 w_Q24 = silk_LSHIFT( state->width_prev_Q14, 10 );
193 denom_Q16 = silk_DIV32_16( (opus_int32)1 << 16, STEREO_INTERP_LEN_MS * fs_kHz );
194 delta0_Q13 = -silk_RSHIFT_ROUND( silk_SMULBB( pred_Q13[ 0 ] - state->pred_prev_Q13[ 0 ], denom_Q16 ), 16 );
195 delta1_Q13 = -silk_RSHIFT_ROUND( silk_SMULBB( pred_Q13[ 1 ] - state->pred_prev_Q13[ 1 ], denom_Q16 ), 16 );
196 deltaw_Q24 = silk_LSHIFT( silk_SMULWB( width_Q14 - state->width_prev_Q14, denom_Q16 ), 10 );
197 for( n = 0; n < STEREO_INTERP_LEN_MS * fs_kHz; n++ ) {
198 pred0_Q13 += delta0_Q13;
199 pred1_Q13 += delta1_Q13;
200 w_Q24 += deltaw_Q24;
201 sum = silk_LSHIFT( silk_ADD_LSHIFT( mid[ n ] + mid[ n + 2 ], mid[ n + 1 ], 1 ), 9 ); /* Q11 */
202 sum = silk_SMLAWB( silk_SMULWB( w_Q24, side[ n + 1 ] ), sum, pred0_Q13 ); /* Q8 */
203 sum = silk_SMLAWB( sum, silk_LSHIFT( (opus_int32)mid[ n + 1 ], 11 ), pred1_Q13 ); /* Q8 */
204 x2[ n - 1 ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( sum, 8 ) );
205 }
206
207 pred0_Q13 = -pred_Q13[ 0 ];
208 pred1_Q13 = -pred_Q13[ 1 ];
209 w_Q24 = silk_LSHIFT( width_Q14, 10 );
210 for( n = STEREO_INTERP_LEN_MS * fs_kHz; n < frame_length; n++ ) {
211 sum = silk_LSHIFT( silk_ADD_LSHIFT( mid[ n ] + mid[ n + 2 ], mid[ n + 1 ], 1 ), 9 ); /* Q11 */
212 sum = silk_SMLAWB( silk_SMULWB( w_Q24, side[ n + 1 ] ), sum, pred0_Q13 ); /* Q8 */
213 sum = silk_SMLAWB( sum, silk_LSHIFT( (opus_int32)mid[ n + 1 ], 11 ), pred1_Q13 ); /* Q8 */
214 x2[ n - 1 ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( sum, 8 ) );
215 }
216 state->pred_prev_Q13[ 0 ] = (opus_int16)pred_Q13[ 0 ];
217 state->pred_prev_Q13[ 1 ] = (opus_int16)pred_Q13[ 1 ];
218 state->width_prev_Q14 = (opus_int16)width_Q14;
219}