Chris@69: /*********************************************************************** Chris@69: Copyright (c) 2006-2011, Skype Limited. All rights reserved. Chris@69: Redistribution and use in source and binary forms, with or without Chris@69: modification, are permitted provided that the following conditions Chris@69: are met: Chris@69: - Redistributions of source code must retain the above copyright notice, Chris@69: this list of conditions and the following disclaimer. Chris@69: - Redistributions in binary form must reproduce the above copyright Chris@69: notice, this list of conditions and the following disclaimer in the Chris@69: documentation and/or other materials provided with the distribution. Chris@69: - Neither the name of Internet Society, IETF or IETF Trust, nor the Chris@69: names of specific contributors, may be used to endorse or promote Chris@69: products derived from this software without specific prior written Chris@69: permission. Chris@69: THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" Chris@69: AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE Chris@69: IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE Chris@69: ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE Chris@69: LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR Chris@69: CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF Chris@69: SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS Chris@69: INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN Chris@69: CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) Chris@69: ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE Chris@69: POSSIBILITY OF SUCH DAMAGE. Chris@69: ***********************************************************************/ Chris@69: Chris@69: #ifdef HAVE_CONFIG_H Chris@69: #include "config.h" Chris@69: #endif Chris@69: Chris@69: #include "main.h" Chris@69: #include "stack_alloc.h" Chris@69: Chris@69: /* Convert Left/Right stereo signal to adaptive Mid/Side representation */ Chris@69: void silk_stereo_LR_to_MS( Chris@69: stereo_enc_state *state, /* I/O State */ Chris@69: opus_int16 x1[], /* I/O Left input signal, becomes mid signal */ Chris@69: opus_int16 x2[], /* I/O Right input signal, becomes side signal */ Chris@69: opus_int8 ix[ 2 ][ 3 ], /* O Quantization indices */ Chris@69: opus_int8 *mid_only_flag, /* O Flag: only mid signal coded */ Chris@69: opus_int32 mid_side_rates_bps[], /* O Bitrates for mid and side signals */ Chris@69: opus_int32 total_rate_bps, /* I Total bitrate */ Chris@69: opus_int prev_speech_act_Q8, /* I Speech activity level in previous frame */ Chris@69: opus_int toMono, /* I Last frame before a stereo->mono transition */ Chris@69: opus_int fs_kHz, /* I Sample rate (kHz) */ Chris@69: opus_int frame_length /* I Number of samples */ Chris@69: ) Chris@69: { Chris@69: opus_int n, is10msFrame, denom_Q16, delta0_Q13, delta1_Q13; Chris@69: opus_int32 sum, diff, smooth_coef_Q16, pred_Q13[ 2 ], pred0_Q13, pred1_Q13; Chris@69: opus_int32 LP_ratio_Q14, HP_ratio_Q14, frac_Q16, frac_3_Q16, min_mid_rate_bps, width_Q14, w_Q24, deltaw_Q24; Chris@69: VARDECL( opus_int16, side ); Chris@69: VARDECL( opus_int16, LP_mid ); Chris@69: VARDECL( opus_int16, HP_mid ); Chris@69: VARDECL( opus_int16, LP_side ); Chris@69: VARDECL( opus_int16, HP_side ); Chris@69: opus_int16 *mid = &x1[ -2 ]; Chris@69: SAVE_STACK; Chris@69: Chris@69: ALLOC( side, frame_length + 2, opus_int16 ); Chris@69: /* Convert to basic mid/side signals */ Chris@69: for( n = 0; n < frame_length + 2; n++ ) { Chris@69: sum = x1[ n - 2 ] + (opus_int32)x2[ n - 2 ]; Chris@69: diff = x1[ n - 2 ] - (opus_int32)x2[ n - 2 ]; Chris@69: mid[ n ] = (opus_int16)silk_RSHIFT_ROUND( sum, 1 ); Chris@69: side[ n ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( diff, 1 ) ); Chris@69: } Chris@69: Chris@69: /* Buffering */ Chris@69: silk_memcpy( mid, state->sMid, 2 * sizeof( opus_int16 ) ); Chris@69: silk_memcpy( side, state->sSide, 2 * sizeof( opus_int16 ) ); Chris@69: silk_memcpy( state->sMid, &mid[ frame_length ], 2 * sizeof( opus_int16 ) ); Chris@69: silk_memcpy( state->sSide, &side[ frame_length ], 2 * sizeof( opus_int16 ) ); Chris@69: Chris@69: /* LP and HP filter mid signal */ Chris@69: ALLOC( LP_mid, frame_length, opus_int16 ); Chris@69: ALLOC( HP_mid, frame_length, opus_int16 ); Chris@69: for( n = 0; n < frame_length; n++ ) { Chris@69: sum = silk_RSHIFT_ROUND( silk_ADD_LSHIFT( mid[ n ] + (opus_int32)mid[ n + 2 ], mid[ n + 1 ], 1 ), 2 ); Chris@69: LP_mid[ n ] = sum; Chris@69: HP_mid[ n ] = mid[ n + 1 ] - sum; Chris@69: } Chris@69: Chris@69: /* LP and HP filter side signal */ Chris@69: ALLOC( LP_side, frame_length, opus_int16 ); Chris@69: ALLOC( HP_side, frame_length, opus_int16 ); Chris@69: for( n = 0; n < frame_length; n++ ) { Chris@69: sum = silk_RSHIFT_ROUND( silk_ADD_LSHIFT( side[ n ] + (opus_int32)side[ n + 2 ], side[ n + 1 ], 1 ), 2 ); Chris@69: LP_side[ n ] = sum; Chris@69: HP_side[ n ] = side[ n + 1 ] - sum; Chris@69: } Chris@69: Chris@69: /* Find energies and predictors */ Chris@69: is10msFrame = frame_length == 10 * fs_kHz; Chris@69: smooth_coef_Q16 = is10msFrame ? Chris@69: SILK_FIX_CONST( STEREO_RATIO_SMOOTH_COEF / 2, 16 ) : Chris@69: SILK_FIX_CONST( STEREO_RATIO_SMOOTH_COEF, 16 ); Chris@69: smooth_coef_Q16 = silk_SMULWB( silk_SMULBB( prev_speech_act_Q8, prev_speech_act_Q8 ), smooth_coef_Q16 ); Chris@69: Chris@69: 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 ); Chris@69: 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 ); Chris@69: /* Ratio of the norms of residual and mid signals */ Chris@69: frac_Q16 = silk_SMLABB( HP_ratio_Q14, LP_ratio_Q14, 3 ); Chris@69: frac_Q16 = silk_min( frac_Q16, SILK_FIX_CONST( 1, 16 ) ); Chris@69: Chris@69: /* Determine bitrate distribution between mid and side, and possibly reduce stereo width */ Chris@69: total_rate_bps -= is10msFrame ? 1200 : 600; /* Subtract approximate bitrate for coding stereo parameters */ Chris@69: if( total_rate_bps < 1 ) { Chris@69: total_rate_bps = 1; Chris@69: } Chris@69: min_mid_rate_bps = silk_SMLABB( 2000, fs_kHz, 600 ); Chris@69: silk_assert( min_mid_rate_bps < 32767 ); Chris@69: /* Default bitrate distribution: 8 parts for Mid and (5+3*frac) parts for Side. so: mid_rate = ( 8 / ( 13 + 3 * frac ) ) * total_ rate */ Chris@69: frac_3_Q16 = silk_MUL( 3, frac_Q16 ); Chris@69: mid_side_rates_bps[ 0 ] = silk_DIV32_varQ( total_rate_bps, SILK_FIX_CONST( 8 + 5, 16 ) + frac_3_Q16, 16+3 ); Chris@69: /* If Mid bitrate below minimum, reduce stereo width */ Chris@69: if( mid_side_rates_bps[ 0 ] < min_mid_rate_bps ) { Chris@69: mid_side_rates_bps[ 0 ] = min_mid_rate_bps; Chris@69: mid_side_rates_bps[ 1 ] = total_rate_bps - mid_side_rates_bps[ 0 ]; Chris@69: /* width = 4 * ( 2 * side_rate - min_rate ) / ( ( 1 + 3 * frac ) * min_rate ) */ Chris@69: width_Q14 = silk_DIV32_varQ( silk_LSHIFT( mid_side_rates_bps[ 1 ], 1 ) - min_mid_rate_bps, Chris@69: silk_SMULWB( SILK_FIX_CONST( 1, 16 ) + frac_3_Q16, min_mid_rate_bps ), 14+2 ); Chris@69: width_Q14 = silk_LIMIT( width_Q14, 0, SILK_FIX_CONST( 1, 14 ) ); Chris@69: } else { Chris@69: mid_side_rates_bps[ 1 ] = total_rate_bps - mid_side_rates_bps[ 0 ]; Chris@69: width_Q14 = SILK_FIX_CONST( 1, 14 ); Chris@69: } Chris@69: Chris@69: /* Smoother */ Chris@69: state->smth_width_Q14 = (opus_int16)silk_SMLAWB( state->smth_width_Q14, width_Q14 - state->smth_width_Q14, smooth_coef_Q16 ); Chris@69: Chris@69: /* At very low bitrates or for inputs that are nearly amplitude panned, switch to panned-mono coding */ Chris@69: *mid_only_flag = 0; Chris@69: if( toMono ) { Chris@69: /* Last frame before stereo->mono transition; collapse stereo width */ Chris@69: width_Q14 = 0; Chris@69: pred_Q13[ 0 ] = 0; Chris@69: pred_Q13[ 1 ] = 0; Chris@69: silk_stereo_quant_pred( pred_Q13, ix ); Chris@69: } else if( state->width_prev_Q14 == 0 && Chris@69: ( 8 * total_rate_bps < 13 * min_mid_rate_bps || silk_SMULWB( frac_Q16, state->smth_width_Q14 ) < SILK_FIX_CONST( 0.05, 14 ) ) ) Chris@69: { Chris@69: /* Code as panned-mono; previous frame already had zero width */ Chris@69: /* Scale down and quantize predictors */ Chris@69: pred_Q13[ 0 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 0 ] ), 14 ); Chris@69: pred_Q13[ 1 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 1 ] ), 14 ); Chris@69: silk_stereo_quant_pred( pred_Q13, ix ); Chris@69: /* Collapse stereo width */ Chris@69: width_Q14 = 0; Chris@69: pred_Q13[ 0 ] = 0; Chris@69: pred_Q13[ 1 ] = 0; Chris@69: mid_side_rates_bps[ 0 ] = total_rate_bps; Chris@69: mid_side_rates_bps[ 1 ] = 0; Chris@69: *mid_only_flag = 1; Chris@69: } else if( state->width_prev_Q14 != 0 && Chris@69: ( 8 * total_rate_bps < 11 * min_mid_rate_bps || silk_SMULWB( frac_Q16, state->smth_width_Q14 ) < SILK_FIX_CONST( 0.02, 14 ) ) ) Chris@69: { Chris@69: /* Transition to zero-width stereo */ Chris@69: /* Scale down and quantize predictors */ Chris@69: pred_Q13[ 0 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 0 ] ), 14 ); Chris@69: pred_Q13[ 1 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 1 ] ), 14 ); Chris@69: silk_stereo_quant_pred( pred_Q13, ix ); Chris@69: /* Collapse stereo width */ Chris@69: width_Q14 = 0; Chris@69: pred_Q13[ 0 ] = 0; Chris@69: pred_Q13[ 1 ] = 0; Chris@69: } else if( state->smth_width_Q14 > SILK_FIX_CONST( 0.95, 14 ) ) { Chris@69: /* Full-width stereo coding */ Chris@69: silk_stereo_quant_pred( pred_Q13, ix ); Chris@69: width_Q14 = SILK_FIX_CONST( 1, 14 ); Chris@69: } else { Chris@69: /* Reduced-width stereo coding; scale down and quantize predictors */ Chris@69: pred_Q13[ 0 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 0 ] ), 14 ); Chris@69: pred_Q13[ 1 ] = silk_RSHIFT( silk_SMULBB( state->smth_width_Q14, pred_Q13[ 1 ] ), 14 ); Chris@69: silk_stereo_quant_pred( pred_Q13, ix ); Chris@69: width_Q14 = state->smth_width_Q14; Chris@69: } Chris@69: Chris@69: /* Make sure to keep on encoding until the tapered output has been transmitted */ Chris@69: if( *mid_only_flag == 1 ) { Chris@69: state->silent_side_len += frame_length - STEREO_INTERP_LEN_MS * fs_kHz; Chris@69: if( state->silent_side_len < LA_SHAPE_MS * fs_kHz ) { Chris@69: *mid_only_flag = 0; Chris@69: } else { Chris@69: /* Limit to avoid wrapping around */ Chris@69: state->silent_side_len = 10000; Chris@69: } Chris@69: } else { Chris@69: state->silent_side_len = 0; Chris@69: } Chris@69: Chris@69: if( *mid_only_flag == 0 && mid_side_rates_bps[ 1 ] < 1 ) { Chris@69: mid_side_rates_bps[ 1 ] = 1; Chris@69: mid_side_rates_bps[ 0 ] = silk_max_int( 1, total_rate_bps - mid_side_rates_bps[ 1 ]); Chris@69: } Chris@69: Chris@69: /* Interpolate predictors and subtract prediction from side channel */ Chris@69: pred0_Q13 = -state->pred_prev_Q13[ 0 ]; Chris@69: pred1_Q13 = -state->pred_prev_Q13[ 1 ]; Chris@69: w_Q24 = silk_LSHIFT( state->width_prev_Q14, 10 ); Chris@69: denom_Q16 = silk_DIV32_16( (opus_int32)1 << 16, STEREO_INTERP_LEN_MS * fs_kHz ); Chris@69: delta0_Q13 = -silk_RSHIFT_ROUND( silk_SMULBB( pred_Q13[ 0 ] - state->pred_prev_Q13[ 0 ], denom_Q16 ), 16 ); Chris@69: delta1_Q13 = -silk_RSHIFT_ROUND( silk_SMULBB( pred_Q13[ 1 ] - state->pred_prev_Q13[ 1 ], denom_Q16 ), 16 ); Chris@69: deltaw_Q24 = silk_LSHIFT( silk_SMULWB( width_Q14 - state->width_prev_Q14, denom_Q16 ), 10 ); Chris@69: for( n = 0; n < STEREO_INTERP_LEN_MS * fs_kHz; n++ ) { Chris@69: pred0_Q13 += delta0_Q13; Chris@69: pred1_Q13 += delta1_Q13; Chris@69: w_Q24 += deltaw_Q24; Chris@69: sum = silk_LSHIFT( silk_ADD_LSHIFT( mid[ n ] + (opus_int32)mid[ n + 2 ], mid[ n + 1 ], 1 ), 9 ); /* Q11 */ Chris@69: sum = silk_SMLAWB( silk_SMULWB( w_Q24, side[ n + 1 ] ), sum, pred0_Q13 ); /* Q8 */ Chris@69: sum = silk_SMLAWB( sum, silk_LSHIFT( (opus_int32)mid[ n + 1 ], 11 ), pred1_Q13 ); /* Q8 */ Chris@69: x2[ n - 1 ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( sum, 8 ) ); Chris@69: } Chris@69: Chris@69: pred0_Q13 = -pred_Q13[ 0 ]; Chris@69: pred1_Q13 = -pred_Q13[ 1 ]; Chris@69: w_Q24 = silk_LSHIFT( width_Q14, 10 ); Chris@69: for( n = STEREO_INTERP_LEN_MS * fs_kHz; n < frame_length; n++ ) { Chris@69: sum = silk_LSHIFT( silk_ADD_LSHIFT( mid[ n ] + (opus_int32)mid[ n + 2 ], mid[ n + 1 ], 1 ), 9 ); /* Q11 */ Chris@69: sum = silk_SMLAWB( silk_SMULWB( w_Q24, side[ n + 1 ] ), sum, pred0_Q13 ); /* Q8 */ Chris@69: sum = silk_SMLAWB( sum, silk_LSHIFT( (opus_int32)mid[ n + 1 ], 11 ), pred1_Q13 ); /* Q8 */ Chris@69: x2[ n - 1 ] = (opus_int16)silk_SAT16( silk_RSHIFT_ROUND( sum, 8 ) ); Chris@69: } Chris@69: state->pred_prev_Q13[ 0 ] = (opus_int16)pred_Q13[ 0 ]; Chris@69: state->pred_prev_Q13[ 1 ] = (opus_int16)pred_Q13[ 1 ]; Chris@69: state->width_prev_Q14 = (opus_int16)width_Q14; Chris@69: RESTORE_STACK; Chris@69: }