Gitiles
Code Review Sign In
review.jami.net / jami-client-android / ec71b2d0acf6e348a7ebe9386f10fecf10b3b49f / . / jni / libopus / silk / enc_API.c
blob: ec7915ce242e064a23a864d1a13b64783d832d34 [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#include "define.h"
32#include "API.h"
33#include "control.h"
34#include "typedef.h"
35#include "structs.h"
36#include "tuning_parameters.h"
37#ifdef FIXED_POINT
38#include "main_FIX.h"
39#else
40#include "main_FLP.h"
41#endif
42
43/***************************************/
44/* Read control structure from encoder */
45/***************************************/
46static opus_int silk_QueryEncoder( /* O Returns error code */
47 const void *encState, /* I State */
48 silk_EncControlStruct *encStatus /* O Encoder Status */
49);
50
51/****************************************/
52/* Encoder functions */
53/****************************************/
54
55opus_int silk_Get_Encoder_Size( /* O Returns error code */
56 opus_int *encSizeBytes /* O Number of bytes in SILK encoder state */
57)
58{
59 opus_int ret = SILK_NO_ERROR;
60
61 *encSizeBytes = sizeof( silk_encoder );
62
63 return ret;
64}
65
66/*************************/
67/* Init or Reset encoder */
68/*************************/
69opus_int silk_InitEncoder( /* O Returns error code */
70 void *encState, /* I/O State */
71 silk_EncControlStruct *encStatus /* O Encoder Status */
72)
73{
74 silk_encoder *psEnc;
75 opus_int n, ret = SILK_NO_ERROR;
76
77 psEnc = (silk_encoder *)encState;
78
79 /* Reset encoder */
80 silk_memset( psEnc, 0, sizeof( silk_encoder ) );
81 for( n = 0; n < ENCODER_NUM_CHANNELS; n++ ) {
82 if( ret += silk_init_encoder( &psEnc->state_Fxx[ n ] ) ) {
83 silk_assert( 0 );
84 }
85 }
86
87 psEnc->nChannelsAPI = 1;
88 psEnc->nChannelsInternal = 1;
89
90 /* Read control structure */
91 if( ret += silk_QueryEncoder( encState, encStatus ) ) {
92 silk_assert( 0 );
93 }
94
95 return ret;
96}
97
98/***************************************/
99/* Read control structure from encoder */
100/***************************************/
101static opus_int silk_QueryEncoder( /* O Returns error code */
102 const void *encState, /* I State */
103 silk_EncControlStruct *encStatus /* O Encoder Status */
104)
105{
106 opus_int ret = SILK_NO_ERROR;
107 silk_encoder_state_Fxx *state_Fxx;
108 silk_encoder *psEnc = (silk_encoder *)encState;
109
110 state_Fxx = psEnc->state_Fxx;
111
112 encStatus->nChannelsAPI = psEnc->nChannelsAPI;
113 encStatus->nChannelsInternal = psEnc->nChannelsInternal;
114 encStatus->API_sampleRate = state_Fxx[ 0 ].sCmn.API_fs_Hz;
115 encStatus->maxInternalSampleRate = state_Fxx[ 0 ].sCmn.maxInternal_fs_Hz;
116 encStatus->minInternalSampleRate = state_Fxx[ 0 ].sCmn.minInternal_fs_Hz;
117 encStatus->desiredInternalSampleRate = state_Fxx[ 0 ].sCmn.desiredInternal_fs_Hz;
118 encStatus->payloadSize_ms = state_Fxx[ 0 ].sCmn.PacketSize_ms;
119 encStatus->bitRate = state_Fxx[ 0 ].sCmn.TargetRate_bps;
120 encStatus->packetLossPercentage = state_Fxx[ 0 ].sCmn.PacketLoss_perc;
121 encStatus->complexity = state_Fxx[ 0 ].sCmn.Complexity;
122 encStatus->useInBandFEC = state_Fxx[ 0 ].sCmn.useInBandFEC;
123 encStatus->useDTX = state_Fxx[ 0 ].sCmn.useDTX;
124 encStatus->useCBR = state_Fxx[ 0 ].sCmn.useCBR;
125 encStatus->internalSampleRate = silk_SMULBB( state_Fxx[ 0 ].sCmn.fs_kHz, 1000 );
126 encStatus->allowBandwidthSwitch = state_Fxx[ 0 ].sCmn.allow_bandwidth_switch;
127 encStatus->inWBmodeWithoutVariableLP = state_Fxx[ 0 ].sCmn.fs_kHz == 16 && state_Fxx[ 0 ].sCmn.sLP.mode == 0;
128
129 return ret;
130}
131
132
133/**************************/
134/* Encode frame with Silk */
135/**************************/
136/* Note: if prefillFlag is set, the input must contain 10 ms of audio, irrespective of what */
137/* encControl->payloadSize_ms is set to */
138opus_int silk_Encode( /* O Returns error code */
139 void *encState, /* I/O State */
140 silk_EncControlStruct *encControl, /* I Control status */
141 const opus_int16 *samplesIn, /* I Speech sample input vector */
142 opus_int nSamplesIn, /* I Number of samples in input vector */
143 ec_enc *psRangeEnc, /* I/O Compressor data structure */
144 opus_int32 *nBytesOut, /* I/O Number of bytes in payload (input: Max bytes) */
145 const opus_int prefillFlag /* I Flag to indicate prefilling buffers no coding */
146)
147{
148 opus_int n, i, nBits, flags, tmp_payloadSize_ms = 0, tmp_complexity = 0, ret = 0;
149 opus_int nSamplesToBuffer, nBlocksOf10ms, nSamplesFromInput = 0;
150 opus_int speech_act_thr_for_switch_Q8;
151 opus_int32 TargetRate_bps, MStargetRates_bps[ 2 ], channelRate_bps, LBRR_symbol, sum;
152 silk_encoder *psEnc = ( silk_encoder * )encState;
153 opus_int16 buf[ MAX_FRAME_LENGTH_MS * MAX_API_FS_KHZ ];
154 opus_int transition, curr_block, tot_blocks;
155
156 psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded = psEnc->state_Fxx[ 1 ].sCmn.nFramesEncoded = 0;
157
158 /* Check values in encoder control structure */
159 if( ( ret = check_control_input( encControl ) != 0 ) ) {
160 silk_assert( 0 );
161 return ret;
162 }
163
164 encControl->switchReady = 0;
165
166 if( encControl->nChannelsInternal > psEnc->nChannelsInternal ) {
167 /* Mono -> Stereo transition: init state of second channel and stereo state */
168 ret += silk_init_encoder( &psEnc->state_Fxx[ 1 ] );
169 silk_memset( psEnc->sStereo.pred_prev_Q13, 0, sizeof( psEnc->sStereo.pred_prev_Q13 ) );
170 silk_memset( psEnc->sStereo.sSide, 0, sizeof( psEnc->sStereo.sSide ) );
171 psEnc->sStereo.mid_side_amp_Q0[ 0 ] = 0;
172 psEnc->sStereo.mid_side_amp_Q0[ 1 ] = 1;
173 psEnc->sStereo.mid_side_amp_Q0[ 2 ] = 0;
174 psEnc->sStereo.mid_side_amp_Q0[ 3 ] = 1;
175 psEnc->sStereo.width_prev_Q14 = 0;
176 psEnc->sStereo.smth_width_Q14 = SILK_FIX_CONST( 1, 14 );
177 if( psEnc->nChannelsAPI == 2 ) {
178 silk_memcpy( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state, &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, sizeof( silk_resampler_state_struct ) );
179 silk_memcpy( &psEnc->state_Fxx[ 1 ].sCmn.In_HP_State, &psEnc->state_Fxx[ 0 ].sCmn.In_HP_State, sizeof( psEnc->state_Fxx[ 1 ].sCmn.In_HP_State ) );
180 }
181 }
182
183 transition = (encControl->payloadSize_ms != psEnc->state_Fxx[ 0 ].sCmn.PacketSize_ms) || (psEnc->nChannelsInternal != encControl->nChannelsInternal);
184
185 psEnc->nChannelsAPI = encControl->nChannelsAPI;
186 psEnc->nChannelsInternal = encControl->nChannelsInternal;
187
188 nBlocksOf10ms = silk_DIV32( 100 * nSamplesIn, encControl->API_sampleRate );
189 tot_blocks = ( nBlocksOf10ms > 1 ) ? nBlocksOf10ms >> 1 : 1;
190 curr_block = 0;
191 if( prefillFlag ) {
192 /* Only accept input length of 10 ms */
193 if( nBlocksOf10ms != 1 ) {
194 ret = SILK_ENC_INPUT_INVALID_NO_OF_SAMPLES;
195 silk_assert( 0 );
196 return ret;
197 }
198 /* Reset Encoder */
199 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
200 if( (ret = silk_init_encoder( &psEnc->state_Fxx[ n ] ) ) != 0 ) {
201 silk_assert( 0 );
202 }
203 }
204 tmp_payloadSize_ms = encControl->payloadSize_ms;
205 encControl->payloadSize_ms = 10;
206 tmp_complexity = encControl->complexity;
207 encControl->complexity = 0;
208 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
209 psEnc->state_Fxx[ n ].sCmn.controlled_since_last_payload = 0;
210 psEnc->state_Fxx[ n ].sCmn.prefillFlag = 1;
211 }
212 } else {
213 /* Only accept input lengths that are a multiple of 10 ms */
214 if( nBlocksOf10ms * encControl->API_sampleRate != 100 * nSamplesIn || nSamplesIn < 0 ) {
215 ret = SILK_ENC_INPUT_INVALID_NO_OF_SAMPLES;
216 silk_assert( 0 );
217 return ret;
218 }
219 /* Make sure no more than one packet can be produced */
220 if( 1000 * (opus_int32)nSamplesIn > encControl->payloadSize_ms * encControl->API_sampleRate ) {
221 ret = SILK_ENC_INPUT_INVALID_NO_OF_SAMPLES;
222 silk_assert( 0 );
223 return ret;
224 }
225 }
226
227 TargetRate_bps = silk_RSHIFT32( encControl->bitRate, encControl->nChannelsInternal - 1 );
228 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
229 /* Force the side channel to the same rate as the mid */
230 opus_int force_fs_kHz = (n==1) ? psEnc->state_Fxx[0].sCmn.fs_kHz : 0;
231 if( ( ret = silk_control_encoder( &psEnc->state_Fxx[ n ], encControl, TargetRate_bps, psEnc->allowBandwidthSwitch, n, force_fs_kHz ) ) != 0 ) {
232 silk_assert( 0 );
233 return ret;
234 }
235 if( psEnc->state_Fxx[n].sCmn.first_frame_after_reset || transition ) {
236 for( i = 0; i < psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket; i++ ) {
237 psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i ] = 0;
238 }
239 }
240 psEnc->state_Fxx[ n ].sCmn.inDTX = psEnc->state_Fxx[ n ].sCmn.useDTX;
241 }
242 silk_assert( encControl->nChannelsInternal == 1 || psEnc->state_Fxx[ 0 ].sCmn.fs_kHz == psEnc->state_Fxx[ 1 ].sCmn.fs_kHz );
243
244 /* Input buffering/resampling and encoding */
245 while( 1 ) {
246 nSamplesToBuffer = psEnc->state_Fxx[ 0 ].sCmn.frame_length - psEnc->state_Fxx[ 0 ].sCmn.inputBufIx;
247 nSamplesToBuffer = silk_min( nSamplesToBuffer, 10 * nBlocksOf10ms * psEnc->state_Fxx[ 0 ].sCmn.fs_kHz );
248 nSamplesFromInput = silk_DIV32_16( nSamplesToBuffer * psEnc->state_Fxx[ 0 ].sCmn.API_fs_Hz, psEnc->state_Fxx[ 0 ].sCmn.fs_kHz * 1000 );
249 /* Resample and write to buffer */
250 if( encControl->nChannelsAPI == 2 && encControl->nChannelsInternal == 2 ) {
251 opus_int id = psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded;
252 for( n = 0; n < nSamplesFromInput; n++ ) {
253 buf[ n ] = samplesIn[ 2 * n ];
254 }
255 /* Making sure to start both resamplers from the same state when switching from mono to stereo */
256 if( psEnc->nPrevChannelsInternal == 1 && id==0 ) {
257 silk_memcpy( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state, &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, sizeof(psEnc->state_Fxx[ 1 ].sCmn.resampler_state));
258 }
259
260 ret += silk_resampler( &psEnc->state_Fxx[ 0 ].sCmn.resampler_state,
261 &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput );
262 psEnc->state_Fxx[ 0 ].sCmn.inputBufIx += nSamplesToBuffer;
263
264 nSamplesToBuffer = psEnc->state_Fxx[ 1 ].sCmn.frame_length - psEnc->state_Fxx[ 1 ].sCmn.inputBufIx;
265 nSamplesToBuffer = silk_min( nSamplesToBuffer, 10 * nBlocksOf10ms * psEnc->state_Fxx[ 1 ].sCmn.fs_kHz );
266 for( n = 0; n < nSamplesFromInput; n++ ) {
267 buf[ n ] = samplesIn[ 2 * n + 1 ];
268 }
269 ret += silk_resampler( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state,
270 &psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ psEnc->state_Fxx[ 1 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput );
271
272 psEnc->state_Fxx[ 1 ].sCmn.inputBufIx += nSamplesToBuffer;
273 } else if( encControl->nChannelsAPI == 2 && encControl->nChannelsInternal == 1 ) {
274 /* Combine left and right channels before resampling */
275 for( n = 0; n < nSamplesFromInput; n++ ) {
276 sum = samplesIn[ 2 * n ] + samplesIn[ 2 * n + 1 ];
277 buf[ n ] = (opus_int16)silk_RSHIFT_ROUND( sum, 1 );
278 }
279 ret += silk_resampler( &psEnc->state_Fxx[ 0 ].sCmn.resampler_state,
280 &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput );
281 /* On the first mono frame, average the results for the two resampler states */
282 if( psEnc->nPrevChannelsInternal == 2 && psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded == 0 ) {
283 ret += silk_resampler( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state,
284 &psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ psEnc->state_Fxx[ 1 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput );
285 for( n = 0; n < psEnc->state_Fxx[ 0 ].sCmn.frame_length; n++ ) {
286 psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx+n+2 ] =
287 silk_RSHIFT(psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx+n+2 ]
288 + psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ psEnc->state_Fxx[ 1 ].sCmn.inputBufIx+n+2 ], 1);
289 }
290 }
291 psEnc->state_Fxx[ 0 ].sCmn.inputBufIx += nSamplesToBuffer;
292 } else {
293 silk_assert( encControl->nChannelsAPI == 1 && encControl->nChannelsInternal == 1 );
294 silk_memcpy(buf, samplesIn, nSamplesFromInput*sizeof(opus_int16));
295 ret += silk_resampler( &psEnc->state_Fxx[ 0 ].sCmn.resampler_state,
296 &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput );
297 psEnc->state_Fxx[ 0 ].sCmn.inputBufIx += nSamplesToBuffer;
298 }
299
300 samplesIn += nSamplesFromInput * encControl->nChannelsAPI;
301 nSamplesIn -= nSamplesFromInput;
302
303 /* Default */
304 psEnc->allowBandwidthSwitch = 0;
305
306 /* Silk encoder */
307 if( psEnc->state_Fxx[ 0 ].sCmn.inputBufIx >= psEnc->state_Fxx[ 0 ].sCmn.frame_length ) {
308 /* Enough data in input buffer, so encode */
309 silk_assert( psEnc->state_Fxx[ 0 ].sCmn.inputBufIx == psEnc->state_Fxx[ 0 ].sCmn.frame_length );
310 silk_assert( encControl->nChannelsInternal == 1 || psEnc->state_Fxx[ 1 ].sCmn.inputBufIx == psEnc->state_Fxx[ 1 ].sCmn.frame_length );
311
312 /* Deal with LBRR data */
313 if( psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded == 0 && !prefillFlag ) {
314 /* Create space at start of payload for VAD and FEC flags */
315 opus_uint8 iCDF[ 2 ] = { 0, 0 };
316 iCDF[ 0 ] = 256 - silk_RSHIFT( 256, ( psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket + 1 ) * encControl->nChannelsInternal );
317 ec_enc_icdf( psRangeEnc, 0, iCDF, 8 );
318
319 /* Encode any LBRR data from previous packet */
320 /* Encode LBRR flags */
321 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
322 LBRR_symbol = 0;
323 for( i = 0; i < psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket; i++ ) {
324 LBRR_symbol |= silk_LSHIFT( psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i ], i );
325 }
326 psEnc->state_Fxx[ n ].sCmn.LBRR_flag = LBRR_symbol > 0 ? 1 : 0;
327 if( LBRR_symbol && psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket > 1 ) {
328 ec_enc_icdf( psRangeEnc, LBRR_symbol - 1, silk_LBRR_flags_iCDF_ptr[ psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket - 2 ], 8 );
329 }
330 }
331
332 /* Code LBRR indices and excitation signals */
333 for( i = 0; i < psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket; i++ ) {
334 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
335 if( psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i ] ) {
336 opus_int condCoding;
337
338 if( encControl->nChannelsInternal == 2 && n == 0 ) {
339 silk_stereo_encode_pred( psRangeEnc, psEnc->sStereo.predIx[ i ] );
340 /* For LBRR data there's no need to code the mid-only flag if the side-channel LBRR flag is set */
341 if( psEnc->state_Fxx[ 1 ].sCmn.LBRR_flags[ i ] == 0 ) {
342 silk_stereo_encode_mid_only( psRangeEnc, psEnc->sStereo.mid_only_flags[ i ] );
343 }
344 }
345 /* Use conditional coding if previous frame available */
346 if( i > 0 && psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i - 1 ] ) {
347 condCoding = CODE_CONDITIONALLY;
348 } else {
349 condCoding = CODE_INDEPENDENTLY;
350 }
351 silk_encode_indices( &psEnc->state_Fxx[ n ].sCmn, psRangeEnc, i, 1, condCoding );
352 silk_encode_pulses( psRangeEnc, psEnc->state_Fxx[ n ].sCmn.indices_LBRR[i].signalType, psEnc->state_Fxx[ n ].sCmn.indices_LBRR[i].quantOffsetType,
353 psEnc->state_Fxx[ n ].sCmn.pulses_LBRR[ i ], psEnc->state_Fxx[ n ].sCmn.frame_length );
354 }
355 }
356 }
357
358 /* Reset LBRR flags */
359 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
360 silk_memset( psEnc->state_Fxx[ n ].sCmn.LBRR_flags, 0, sizeof( psEnc->state_Fxx[ n ].sCmn.LBRR_flags ) );
361 }
362 }
363
364 silk_HP_variable_cutoff( psEnc->state_Fxx );
365
366 /* Total target bits for packet */
367 nBits = silk_DIV32_16( silk_MUL( encControl->bitRate, encControl->payloadSize_ms ), 1000 );
368 /* Subtract half of the bits already used */
369 if( !prefillFlag ) {
370 nBits -= ec_tell( psRangeEnc ) >> 1;
371 }
372 /* Divide by number of uncoded frames left in packet */
373 nBits = silk_DIV32_16( nBits, psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket - psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded );
374 /* Convert to bits/second */
375 if( encControl->payloadSize_ms == 10 ) {
376 TargetRate_bps = silk_SMULBB( nBits, 100 );
377 } else {
378 TargetRate_bps = silk_SMULBB( nBits, 50 );
379 }
380 /* Subtract fraction of bits in excess of target in previous packets */
381 TargetRate_bps -= silk_DIV32_16( silk_MUL( psEnc->nBitsExceeded, 1000 ), BITRESERVOIR_DECAY_TIME_MS );
382 /* Never exceed input bitrate */
383 TargetRate_bps = silk_LIMIT( TargetRate_bps, encControl->bitRate, 5000 );
384
385 /* Convert Left/Right to Mid/Side */
386 if( encControl->nChannelsInternal == 2 ) {
387 silk_stereo_LR_to_MS( &psEnc->sStereo, &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ 2 ], &psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ 2 ],
388 psEnc->sStereo.predIx[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ], &psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ],
389 MStargetRates_bps, TargetRate_bps, psEnc->state_Fxx[ 0 ].sCmn.speech_activity_Q8, encControl->toMono,
390 psEnc->state_Fxx[ 0 ].sCmn.fs_kHz, psEnc->state_Fxx[ 0 ].sCmn.frame_length );
391 if( psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] == 0 ) {
392 /* Reset side channel encoder memory for first frame with side coding */
393 if( psEnc->prev_decode_only_middle == 1 ) {
394 silk_memset( &psEnc->state_Fxx[ 1 ].sShape, 0, sizeof( psEnc->state_Fxx[ 1 ].sShape ) );
395 silk_memset( &psEnc->state_Fxx[ 1 ].sPrefilt, 0, sizeof( psEnc->state_Fxx[ 1 ].sPrefilt ) );
396 silk_memset( &psEnc->state_Fxx[ 1 ].sCmn.sNSQ, 0, sizeof( psEnc->state_Fxx[ 1 ].sCmn.sNSQ ) );
397 silk_memset( psEnc->state_Fxx[ 1 ].sCmn.prev_NLSFq_Q15, 0, sizeof( psEnc->state_Fxx[ 1 ].sCmn.prev_NLSFq_Q15 ) );
398 silk_memset( &psEnc->state_Fxx[ 1 ].sCmn.sLP.In_LP_State, 0, sizeof( psEnc->state_Fxx[ 1 ].sCmn.sLP.In_LP_State ) );
399 psEnc->state_Fxx[ 1 ].sCmn.prevLag = 100;
400 psEnc->state_Fxx[ 1 ].sCmn.sNSQ.lagPrev = 100;
401 psEnc->state_Fxx[ 1 ].sShape.LastGainIndex = 10;
402 psEnc->state_Fxx[ 1 ].sCmn.prevSignalType = TYPE_NO_VOICE_ACTIVITY;
403 psEnc->state_Fxx[ 1 ].sCmn.sNSQ.prev_gain_Q16 = 65536;
404 psEnc->state_Fxx[ 1 ].sCmn.first_frame_after_reset = 1;
405 }
406 silk_encode_do_VAD_Fxx( &psEnc->state_Fxx[ 1 ] );
407 } else {
408 psEnc->state_Fxx[ 1 ].sCmn.VAD_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] = 0;
409 }
410 if( !prefillFlag ) {
411 silk_stereo_encode_pred( psRangeEnc, psEnc->sStereo.predIx[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] );
412 if( psEnc->state_Fxx[ 1 ].sCmn.VAD_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] == 0 ) {
413 silk_stereo_encode_mid_only( psRangeEnc, psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] );
414 }
415 }
416 } else {
417 /* Buffering */
418 silk_memcpy( psEnc->state_Fxx[ 0 ].sCmn.inputBuf, psEnc->sStereo.sMid, 2 * sizeof( opus_int16 ) );
419 silk_memcpy( psEnc->sStereo.sMid, &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.frame_length ], 2 * sizeof( opus_int16 ) );
420 }
421 silk_encode_do_VAD_Fxx( &psEnc->state_Fxx[ 0 ] );
422
423 /* Encode */
424 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
425 opus_int maxBits, useCBR;
426
427 /* Handling rate constraints */
428 maxBits = encControl->maxBits;
429 if( tot_blocks == 2 && curr_block == 0 ) {
430 maxBits = maxBits * 3 / 5;
431 } else if( tot_blocks == 3 ) {
432 if( curr_block == 0 ) {
433 maxBits = maxBits * 2 / 5;
434 } else if( curr_block == 1 ) {
435 maxBits = maxBits * 3 / 4;
436 }
437 }
438 useCBR = encControl->useCBR && curr_block == tot_blocks - 1;
439
440 if( encControl->nChannelsInternal == 1 ) {
441 channelRate_bps = TargetRate_bps;
442 } else {
443 channelRate_bps = MStargetRates_bps[ n ];
444 if( n == 0 && MStargetRates_bps[ 1 ] > 0 ) {
445 useCBR = 0;
446 /* Give mid up to 1/2 of the max bits for that frame */
447 maxBits -= encControl->maxBits / ( tot_blocks * 2 );
448 }
449 }
450
451 if( channelRate_bps > 0 ) {
452 opus_int condCoding;
453
454 silk_control_SNR( &psEnc->state_Fxx[ n ].sCmn, channelRate_bps );
455
456 /* Use independent coding if no previous frame available */
457 if( psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded - n <= 0 ) {
458 condCoding = CODE_INDEPENDENTLY;
459 } else if( n > 0 && psEnc->prev_decode_only_middle ) {
460 /* If we skipped a side frame in this packet, we don't
461 need LTP scaling; the LTP state is well-defined. */
462 condCoding = CODE_INDEPENDENTLY_NO_LTP_SCALING;
463 } else {
464 condCoding = CODE_CONDITIONALLY;
465 }
466 if( ( ret = silk_encode_frame_Fxx( &psEnc->state_Fxx[ n ], nBytesOut, psRangeEnc, condCoding, maxBits, useCBR ) ) != 0 ) {
467 silk_assert( 0 );
468 }
469 }
470 psEnc->state_Fxx[ n ].sCmn.controlled_since_last_payload = 0;
471 psEnc->state_Fxx[ n ].sCmn.inputBufIx = 0;
472 psEnc->state_Fxx[ n ].sCmn.nFramesEncoded++;
473 }
474 psEnc->prev_decode_only_middle = psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded - 1 ];
475
476 /* Insert VAD and FEC flags at beginning of bitstream */
477 if( *nBytesOut > 0 && psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded == psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket) {
478 flags = 0;
479 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
480 for( i = 0; i < psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket; i++ ) {
481 flags = silk_LSHIFT( flags, 1 );
482 flags |= psEnc->state_Fxx[ n ].sCmn.VAD_flags[ i ];
483 }
484 flags = silk_LSHIFT( flags, 1 );
485 flags |= psEnc->state_Fxx[ n ].sCmn.LBRR_flag;
486 }
487 if( !prefillFlag ) {
488 ec_enc_patch_initial_bits( psRangeEnc, flags, ( psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket + 1 ) * encControl->nChannelsInternal );
489 }
490
491 /* Return zero bytes if all channels DTXed */
492 if( psEnc->state_Fxx[ 0 ].sCmn.inDTX && ( encControl->nChannelsInternal == 1 || psEnc->state_Fxx[ 1 ].sCmn.inDTX ) ) {
493 *nBytesOut = 0;
494 }
495
496 psEnc->nBitsExceeded += *nBytesOut * 8;
497 psEnc->nBitsExceeded -= silk_DIV32_16( silk_MUL( encControl->bitRate, encControl->payloadSize_ms ), 1000 );
498 psEnc->nBitsExceeded = silk_LIMIT( psEnc->nBitsExceeded, 0, 10000 );
499
500 /* Update flag indicating if bandwidth switching is allowed */
501 speech_act_thr_for_switch_Q8 = silk_SMLAWB( SILK_FIX_CONST( SPEECH_ACTIVITY_DTX_THRES, 8 ),
502 SILK_FIX_CONST( ( 1 - SPEECH_ACTIVITY_DTX_THRES ) / MAX_BANDWIDTH_SWITCH_DELAY_MS, 16 + 8 ), psEnc->timeSinceSwitchAllowed_ms );
503 if( psEnc->state_Fxx[ 0 ].sCmn.speech_activity_Q8 < speech_act_thr_for_switch_Q8 ) {
504 psEnc->allowBandwidthSwitch = 1;
505 psEnc->timeSinceSwitchAllowed_ms = 0;
506 } else {
507 psEnc->allowBandwidthSwitch = 0;
508 psEnc->timeSinceSwitchAllowed_ms += encControl->payloadSize_ms;
509 }
510 }
511
512 if( nSamplesIn == 0 ) {
513 break;
514 }
515 } else {
516 break;
517 }
518 curr_block++;
519 }
520
521 psEnc->nPrevChannelsInternal = encControl->nChannelsInternal;
522
523 encControl->allowBandwidthSwitch = psEnc->allowBandwidthSwitch;
524 encControl->inWBmodeWithoutVariableLP = psEnc->state_Fxx[ 0 ].sCmn.fs_kHz == 16 && psEnc->state_Fxx[ 0 ].sCmn.sLP.mode == 0;
525 encControl->internalSampleRate = silk_SMULBB( psEnc->state_Fxx[ 0 ].sCmn.fs_kHz, 1000 );
526 encControl->stereoWidth_Q14 = encControl->toMono ? 0 : psEnc->sStereo.smth_width_Q14;
527 if( prefillFlag ) {
528 encControl->payloadSize_ms = tmp_payloadSize_ms;
529 encControl->complexity = tmp_complexity;
530 for( n = 0; n < encControl->nChannelsInternal; n++ ) {
531 psEnc->state_Fxx[ n ].sCmn.controlled_since_last_payload = 0;
532 psEnc->state_Fxx[ n ].sCmn.prefillFlag = 0;
533 }
534 }
535
536 return ret;
537}
538