| 1 | /* |
| 2 | * RFC 3389 comfort noise generator |
| 3 | * Copyright (c) 2012 Martin Storsjo |
| 4 | * |
| 5 | * This file is part of FFmpeg. |
| 6 | * |
| 7 | * FFmpeg is free software; you can redistribute it and/or |
| 8 | * modify it under the terms of the GNU Lesser General Public |
| 9 | * License as published by the Free Software Foundation; either |
| 10 | * version 2.1 of the License, or (at your option) any later version. |
| 11 | * |
| 12 | * FFmpeg is distributed in the hope that it will be useful, |
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 15 | * Lesser General Public License for more details. |
| 16 | * |
| 17 | * You should have received a copy of the GNU Lesser General Public |
| 18 | * License along with FFmpeg; if not, write to the Free Software |
| 19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 20 | */ |
| 21 | |
| 22 | #include <math.h> |
| 23 | |
| 24 | #include "libavutil/common.h" |
| 25 | #include "avcodec.h" |
| 26 | #include "celp_filters.h" |
| 27 | #include "internal.h" |
| 28 | #include "libavutil/lfg.h" |
| 29 | |
| 30 | typedef struct CNGContext { |
| 31 | float *refl_coef, *target_refl_coef; |
| 32 | float *lpc_coef; |
| 33 | int order; |
| 34 | int energy, target_energy; |
| 35 | int inited; |
| 36 | float *filter_out; |
| 37 | float *excitation; |
| 38 | AVLFG lfg; |
| 39 | } CNGContext; |
| 40 | |
| 41 | static av_cold int cng_decode_close(AVCodecContext *avctx) |
| 42 | { |
| 43 | CNGContext *p = avctx->priv_data; |
| 44 | av_freep(&p->refl_coef); |
| 45 | av_freep(&p->target_refl_coef); |
| 46 | av_freep(&p->lpc_coef); |
| 47 | av_freep(&p->filter_out); |
| 48 | av_freep(&p->excitation); |
| 49 | return 0; |
| 50 | } |
| 51 | |
| 52 | static av_cold int cng_decode_init(AVCodecContext *avctx) |
| 53 | { |
| 54 | CNGContext *p = avctx->priv_data; |
| 55 | |
| 56 | avctx->sample_fmt = AV_SAMPLE_FMT_S16; |
| 57 | avctx->channels = 1; |
| 58 | avctx->sample_rate = 8000; |
| 59 | |
| 60 | p->order = 12; |
| 61 | avctx->frame_size = 640; |
| 62 | p->refl_coef = av_mallocz_array(p->order, sizeof(*p->refl_coef)); |
| 63 | p->target_refl_coef = av_mallocz_array(p->order, sizeof(*p->target_refl_coef)); |
| 64 | p->lpc_coef = av_mallocz_array(p->order, sizeof(*p->lpc_coef)); |
| 65 | p->filter_out = av_mallocz_array(avctx->frame_size + p->order, |
| 66 | sizeof(*p->filter_out)); |
| 67 | p->excitation = av_mallocz_array(avctx->frame_size, sizeof(*p->excitation)); |
| 68 | if (!p->refl_coef || !p->target_refl_coef || !p->lpc_coef || |
| 69 | !p->filter_out || !p->excitation) { |
| 70 | cng_decode_close(avctx); |
| 71 | return AVERROR(ENOMEM); |
| 72 | } |
| 73 | |
| 74 | av_lfg_init(&p->lfg, 0); |
| 75 | |
| 76 | return 0; |
| 77 | } |
| 78 | |
| 79 | static void make_lpc_coefs(float *lpc, const float *refl, int order) |
| 80 | { |
| 81 | float buf[100]; |
| 82 | float *next, *cur; |
| 83 | int m, i; |
| 84 | next = buf; |
| 85 | cur = lpc; |
| 86 | for (m = 0; m < order; m++) { |
| 87 | next[m] = refl[m]; |
| 88 | for (i = 0; i < m; i++) |
| 89 | next[i] = cur[i] + refl[m] * cur[m - i - 1]; |
| 90 | FFSWAP(float*, next, cur); |
| 91 | } |
| 92 | if (cur != lpc) |
| 93 | memcpy(lpc, cur, sizeof(*lpc) * order); |
| 94 | } |
| 95 | |
| 96 | static void cng_decode_flush(AVCodecContext *avctx) |
| 97 | { |
| 98 | CNGContext *p = avctx->priv_data; |
| 99 | p->inited = 0; |
| 100 | } |
| 101 | |
| 102 | static int cng_decode_frame(AVCodecContext *avctx, void *data, |
| 103 | int *got_frame_ptr, AVPacket *avpkt) |
| 104 | { |
| 105 | AVFrame *frame = data; |
| 106 | CNGContext *p = avctx->priv_data; |
| 107 | int buf_size = avpkt->size; |
| 108 | int ret, i; |
| 109 | int16_t *buf_out; |
| 110 | float e = 1.0; |
| 111 | float scaling; |
| 112 | |
| 113 | if (avpkt->size) { |
| 114 | int dbov = -avpkt->data[0]; |
| 115 | p->target_energy = 1081109975 * pow(10, dbov / 10.0) * 0.75; |
| 116 | memset(p->target_refl_coef, 0, p->order * sizeof(*p->target_refl_coef)); |
| 117 | for (i = 0; i < FFMIN(avpkt->size - 1, p->order); i++) { |
| 118 | p->target_refl_coef[i] = (avpkt->data[1 + i] - 127) / 128.0; |
| 119 | } |
| 120 | } |
| 121 | |
| 122 | if (p->inited) { |
| 123 | p->energy = p->energy / 2 + p->target_energy / 2; |
| 124 | for (i = 0; i < p->order; i++) |
| 125 | p->refl_coef[i] = 0.6 *p->refl_coef[i] + 0.4 * p->target_refl_coef[i]; |
| 126 | } else { |
| 127 | p->energy = p->target_energy; |
| 128 | memcpy(p->refl_coef, p->target_refl_coef, p->order * sizeof(*p->refl_coef)); |
| 129 | p->inited = 1; |
| 130 | } |
| 131 | make_lpc_coefs(p->lpc_coef, p->refl_coef, p->order); |
| 132 | |
| 133 | for (i = 0; i < p->order; i++) |
| 134 | e *= 1.0 - p->refl_coef[i]*p->refl_coef[i]; |
| 135 | |
| 136 | scaling = sqrt(e * p->energy / 1081109975); |
| 137 | for (i = 0; i < avctx->frame_size; i++) { |
| 138 | int r = (av_lfg_get(&p->lfg) & 0xffff) - 0x8000; |
| 139 | p->excitation[i] = scaling * r; |
| 140 | } |
| 141 | ff_celp_lp_synthesis_filterf(p->filter_out + p->order, p->lpc_coef, |
| 142 | p->excitation, avctx->frame_size, p->order); |
| 143 | |
| 144 | frame->nb_samples = avctx->frame_size; |
| 145 | if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) |
| 146 | return ret; |
| 147 | buf_out = (int16_t *)frame->data[0]; |
| 148 | for (i = 0; i < avctx->frame_size; i++) |
| 149 | buf_out[i] = p->filter_out[i + p->order]; |
| 150 | memcpy(p->filter_out, p->filter_out + avctx->frame_size, |
| 151 | p->order * sizeof(*p->filter_out)); |
| 152 | |
| 153 | *got_frame_ptr = 1; |
| 154 | |
| 155 | return buf_size; |
| 156 | } |
| 157 | |
| 158 | AVCodec ff_comfortnoise_decoder = { |
| 159 | .name = "comfortnoise", |
| 160 | .long_name = NULL_IF_CONFIG_SMALL("RFC 3389 comfort noise generator"), |
| 161 | .type = AVMEDIA_TYPE_AUDIO, |
| 162 | .id = AV_CODEC_ID_COMFORT_NOISE, |
| 163 | .priv_data_size = sizeof(CNGContext), |
| 164 | .init = cng_decode_init, |
| 165 | .decode = cng_decode_frame, |
| 166 | .flush = cng_decode_flush, |
| 167 | .close = cng_decode_close, |
| 168 | .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16, |
| 169 | AV_SAMPLE_FMT_NONE }, |
| 170 | .capabilities = CODEC_CAP_DELAY | CODEC_CAP_DR1, |
| 171 | }; |