Line | Count | Source (jump to first uncovered line) |
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 | | #include "define.h" |
32 | | #include "API.h" |
33 | | #include "control.h" |
34 | | #include "typedef.h" |
35 | | #include "stack_alloc.h" |
36 | | #include "structs.h" |
37 | | #include "tuning_parameters.h" |
38 | | #ifdef FIXED_POINT |
39 | | #include "main_FIX.h" |
40 | | #else |
41 | | #include "main_FLP.h" |
42 | | #endif |
43 | | |
44 | | #ifdef ENABLE_DRED |
45 | | #include "dred_encoder.h" |
46 | | #endif |
47 | | |
48 | | /***************************************/ |
49 | | /* Read control structure from encoder */ |
50 | | /***************************************/ |
51 | | static opus_int silk_QueryEncoder( /* O Returns error code */ |
52 | | const void *encState, /* I State */ |
53 | | silk_EncControlStruct *encStatus /* O Encoder Status */ |
54 | | ); |
55 | | |
56 | | /****************************************/ |
57 | | /* Encoder functions */ |
58 | | /****************************************/ |
59 | | |
60 | | opus_int silk_Get_Encoder_Size( /* O Returns error code */ |
61 | | opus_int *encSizeBytes /* O Number of bytes in SILK encoder state */ |
62 | | ) |
63 | 0 | { |
64 | 0 | opus_int ret = SILK_NO_ERROR; |
65 | |
|
66 | 0 | *encSizeBytes = sizeof( silk_encoder ); |
67 | |
|
68 | 0 | return ret; |
69 | 0 | } |
70 | | |
71 | | /*************************/ |
72 | | /* Init or Reset encoder */ |
73 | | /*************************/ |
74 | | opus_int silk_InitEncoder( /* O Returns error code */ |
75 | | void *encState, /* I/O State */ |
76 | | int arch, /* I Run-time architecture */ |
77 | | silk_EncControlStruct *encStatus /* O Encoder Status */ |
78 | | ) |
79 | 0 | { |
80 | 0 | silk_encoder *psEnc; |
81 | 0 | opus_int n, ret = SILK_NO_ERROR; |
82 | |
|
83 | 0 | psEnc = (silk_encoder *)encState; |
84 | | |
85 | | /* Reset encoder */ |
86 | 0 | silk_memset( psEnc, 0, sizeof( silk_encoder ) ); |
87 | 0 | for( n = 0; n < ENCODER_NUM_CHANNELS; n++ ) { |
88 | 0 | if( ret += silk_init_encoder( &psEnc->state_Fxx[ n ], arch ) ) { |
89 | 0 | celt_assert( 0 ); |
90 | 0 | } |
91 | 0 | } |
92 | | |
93 | 0 | psEnc->nChannelsAPI = 1; |
94 | 0 | psEnc->nChannelsInternal = 1; |
95 | | |
96 | | /* Read control structure */ |
97 | 0 | if( ret += silk_QueryEncoder( encState, encStatus ) ) { |
98 | 0 | celt_assert( 0 ); |
99 | 0 | } |
100 | | |
101 | 0 | return ret; |
102 | 0 | } |
103 | | |
104 | | /***************************************/ |
105 | | /* Read control structure from encoder */ |
106 | | /***************************************/ |
107 | | static opus_int silk_QueryEncoder( /* O Returns error code */ |
108 | | const void *encState, /* I State */ |
109 | | silk_EncControlStruct *encStatus /* O Encoder Status */ |
110 | | ) |
111 | 0 | { |
112 | 0 | opus_int ret = SILK_NO_ERROR; |
113 | 0 | silk_encoder_state_Fxx *state_Fxx; |
114 | 0 | silk_encoder *psEnc = (silk_encoder *)encState; |
115 | |
|
116 | 0 | state_Fxx = psEnc->state_Fxx; |
117 | |
|
118 | 0 | encStatus->nChannelsAPI = psEnc->nChannelsAPI; |
119 | 0 | encStatus->nChannelsInternal = psEnc->nChannelsInternal; |
120 | 0 | encStatus->API_sampleRate = state_Fxx[ 0 ].sCmn.API_fs_Hz; |
121 | 0 | encStatus->maxInternalSampleRate = state_Fxx[ 0 ].sCmn.maxInternal_fs_Hz; |
122 | 0 | encStatus->minInternalSampleRate = state_Fxx[ 0 ].sCmn.minInternal_fs_Hz; |
123 | 0 | encStatus->desiredInternalSampleRate = state_Fxx[ 0 ].sCmn.desiredInternal_fs_Hz; |
124 | 0 | encStatus->payloadSize_ms = state_Fxx[ 0 ].sCmn.PacketSize_ms; |
125 | 0 | encStatus->bitRate = state_Fxx[ 0 ].sCmn.TargetRate_bps; |
126 | 0 | encStatus->packetLossPercentage = state_Fxx[ 0 ].sCmn.PacketLoss_perc; |
127 | 0 | encStatus->complexity = state_Fxx[ 0 ].sCmn.Complexity; |
128 | 0 | encStatus->useInBandFEC = state_Fxx[ 0 ].sCmn.useInBandFEC; |
129 | 0 | encStatus->useDTX = state_Fxx[ 0 ].sCmn.useDTX; |
130 | 0 | encStatus->useCBR = state_Fxx[ 0 ].sCmn.useCBR; |
131 | 0 | encStatus->internalSampleRate = silk_SMULBB( state_Fxx[ 0 ].sCmn.fs_kHz, 1000 ); |
132 | 0 | encStatus->allowBandwidthSwitch = state_Fxx[ 0 ].sCmn.allow_bandwidth_switch; |
133 | 0 | encStatus->inWBmodeWithoutVariableLP = state_Fxx[ 0 ].sCmn.fs_kHz == 16 && state_Fxx[ 0 ].sCmn.sLP.mode == 0; |
134 | |
|
135 | 0 | return ret; |
136 | 0 | } |
137 | | |
138 | | |
139 | | /**************************/ |
140 | | /* Encode frame with Silk */ |
141 | | /**************************/ |
142 | | /* Note: if prefillFlag is set, the input must contain 10 ms of audio, irrespective of what */ |
143 | | /* encControl->payloadSize_ms is set to */ |
144 | | opus_int silk_Encode( /* O Returns error code */ |
145 | | void *encState, /* I/O State */ |
146 | | silk_EncControlStruct *encControl, /* I Control status */ |
147 | | const opus_int16 *samplesIn, /* I Speech sample input vector */ |
148 | | opus_int nSamplesIn, /* I Number of samples in input vector */ |
149 | | ec_enc *psRangeEnc, /* I/O Compressor data structure */ |
150 | | opus_int32 *nBytesOut, /* I/O Number of bytes in payload (input: Max bytes) */ |
151 | | const opus_int prefillFlag, /* I Flag to indicate prefilling buffers no coding */ |
152 | | opus_int activity /* I Decision of Opus voice activity detector */ |
153 | | ) |
154 | 0 | { |
155 | 0 | opus_int n, i, nBits, flags, tmp_payloadSize_ms = 0, tmp_complexity = 0, ret = 0; |
156 | 0 | opus_int nSamplesToBuffer, nSamplesToBufferMax, nBlocksOf10ms; |
157 | 0 | opus_int nSamplesFromInput = 0, nSamplesFromInputMax; |
158 | 0 | opus_int speech_act_thr_for_switch_Q8; |
159 | 0 | opus_int32 TargetRate_bps, MStargetRates_bps[ 2 ], channelRate_bps, LBRR_symbol, sum; |
160 | 0 | silk_encoder *psEnc = ( silk_encoder * )encState; |
161 | 0 | VARDECL( opus_int16, buf ); |
162 | 0 | opus_int transition, curr_block, tot_blocks; |
163 | 0 | SAVE_STACK; |
164 | |
|
165 | 0 | if (encControl->reducedDependency) |
166 | 0 | { |
167 | 0 | psEnc->state_Fxx[0].sCmn.first_frame_after_reset = 1; |
168 | 0 | psEnc->state_Fxx[1].sCmn.first_frame_after_reset = 1; |
169 | 0 | } |
170 | 0 | psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded = psEnc->state_Fxx[ 1 ].sCmn.nFramesEncoded = 0; |
171 | | |
172 | | /* Check values in encoder control structure */ |
173 | 0 | if( ( ret = check_control_input( encControl ) ) != 0 ) { |
174 | 0 | celt_assert( 0 ); |
175 | 0 | RESTORE_STACK; |
176 | 0 | return ret; |
177 | 0 | } |
178 | | |
179 | 0 | encControl->switchReady = 0; |
180 | |
|
181 | 0 | if( encControl->nChannelsInternal > psEnc->nChannelsInternal ) { |
182 | | /* Mono -> Stereo transition: init state of second channel and stereo state */ |
183 | 0 | ret += silk_init_encoder( &psEnc->state_Fxx[ 1 ], psEnc->state_Fxx[ 0 ].sCmn.arch ); |
184 | 0 | silk_memset( psEnc->sStereo.pred_prev_Q13, 0, sizeof( psEnc->sStereo.pred_prev_Q13 ) ); |
185 | 0 | silk_memset( psEnc->sStereo.sSide, 0, sizeof( psEnc->sStereo.sSide ) ); |
186 | 0 | psEnc->sStereo.mid_side_amp_Q0[ 0 ] = 0; |
187 | 0 | psEnc->sStereo.mid_side_amp_Q0[ 1 ] = 1; |
188 | 0 | psEnc->sStereo.mid_side_amp_Q0[ 2 ] = 0; |
189 | 0 | psEnc->sStereo.mid_side_amp_Q0[ 3 ] = 1; |
190 | 0 | psEnc->sStereo.width_prev_Q14 = 0; |
191 | 0 | psEnc->sStereo.smth_width_Q14 = SILK_FIX_CONST( 1, 14 ); |
192 | 0 | if( psEnc->nChannelsAPI == 2 ) { |
193 | 0 | silk_memcpy( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state, &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, sizeof( silk_resampler_state_struct ) ); |
194 | 0 | 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 ) ); |
195 | 0 | } |
196 | 0 | } |
197 | |
|
198 | 0 | transition = (encControl->payloadSize_ms != psEnc->state_Fxx[ 0 ].sCmn.PacketSize_ms) || (psEnc->nChannelsInternal != encControl->nChannelsInternal); |
199 | |
|
200 | 0 | psEnc->nChannelsAPI = encControl->nChannelsAPI; |
201 | 0 | psEnc->nChannelsInternal = encControl->nChannelsInternal; |
202 | |
|
203 | 0 | nBlocksOf10ms = silk_DIV32( 100 * nSamplesIn, encControl->API_sampleRate ); |
204 | 0 | tot_blocks = ( nBlocksOf10ms > 1 ) ? nBlocksOf10ms >> 1 : 1; |
205 | 0 | curr_block = 0; |
206 | 0 | if( prefillFlag ) { |
207 | 0 | silk_LP_state save_LP; |
208 | | /* Only accept input length of 10 ms */ |
209 | 0 | if( nBlocksOf10ms != 1 ) { |
210 | 0 | celt_assert( 0 ); |
211 | 0 | RESTORE_STACK; |
212 | 0 | return SILK_ENC_INPUT_INVALID_NO_OF_SAMPLES; |
213 | 0 | } |
214 | 0 | if ( prefillFlag == 2 ) { |
215 | 0 | save_LP = psEnc->state_Fxx[ 0 ].sCmn.sLP; |
216 | | /* Save the sampling rate so the bandwidth switching code can keep handling transitions. */ |
217 | 0 | save_LP.saved_fs_kHz = psEnc->state_Fxx[ 0 ].sCmn.fs_kHz; |
218 | 0 | } |
219 | | /* Reset Encoder */ |
220 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
221 | 0 | ret = silk_init_encoder( &psEnc->state_Fxx[ n ], psEnc->state_Fxx[ n ].sCmn.arch ); |
222 | | /* Restore the variable LP state. */ |
223 | 0 | if ( prefillFlag == 2 ) { |
224 | 0 | psEnc->state_Fxx[ n ].sCmn.sLP = save_LP; |
225 | 0 | } |
226 | 0 | celt_assert( !ret ); |
227 | 0 | } |
228 | 0 | tmp_payloadSize_ms = encControl->payloadSize_ms; |
229 | 0 | encControl->payloadSize_ms = 10; |
230 | 0 | tmp_complexity = encControl->complexity; |
231 | 0 | encControl->complexity = 0; |
232 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
233 | 0 | psEnc->state_Fxx[ n ].sCmn.controlled_since_last_payload = 0; |
234 | 0 | psEnc->state_Fxx[ n ].sCmn.prefillFlag = 1; |
235 | 0 | } |
236 | 0 | } else { |
237 | | /* Only accept input lengths that are a multiple of 10 ms */ |
238 | 0 | if( nBlocksOf10ms * encControl->API_sampleRate != 100 * nSamplesIn || nSamplesIn < 0 ) { |
239 | 0 | celt_assert( 0 ); |
240 | 0 | RESTORE_STACK; |
241 | 0 | return SILK_ENC_INPUT_INVALID_NO_OF_SAMPLES; |
242 | 0 | } |
243 | | /* Make sure no more than one packet can be produced */ |
244 | 0 | if( 1000 * (opus_int32)nSamplesIn > encControl->payloadSize_ms * encControl->API_sampleRate ) { |
245 | 0 | celt_assert( 0 ); |
246 | 0 | RESTORE_STACK; |
247 | 0 | return SILK_ENC_INPUT_INVALID_NO_OF_SAMPLES; |
248 | 0 | } |
249 | 0 | } |
250 | | |
251 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
252 | | /* Force the side channel to the same rate as the mid */ |
253 | 0 | opus_int force_fs_kHz = (n==1) ? psEnc->state_Fxx[0].sCmn.fs_kHz : 0; |
254 | 0 | if( ( ret = silk_control_encoder( &psEnc->state_Fxx[ n ], encControl, psEnc->allowBandwidthSwitch, n, force_fs_kHz ) ) != 0 ) { |
255 | 0 | silk_assert( 0 ); |
256 | 0 | RESTORE_STACK; |
257 | 0 | return ret; |
258 | 0 | } |
259 | 0 | if( psEnc->state_Fxx[n].sCmn.first_frame_after_reset || transition ) { |
260 | 0 | for( i = 0; i < psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket; i++ ) { |
261 | 0 | psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i ] = 0; |
262 | 0 | } |
263 | 0 | } |
264 | 0 | psEnc->state_Fxx[ n ].sCmn.inDTX = psEnc->state_Fxx[ n ].sCmn.useDTX; |
265 | 0 | } |
266 | 0 | celt_assert( encControl->nChannelsInternal == 1 || psEnc->state_Fxx[ 0 ].sCmn.fs_kHz == psEnc->state_Fxx[ 1 ].sCmn.fs_kHz ); |
267 | | |
268 | | /* Input buffering/resampling and encoding */ |
269 | 0 | nSamplesToBufferMax = |
270 | 0 | 10 * nBlocksOf10ms * psEnc->state_Fxx[ 0 ].sCmn.fs_kHz; |
271 | 0 | nSamplesFromInputMax = |
272 | 0 | silk_DIV32_16( nSamplesToBufferMax * |
273 | 0 | psEnc->state_Fxx[ 0 ].sCmn.API_fs_Hz, |
274 | 0 | psEnc->state_Fxx[ 0 ].sCmn.fs_kHz * 1000 ); |
275 | 0 | ALLOC( buf, nSamplesFromInputMax, opus_int16 ); |
276 | 0 | while( 1 ) { |
277 | 0 | int curr_nBitsUsedLBRR = 0; |
278 | 0 | nSamplesToBuffer = psEnc->state_Fxx[ 0 ].sCmn.frame_length - psEnc->state_Fxx[ 0 ].sCmn.inputBufIx; |
279 | 0 | nSamplesToBuffer = silk_min( nSamplesToBuffer, nSamplesToBufferMax ); |
280 | 0 | nSamplesFromInput = silk_DIV32_16( nSamplesToBuffer * psEnc->state_Fxx[ 0 ].sCmn.API_fs_Hz, psEnc->state_Fxx[ 0 ].sCmn.fs_kHz * 1000 ); |
281 | | /* Resample and write to buffer */ |
282 | 0 | if( encControl->nChannelsAPI == 2 && encControl->nChannelsInternal == 2 ) { |
283 | 0 | opus_int id = psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded; |
284 | 0 | for( n = 0; n < nSamplesFromInput; n++ ) { |
285 | 0 | buf[ n ] = samplesIn[ 2 * n ]; |
286 | 0 | } |
287 | | /* Making sure to start both resamplers from the same state when switching from mono to stereo */ |
288 | 0 | if( psEnc->nPrevChannelsInternal == 1 && id==0 ) { |
289 | 0 | silk_memcpy( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state, &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, sizeof(psEnc->state_Fxx[ 1 ].sCmn.resampler_state)); |
290 | 0 | } |
291 | |
|
292 | 0 | ret += silk_resampler( &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, |
293 | 0 | &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput ); |
294 | 0 | psEnc->state_Fxx[ 0 ].sCmn.inputBufIx += nSamplesToBuffer; |
295 | |
|
296 | 0 | nSamplesToBuffer = psEnc->state_Fxx[ 1 ].sCmn.frame_length - psEnc->state_Fxx[ 1 ].sCmn.inputBufIx; |
297 | 0 | nSamplesToBuffer = silk_min( nSamplesToBuffer, 10 * nBlocksOf10ms * psEnc->state_Fxx[ 1 ].sCmn.fs_kHz ); |
298 | 0 | for( n = 0; n < nSamplesFromInput; n++ ) { |
299 | 0 | buf[ n ] = samplesIn[ 2 * n + 1 ]; |
300 | 0 | } |
301 | 0 | ret += silk_resampler( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state, |
302 | 0 | &psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ psEnc->state_Fxx[ 1 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput ); |
303 | |
|
304 | 0 | psEnc->state_Fxx[ 1 ].sCmn.inputBufIx += nSamplesToBuffer; |
305 | 0 | } else if( encControl->nChannelsAPI == 2 && encControl->nChannelsInternal == 1 ) { |
306 | | /* Combine left and right channels before resampling */ |
307 | 0 | for( n = 0; n < nSamplesFromInput; n++ ) { |
308 | 0 | sum = samplesIn[ 2 * n ] + samplesIn[ 2 * n + 1 ]; |
309 | 0 | buf[ n ] = (opus_int16)silk_RSHIFT_ROUND( sum, 1 ); |
310 | 0 | } |
311 | 0 | ret += silk_resampler( &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, |
312 | 0 | &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput ); |
313 | | /* On the first mono frame, average the results for the two resampler states */ |
314 | 0 | if( psEnc->nPrevChannelsInternal == 2 && psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded == 0 ) { |
315 | 0 | ret += silk_resampler( &psEnc->state_Fxx[ 1 ].sCmn.resampler_state, |
316 | 0 | &psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ psEnc->state_Fxx[ 1 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput ); |
317 | 0 | for( n = 0; n < psEnc->state_Fxx[ 0 ].sCmn.frame_length; n++ ) { |
318 | 0 | psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx+n+2 ] = |
319 | 0 | silk_RSHIFT(psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx+n+2 ] |
320 | 0 | + psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ psEnc->state_Fxx[ 1 ].sCmn.inputBufIx+n+2 ], 1); |
321 | 0 | } |
322 | 0 | } |
323 | 0 | psEnc->state_Fxx[ 0 ].sCmn.inputBufIx += nSamplesToBuffer; |
324 | 0 | } else { |
325 | 0 | celt_assert( encControl->nChannelsAPI == 1 && encControl->nChannelsInternal == 1 ); |
326 | 0 | silk_memcpy(buf, samplesIn, nSamplesFromInput*sizeof(opus_int16)); |
327 | 0 | ret += silk_resampler( &psEnc->state_Fxx[ 0 ].sCmn.resampler_state, |
328 | 0 | &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.inputBufIx + 2 ], buf, nSamplesFromInput ); |
329 | 0 | psEnc->state_Fxx[ 0 ].sCmn.inputBufIx += nSamplesToBuffer; |
330 | 0 | } |
331 | | |
332 | 0 | samplesIn += nSamplesFromInput * encControl->nChannelsAPI; |
333 | 0 | nSamplesIn -= nSamplesFromInput; |
334 | | |
335 | | /* Default */ |
336 | 0 | psEnc->allowBandwidthSwitch = 0; |
337 | | |
338 | | /* Silk encoder */ |
339 | 0 | if( psEnc->state_Fxx[ 0 ].sCmn.inputBufIx >= psEnc->state_Fxx[ 0 ].sCmn.frame_length ) { |
340 | | /* Enough data in input buffer, so encode */ |
341 | 0 | celt_assert( psEnc->state_Fxx[ 0 ].sCmn.inputBufIx == psEnc->state_Fxx[ 0 ].sCmn.frame_length ); |
342 | 0 | celt_assert( encControl->nChannelsInternal == 1 || psEnc->state_Fxx[ 1 ].sCmn.inputBufIx == psEnc->state_Fxx[ 1 ].sCmn.frame_length ); |
343 | | |
344 | | /* Deal with LBRR data */ |
345 | 0 | if( psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded == 0 && !prefillFlag ) { |
346 | | /* Create space at start of payload for VAD and FEC flags */ |
347 | 0 | opus_uint8 iCDF[ 2 ] = { 0, 0 }; |
348 | 0 | iCDF[ 0 ] = 256 - silk_RSHIFT( 256, ( psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket + 1 ) * encControl->nChannelsInternal ); |
349 | 0 | ec_enc_icdf( psRangeEnc, 0, iCDF, 8 ); |
350 | 0 | curr_nBitsUsedLBRR = ec_tell( psRangeEnc ); |
351 | | |
352 | | /* Encode any LBRR data from previous packet */ |
353 | | /* Encode LBRR flags */ |
354 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
355 | 0 | LBRR_symbol = 0; |
356 | 0 | for( i = 0; i < psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket; i++ ) { |
357 | 0 | LBRR_symbol |= silk_LSHIFT( psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i ], i ); |
358 | 0 | } |
359 | 0 | psEnc->state_Fxx[ n ].sCmn.LBRR_flag = LBRR_symbol > 0 ? 1 : 0; |
360 | 0 | if( LBRR_symbol && psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket > 1 ) { |
361 | 0 | ec_enc_icdf( psRangeEnc, LBRR_symbol - 1, silk_LBRR_flags_iCDF_ptr[ psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket - 2 ], 8 ); |
362 | 0 | } |
363 | 0 | } |
364 | | |
365 | | /* Code LBRR indices and excitation signals */ |
366 | 0 | for( i = 0; i < psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket; i++ ) { |
367 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
368 | 0 | if( psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i ] ) { |
369 | 0 | opus_int condCoding; |
370 | |
|
371 | 0 | if( encControl->nChannelsInternal == 2 && n == 0 ) { |
372 | 0 | silk_stereo_encode_pred( psRangeEnc, psEnc->sStereo.predIx[ i ] ); |
373 | | /* For LBRR data there's no need to code the mid-only flag if the side-channel LBRR flag is set */ |
374 | 0 | if( psEnc->state_Fxx[ 1 ].sCmn.LBRR_flags[ i ] == 0 ) { |
375 | 0 | silk_stereo_encode_mid_only( psRangeEnc, psEnc->sStereo.mid_only_flags[ i ] ); |
376 | 0 | } |
377 | 0 | } |
378 | | /* Use conditional coding if previous frame available */ |
379 | 0 | if( i > 0 && psEnc->state_Fxx[ n ].sCmn.LBRR_flags[ i - 1 ] ) { |
380 | 0 | condCoding = CODE_CONDITIONALLY; |
381 | 0 | } else { |
382 | 0 | condCoding = CODE_INDEPENDENTLY; |
383 | 0 | } |
384 | 0 | silk_encode_indices( &psEnc->state_Fxx[ n ].sCmn, psRangeEnc, i, 1, condCoding ); |
385 | 0 | silk_encode_pulses( psRangeEnc, psEnc->state_Fxx[ n ].sCmn.indices_LBRR[i].signalType, psEnc->state_Fxx[ n ].sCmn.indices_LBRR[i].quantOffsetType, |
386 | 0 | psEnc->state_Fxx[ n ].sCmn.pulses_LBRR[ i ], psEnc->state_Fxx[ n ].sCmn.frame_length ); |
387 | 0 | } |
388 | 0 | } |
389 | 0 | } |
390 | | |
391 | | /* Reset LBRR flags */ |
392 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
393 | 0 | silk_memset( psEnc->state_Fxx[ n ].sCmn.LBRR_flags, 0, sizeof( psEnc->state_Fxx[ n ].sCmn.LBRR_flags ) ); |
394 | 0 | } |
395 | 0 | curr_nBitsUsedLBRR = ec_tell( psRangeEnc ) - curr_nBitsUsedLBRR; |
396 | 0 | } |
397 | |
|
398 | 0 | silk_HP_variable_cutoff( psEnc->state_Fxx ); |
399 | | |
400 | | /* Total target bits for packet */ |
401 | 0 | nBits = silk_DIV32_16( silk_MUL( encControl->bitRate, encControl->payloadSize_ms ), 1000 ); |
402 | | /* Subtract bits used for LBRR */ |
403 | 0 | if( !prefillFlag ) { |
404 | | /* psEnc->nBitsUsedLBRR is an exponential moving average of the LBRR usage, |
405 | | except that for the first LBRR frame it does no averaging and for the first |
406 | | frame after after LBRR, it goes back to zero immediately. */ |
407 | 0 | if ( curr_nBitsUsedLBRR < 10 ) { |
408 | 0 | psEnc->nBitsUsedLBRR = 0; |
409 | 0 | } else if ( psEnc->nBitsUsedLBRR < 10) { |
410 | 0 | psEnc->nBitsUsedLBRR = curr_nBitsUsedLBRR; |
411 | 0 | } else { |
412 | 0 | psEnc->nBitsUsedLBRR = ( psEnc->nBitsUsedLBRR + curr_nBitsUsedLBRR ) / 2; |
413 | 0 | } |
414 | 0 | nBits -= psEnc->nBitsUsedLBRR; |
415 | 0 | } |
416 | | /* Divide by number of uncoded frames left in packet */ |
417 | 0 | nBits = silk_DIV32_16( nBits, psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket ); |
418 | | /* Convert to bits/second */ |
419 | 0 | if( encControl->payloadSize_ms == 10 ) { |
420 | 0 | TargetRate_bps = silk_SMULBB( nBits, 100 ); |
421 | 0 | } else { |
422 | 0 | TargetRate_bps = silk_SMULBB( nBits, 50 ); |
423 | 0 | } |
424 | | /* Subtract fraction of bits in excess of target in previous frames and packets */ |
425 | 0 | TargetRate_bps -= silk_DIV32_16( silk_MUL( psEnc->nBitsExceeded, 1000 ), BITRESERVOIR_DECAY_TIME_MS ); |
426 | 0 | if( !prefillFlag && psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded > 0 ) { |
427 | | /* Compare actual vs target bits so far in this packet */ |
428 | 0 | opus_int32 bitsBalance = ec_tell( psRangeEnc ) - psEnc->nBitsUsedLBRR - nBits * psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded; |
429 | 0 | TargetRate_bps -= silk_DIV32_16( silk_MUL( bitsBalance, 1000 ), BITRESERVOIR_DECAY_TIME_MS ); |
430 | 0 | } |
431 | | /* Never exceed input bitrate */ |
432 | 0 | TargetRate_bps = silk_LIMIT( TargetRate_bps, encControl->bitRate, 5000 ); |
433 | | |
434 | | /* Convert Left/Right to Mid/Side */ |
435 | 0 | if( encControl->nChannelsInternal == 2 ) { |
436 | 0 | silk_stereo_LR_to_MS( &psEnc->sStereo, &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ 2 ], &psEnc->state_Fxx[ 1 ].sCmn.inputBuf[ 2 ], |
437 | 0 | psEnc->sStereo.predIx[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ], &psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ], |
438 | 0 | MStargetRates_bps, TargetRate_bps, psEnc->state_Fxx[ 0 ].sCmn.speech_activity_Q8, encControl->toMono, |
439 | 0 | psEnc->state_Fxx[ 0 ].sCmn.fs_kHz, psEnc->state_Fxx[ 0 ].sCmn.frame_length ); |
440 | 0 | if( psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] == 0 ) { |
441 | | /* Reset side channel encoder memory for first frame with side coding */ |
442 | 0 | if( psEnc->prev_decode_only_middle == 1 ) { |
443 | 0 | silk_memset( &psEnc->state_Fxx[ 1 ].sShape, 0, sizeof( psEnc->state_Fxx[ 1 ].sShape ) ); |
444 | 0 | silk_memset( &psEnc->state_Fxx[ 1 ].sCmn.sNSQ, 0, sizeof( psEnc->state_Fxx[ 1 ].sCmn.sNSQ ) ); |
445 | 0 | silk_memset( psEnc->state_Fxx[ 1 ].sCmn.prev_NLSFq_Q15, 0, sizeof( psEnc->state_Fxx[ 1 ].sCmn.prev_NLSFq_Q15 ) ); |
446 | 0 | silk_memset( &psEnc->state_Fxx[ 1 ].sCmn.sLP.In_LP_State, 0, sizeof( psEnc->state_Fxx[ 1 ].sCmn.sLP.In_LP_State ) ); |
447 | 0 | psEnc->state_Fxx[ 1 ].sCmn.prevLag = 100; |
448 | 0 | psEnc->state_Fxx[ 1 ].sCmn.sNSQ.lagPrev = 100; |
449 | 0 | psEnc->state_Fxx[ 1 ].sShape.LastGainIndex = 10; |
450 | 0 | psEnc->state_Fxx[ 1 ].sCmn.prevSignalType = TYPE_NO_VOICE_ACTIVITY; |
451 | 0 | psEnc->state_Fxx[ 1 ].sCmn.sNSQ.prev_gain_Q16 = 65536; |
452 | 0 | psEnc->state_Fxx[ 1 ].sCmn.first_frame_after_reset = 1; |
453 | 0 | } |
454 | 0 | silk_encode_do_VAD_Fxx( &psEnc->state_Fxx[ 1 ], activity ); |
455 | 0 | } else { |
456 | 0 | psEnc->state_Fxx[ 1 ].sCmn.VAD_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] = 0; |
457 | 0 | } |
458 | 0 | if( !prefillFlag ) { |
459 | 0 | silk_stereo_encode_pred( psRangeEnc, psEnc->sStereo.predIx[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] ); |
460 | 0 | if( psEnc->state_Fxx[ 1 ].sCmn.VAD_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] == 0 ) { |
461 | 0 | silk_stereo_encode_mid_only( psRangeEnc, psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded ] ); |
462 | 0 | } |
463 | 0 | } |
464 | 0 | } else { |
465 | | /* Buffering */ |
466 | 0 | silk_memcpy( psEnc->state_Fxx[ 0 ].sCmn.inputBuf, psEnc->sStereo.sMid, 2 * sizeof( opus_int16 ) ); |
467 | 0 | silk_memcpy( psEnc->sStereo.sMid, &psEnc->state_Fxx[ 0 ].sCmn.inputBuf[ psEnc->state_Fxx[ 0 ].sCmn.frame_length ], 2 * sizeof( opus_int16 ) ); |
468 | 0 | } |
469 | 0 | silk_encode_do_VAD_Fxx( &psEnc->state_Fxx[ 0 ], activity ); |
470 | | |
471 | | /* Encode */ |
472 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
473 | 0 | opus_int maxBits, useCBR; |
474 | | |
475 | | /* Handling rate constraints */ |
476 | 0 | maxBits = encControl->maxBits; |
477 | 0 | if( tot_blocks == 2 && curr_block == 0 ) { |
478 | 0 | maxBits = maxBits * 3 / 5; |
479 | 0 | } else if( tot_blocks == 3 ) { |
480 | 0 | if( curr_block == 0 ) { |
481 | 0 | maxBits = maxBits * 2 / 5; |
482 | 0 | } else if( curr_block == 1 ) { |
483 | 0 | maxBits = maxBits * 3 / 4; |
484 | 0 | } |
485 | 0 | } |
486 | 0 | useCBR = encControl->useCBR && curr_block == tot_blocks - 1; |
487 | |
|
488 | 0 | if( encControl->nChannelsInternal == 1 ) { |
489 | 0 | channelRate_bps = TargetRate_bps; |
490 | 0 | } else { |
491 | 0 | channelRate_bps = MStargetRates_bps[ n ]; |
492 | 0 | if( n == 0 && MStargetRates_bps[ 1 ] > 0 ) { |
493 | 0 | useCBR = 0; |
494 | | /* Give mid up to 1/2 of the max bits for that frame */ |
495 | 0 | maxBits -= encControl->maxBits / ( tot_blocks * 2 ); |
496 | 0 | } |
497 | 0 | } |
498 | |
|
499 | 0 | if( channelRate_bps > 0 ) { |
500 | 0 | opus_int condCoding; |
501 | |
|
502 | 0 | silk_control_SNR( &psEnc->state_Fxx[ n ].sCmn, channelRate_bps ); |
503 | | |
504 | | /* Use independent coding if no previous frame available */ |
505 | 0 | if( psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded - n <= 0 ) { |
506 | 0 | condCoding = CODE_INDEPENDENTLY; |
507 | 0 | } else if( n > 0 && psEnc->prev_decode_only_middle ) { |
508 | | /* If we skipped a side frame in this packet, we don't |
509 | | need LTP scaling; the LTP state is well-defined. */ |
510 | 0 | condCoding = CODE_INDEPENDENTLY_NO_LTP_SCALING; |
511 | 0 | } else { |
512 | 0 | condCoding = CODE_CONDITIONALLY; |
513 | 0 | } |
514 | 0 | if( ( ret = silk_encode_frame_Fxx( &psEnc->state_Fxx[ n ], nBytesOut, psRangeEnc, condCoding, maxBits, useCBR ) ) != 0 ) { |
515 | 0 | silk_assert( 0 ); |
516 | 0 | } |
517 | 0 | } |
518 | 0 | psEnc->state_Fxx[ n ].sCmn.controlled_since_last_payload = 0; |
519 | 0 | psEnc->state_Fxx[ n ].sCmn.inputBufIx = 0; |
520 | 0 | psEnc->state_Fxx[ n ].sCmn.nFramesEncoded++; |
521 | 0 | } |
522 | 0 | psEnc->prev_decode_only_middle = psEnc->sStereo.mid_only_flags[ psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded - 1 ]; |
523 | | |
524 | | /* Insert VAD and FEC flags at beginning of bitstream */ |
525 | 0 | if( *nBytesOut > 0 && psEnc->state_Fxx[ 0 ].sCmn.nFramesEncoded == psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket) { |
526 | 0 | flags = 0; |
527 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
528 | 0 | for( i = 0; i < psEnc->state_Fxx[ n ].sCmn.nFramesPerPacket; i++ ) { |
529 | 0 | flags = silk_LSHIFT( flags, 1 ); |
530 | 0 | flags |= psEnc->state_Fxx[ n ].sCmn.VAD_flags[ i ]; |
531 | 0 | } |
532 | 0 | flags = silk_LSHIFT( flags, 1 ); |
533 | 0 | flags |= psEnc->state_Fxx[ n ].sCmn.LBRR_flag; |
534 | 0 | } |
535 | 0 | if( !prefillFlag ) { |
536 | 0 | ec_enc_patch_initial_bits( psRangeEnc, flags, ( psEnc->state_Fxx[ 0 ].sCmn.nFramesPerPacket + 1 ) * encControl->nChannelsInternal ); |
537 | 0 | } |
538 | | |
539 | | /* Return zero bytes if all channels DTXed */ |
540 | 0 | if( psEnc->state_Fxx[ 0 ].sCmn.inDTX && ( encControl->nChannelsInternal == 1 || psEnc->state_Fxx[ 1 ].sCmn.inDTX ) ) { |
541 | 0 | *nBytesOut = 0; |
542 | 0 | } |
543 | |
|
544 | 0 | psEnc->nBitsExceeded += *nBytesOut * 8; |
545 | 0 | psEnc->nBitsExceeded -= silk_DIV32_16( silk_MUL( encControl->bitRate, encControl->payloadSize_ms ), 1000 ); |
546 | 0 | psEnc->nBitsExceeded = silk_LIMIT( psEnc->nBitsExceeded, 0, 10000 ); |
547 | | |
548 | | /* Update flag indicating if bandwidth switching is allowed */ |
549 | 0 | speech_act_thr_for_switch_Q8 = silk_SMLAWB( SILK_FIX_CONST( SPEECH_ACTIVITY_DTX_THRES, 8 ), |
550 | 0 | SILK_FIX_CONST( ( 1 - SPEECH_ACTIVITY_DTX_THRES ) / MAX_BANDWIDTH_SWITCH_DELAY_MS, 16 + 8 ), psEnc->timeSinceSwitchAllowed_ms ); |
551 | 0 | if( psEnc->state_Fxx[ 0 ].sCmn.speech_activity_Q8 < speech_act_thr_for_switch_Q8 ) { |
552 | 0 | psEnc->allowBandwidthSwitch = 1; |
553 | 0 | psEnc->timeSinceSwitchAllowed_ms = 0; |
554 | 0 | } else { |
555 | 0 | psEnc->allowBandwidthSwitch = 0; |
556 | 0 | psEnc->timeSinceSwitchAllowed_ms += encControl->payloadSize_ms; |
557 | 0 | } |
558 | 0 | } |
559 | |
|
560 | 0 | if( nSamplesIn == 0 ) { |
561 | 0 | break; |
562 | 0 | } |
563 | 0 | } else { |
564 | 0 | break; |
565 | 0 | } |
566 | 0 | curr_block++; |
567 | 0 | } |
568 | | |
569 | 0 | psEnc->nPrevChannelsInternal = encControl->nChannelsInternal; |
570 | |
|
571 | 0 | encControl->allowBandwidthSwitch = psEnc->allowBandwidthSwitch; |
572 | 0 | encControl->inWBmodeWithoutVariableLP = psEnc->state_Fxx[ 0 ].sCmn.fs_kHz == 16 && psEnc->state_Fxx[ 0 ].sCmn.sLP.mode == 0; |
573 | 0 | encControl->internalSampleRate = silk_SMULBB( psEnc->state_Fxx[ 0 ].sCmn.fs_kHz, 1000 ); |
574 | 0 | encControl->stereoWidth_Q14 = encControl->toMono ? 0 : psEnc->sStereo.smth_width_Q14; |
575 | 0 | if( prefillFlag ) { |
576 | 0 | encControl->payloadSize_ms = tmp_payloadSize_ms; |
577 | 0 | encControl->complexity = tmp_complexity; |
578 | 0 | for( n = 0; n < encControl->nChannelsInternal; n++ ) { |
579 | 0 | psEnc->state_Fxx[ n ].sCmn.controlled_since_last_payload = 0; |
580 | 0 | psEnc->state_Fxx[ n ].sCmn.prefillFlag = 0; |
581 | 0 | } |
582 | 0 | } |
583 | |
|
584 | 0 | encControl->signalType = psEnc->state_Fxx[0].sCmn.indices.signalType; |
585 | 0 | encControl->offset = silk_Quantization_Offsets_Q10 |
586 | 0 | [ psEnc->state_Fxx[0].sCmn.indices.signalType >> 1 ] |
587 | 0 | [ psEnc->state_Fxx[0].sCmn.indices.quantOffsetType ]; |
588 | 0 | RESTORE_STACK; |
589 | 0 | return ret; |
590 | 0 | } |
591 | | |