| 1 | /* |
| 2 | * Copyright (c) CMU 1993 Computer Science, Speech Group |
| 3 | * Chengxiang Lu and Alex Hauptmann |
| 4 | * Copyright (c) 2005 Steve Underwood <steveu at coppice.org> |
| 5 | * Copyright (c) 2009 Kenan Gillet |
| 6 | * Copyright (c) 2010 Martin Storsjo |
| 7 | * |
| 8 | * This file is part of FFmpeg. |
| 9 | * |
| 10 | * FFmpeg is free software; you can redistribute it and/or |
| 11 | * modify it under the terms of the GNU Lesser General Public |
| 12 | * License as published by the Free Software Foundation; either |
| 13 | * version 2.1 of the License, or (at your option) any later version. |
| 14 | * |
| 15 | * FFmpeg is distributed in the hope that it will be useful, |
| 16 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 18 | * Lesser General Public License for more details. |
| 19 | * |
| 20 | * You should have received a copy of the GNU Lesser General Public |
| 21 | * License along with FFmpeg; if not, write to the Free Software |
| 22 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 23 | */ |
| 24 | |
| 25 | /** |
| 26 | * @file |
| 27 | * G.722 ADPCM audio decoder |
| 28 | * |
| 29 | * This G.722 decoder is a bit-exact implementation of the ITU G.722 |
| 30 | * specification for all three specified bitrates - 64000bps, 56000bps |
| 31 | * and 48000bps. It passes the ITU tests. |
| 32 | * |
| 33 | * @note For the 56000bps and 48000bps bitrates, the lowest 1 or 2 bits |
| 34 | * respectively of each byte are ignored. |
| 35 | */ |
| 36 | |
| 37 | #include "libavutil/channel_layout.h" |
| 38 | #include "libavutil/opt.h" |
| 39 | #include "avcodec.h" |
| 40 | #include "get_bits.h" |
| 41 | #include "g722.h" |
| 42 | #include "internal.h" |
| 43 | |
| 44 | #define OFFSET(x) offsetof(G722Context, x) |
| 45 | #define AD AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_DECODING_PARAM |
| 46 | static const AVOption options[] = { |
| 47 | { "bits_per_codeword", "Bits per G722 codeword", OFFSET(bits_per_codeword), AV_OPT_TYPE_INT, { .i64 = 8 }, 6, 8, AD }, |
| 48 | { NULL } |
| 49 | }; |
| 50 | |
| 51 | static const AVClass g722_decoder_class = { |
| 52 | .class_name = "g722 decoder", |
| 53 | .item_name = av_default_item_name, |
| 54 | .option = options, |
| 55 | .version = LIBAVUTIL_VERSION_INT, |
| 56 | }; |
| 57 | |
| 58 | static av_cold int g722_decode_init(AVCodecContext * avctx) |
| 59 | { |
| 60 | G722Context *c = avctx->priv_data; |
| 61 | |
| 62 | avctx->channels = 1; |
| 63 | avctx->channel_layout = AV_CH_LAYOUT_MONO; |
| 64 | avctx->sample_fmt = AV_SAMPLE_FMT_S16; |
| 65 | |
| 66 | c->band[0].scale_factor = 8; |
| 67 | c->band[1].scale_factor = 2; |
| 68 | c->prev_samples_pos = 22; |
| 69 | |
| 70 | return 0; |
| 71 | } |
| 72 | |
| 73 | static const int16_t low_inv_quant5[32] = { |
| 74 | -35, -35, -2919, -2195, -1765, -1458, -1219, -1023, |
| 75 | -858, -714, -587, -473, -370, -276, -190, -110, |
| 76 | 2919, 2195, 1765, 1458, 1219, 1023, 858, 714, |
| 77 | 587, 473, 370, 276, 190, 110, 35, -35 |
| 78 | }; |
| 79 | |
| 80 | static const int16_t * const low_inv_quants[3] = { ff_g722_low_inv_quant6, |
| 81 | low_inv_quant5, |
| 82 | ff_g722_low_inv_quant4 }; |
| 83 | |
| 84 | static int g722_decode_frame(AVCodecContext *avctx, void *data, |
| 85 | int *got_frame_ptr, AVPacket *avpkt) |
| 86 | { |
| 87 | G722Context *c = avctx->priv_data; |
| 88 | AVFrame *frame = data; |
| 89 | int16_t *out_buf; |
| 90 | int j, ret; |
| 91 | const int skip = 8 - c->bits_per_codeword; |
| 92 | const int16_t *quantizer_table = low_inv_quants[skip]; |
| 93 | GetBitContext gb; |
| 94 | |
| 95 | /* get output buffer */ |
| 96 | frame->nb_samples = avpkt->size * 2; |
| 97 | if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) |
| 98 | return ret; |
| 99 | out_buf = (int16_t *)frame->data[0]; |
| 100 | |
| 101 | init_get_bits(&gb, avpkt->data, avpkt->size * 8); |
| 102 | |
| 103 | for (j = 0; j < avpkt->size; j++) { |
| 104 | int ilow, ihigh, rlow, rhigh, dhigh; |
| 105 | int xout1, xout2; |
| 106 | |
| 107 | ihigh = get_bits(&gb, 2); |
| 108 | ilow = get_bits(&gb, 6 - skip); |
| 109 | skip_bits(&gb, skip); |
| 110 | |
| 111 | rlow = av_clip((c->band[0].scale_factor * quantizer_table[ilow] >> 10) |
| 112 | + c->band[0].s_predictor, -16384, 16383); |
| 113 | |
| 114 | ff_g722_update_low_predictor(&c->band[0], ilow >> (2 - skip)); |
| 115 | |
| 116 | dhigh = c->band[1].scale_factor * ff_g722_high_inv_quant[ihigh] >> 10; |
| 117 | rhigh = av_clip(dhigh + c->band[1].s_predictor, -16384, 16383); |
| 118 | |
| 119 | ff_g722_update_high_predictor(&c->band[1], dhigh, ihigh); |
| 120 | |
| 121 | c->prev_samples[c->prev_samples_pos++] = rlow + rhigh; |
| 122 | c->prev_samples[c->prev_samples_pos++] = rlow - rhigh; |
| 123 | ff_g722_apply_qmf(c->prev_samples + c->prev_samples_pos - 24, |
| 124 | &xout1, &xout2); |
| 125 | *out_buf++ = av_clip_int16(xout1 >> 11); |
| 126 | *out_buf++ = av_clip_int16(xout2 >> 11); |
| 127 | if (c->prev_samples_pos >= PREV_SAMPLES_BUF_SIZE) { |
| 128 | memmove(c->prev_samples, c->prev_samples + c->prev_samples_pos - 22, |
| 129 | 22 * sizeof(c->prev_samples[0])); |
| 130 | c->prev_samples_pos = 22; |
| 131 | } |
| 132 | } |
| 133 | |
| 134 | *got_frame_ptr = 1; |
| 135 | |
| 136 | return avpkt->size; |
| 137 | } |
| 138 | |
| 139 | AVCodec ff_adpcm_g722_decoder = { |
| 140 | .name = "g722", |
| 141 | .long_name = NULL_IF_CONFIG_SMALL("G.722 ADPCM"), |
| 142 | .type = AVMEDIA_TYPE_AUDIO, |
| 143 | .id = AV_CODEC_ID_ADPCM_G722, |
| 144 | .priv_data_size = sizeof(G722Context), |
| 145 | .init = g722_decode_init, |
| 146 | .decode = g722_decode_frame, |
| 147 | .capabilities = CODEC_CAP_DR1, |
| 148 | .priv_class = &g722_decoder_class, |
| 149 | }; |