diff options
author | Paul B Mahol <onemda@gmail.com> | 2020-03-15 15:28:13 +0300 |
---|---|---|
committer | Paul B Mahol <onemda@gmail.com> | 2020-03-17 18:07:25 +0300 |
commit | fed0de3728c201ac15d8cf9bb1acc046a1b5230e (patch) | |
tree | 153ff69faab056dee500fb8a118321d1a5cb85a8 /libavcodec/hcadec.c | |
parent | 0bbc4914da07bfe3351354e0f2fa85e5cb437e98 (diff) |
avcodec: add CRI HCA decoder
Diffstat (limited to 'libavcodec/hcadec.c')
-rw-r--r-- | libavcodec/hcadec.c | 453 |
1 files changed, 453 insertions, 0 deletions
diff --git a/libavcodec/hcadec.c b/libavcodec/hcadec.c new file mode 100644 index 0000000000..4e3f589579 --- /dev/null +++ b/libavcodec/hcadec.c @@ -0,0 +1,453 @@ +/* + * This file is part of FFmpeg. + * + * FFmpeg is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * FFmpeg is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with FFmpeg; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include "libavutil/crc.h" +#include "libavutil/float_dsp.h" +#include "libavutil/intreadwrite.h" +#include "libavutil/tx.h" + +#include "avcodec.h" +#include "get_bits.h" +#include "internal.h" +#include "hca_data.h" + +typedef struct ChannelContext { + float base[128]; + DECLARE_ALIGNED(32, float, imdct_in)[128]; + DECLARE_ALIGNED(32, float, imdct_out)[128]; + DECLARE_ALIGNED(32, float, imdct_prev)[128]; + int8_t scale_factors[128]; + uint8_t scale[128]; + int8_t intensity[8]; + int8_t *hfr_scale; + unsigned count; + int chan_type; +} ChannelContext; + +typedef struct HCAContext { + GetBitContext gb; + + const AVCRC *crc_table; + + ChannelContext ch[16]; + + uint8_t ath[128]; + + int ath_type; + unsigned hfr_group_count; + uint8_t track_count; + uint8_t channel_config; + uint8_t total_band_count; + uint8_t base_band_count; + uint8_t stereo_band_count; + uint8_t bands_per_hfr_group; + + av_tx_fn tx_fn; + AVTXContext *tx_ctx; + AVFloatDSPContext *fdsp; +} HCAContext; + +static void ath_init1(uint8_t *ath, int sample_rate) +{ + unsigned int index; + unsigned int acc = 0; + + for (int i = 0; i < 128; i++) { + acc += sample_rate; + index = acc >> 13; + + if (index >= 654) { + memset(ath+i, 0xFF, (128 - i)); + break; + } + + ath[i] = ath_base_curve[index]; + } +} + +static int ath_init(uint8_t *ath, int type, int sample_rate) +{ + switch (type) { + case 0: + /* nothing to do */ + break; + case 1: + ath_init1(ath, sample_rate); + break; + default: + return AVERROR_INVALIDDATA; + } + + return 0; +} + +static inline unsigned ceil2(unsigned a, unsigned b) +{ + return (b > 0) ? (a / b + ((a % b) ? 1 : 0)) : 0; +} + +static av_cold int decode_init(AVCodecContext *avctx) +{ + HCAContext *c = avctx->priv_data; + GetBitContext *gb = &c->gb; + int8_t r[16] = { 0 }; + float scale = 1.f / 8.f; + unsigned b, chunk; + int version, ret; + + avctx->sample_fmt = AV_SAMPLE_FMT_FLTP; + c->crc_table = av_crc_get_table(AV_CRC_16_ANSI); + + if (avctx->channels <= 0 || avctx->channels > 16) + return AVERROR(EINVAL); + + ret = init_get_bits8(gb, avctx->extradata, avctx->extradata_size); + if (ret < 0) + return ret; + skip_bits_long(gb, 32); + version = get_bits(gb, 16); + skip_bits_long(gb, 16); + + c->ath_type = version >= 0x200 ? 0 : 1; + + if (get_bits_long(gb, 32) != MKBETAG('f', 'm', 't', 0)) + return AVERROR_INVALIDDATA; + skip_bits_long(gb, 32); + skip_bits_long(gb, 32); + skip_bits_long(gb, 32); + + chunk = get_bits_long(gb, 32); + if (chunk == MKBETAG('c', 'o', 'm', 'p')) { + skip_bits_long(gb, 16); + skip_bits_long(gb, 8); + skip_bits_long(gb, 8); + c->track_count = get_bits(gb, 8); + c->channel_config = get_bits(gb, 8); + c->total_band_count = get_bits(gb, 8); + c->base_band_count = get_bits(gb, 8); + c->stereo_band_count = get_bits(gb, 8); + c->bands_per_hfr_group = get_bits(gb, 8); + } else if (chunk == MKBETAG('d', 'e', 'c', 0)) { + skip_bits_long(gb, 16); + skip_bits_long(gb, 8); + skip_bits_long(gb, 8); + c->total_band_count = get_bits(gb, 8) + 1; + c->base_band_count = get_bits(gb, 8) + 1; + c->track_count = get_bits(gb, 4); + c->channel_config = get_bits(gb, 4); + if (!get_bits(gb, 8)) + c->base_band_count = c->total_band_count; + c->stereo_band_count = c->total_band_count - c->base_band_count; + c->bands_per_hfr_group = 0; + } else + return AVERROR_INVALIDDATA; + + while (get_bits_left(gb) >= 32) { + chunk = get_bits_long(gb, 32); + if (chunk == MKBETAG('v', 'b', 'r', 0)) { + skip_bits_long(gb, 16); + skip_bits_long(gb, 16); + } else if (chunk == MKBETAG('a', 't', 'h', 0)) { + c->ath_type = get_bits(gb, 16); + } else if (chunk == MKBETAG('r', 'v', 'a', 0)) { + skip_bits_long(gb, 32); + } else if (chunk == MKBETAG('c', 'o', 'm', 'm')) { + skip_bits_long(gb, get_bits(gb, 8) * 8); + } else if (chunk == MKBETAG('c', 'i', 'p', 'h')) { + skip_bits_long(gb, 16); + } else if (chunk == MKBETAG('l', 'o', 'o', 'p')) { + skip_bits_long(gb, 32); + skip_bits_long(gb, 32); + skip_bits_long(gb, 16); + skip_bits_long(gb, 16); + } else if (chunk == MKBETAG('p', 'a', 'd', 0)) { + break; + } else { + break; + } + } + + ret = ath_init(c->ath, c->ath_type, avctx->sample_rate); + if (ret < 0) + return ret; + + if (!c->track_count) + c->track_count = 1; + + b = avctx->channels / c->track_count; + if (c->stereo_band_count && b > 1) { + int8_t *x = r; + + for (int i = 0; i < c->track_count; i++, x+=b) { + switch (b) { + case 2: + case 3: + x[0] = 1; + x[1] = 2; + break; + case 4: + x[0]=1; x[1] = 2; + if (c->channel_config == 0) { + x[2]=1; + x[3]=2; + } + break; + case 5: + x[0]=1; x[1] = 2; + if (c->channel_config <= 2) { + x[3]=1; + x[4]=2; + } + break; + case 6: + case 7: + x[0] = 1; x[1] = 2; x[4] = 1; x[5] = 2; + break; + case 8: + x[0] = 1; x[1] = 2; x[4] = 1; x[5] = 2; x[6] = 1; x[7] = 2; + break; + } + } + } + + if (c->total_band_count < c->base_band_count) + return AVERROR_INVALIDDATA; + + c->hfr_group_count = ceil2(c->total_band_count - (c->base_band_count + c->stereo_band_count), + c->bands_per_hfr_group); + + if (c->base_band_count + c->stereo_band_count + (unsigned long)c->hfr_group_count > 128ULL) + return AVERROR_INVALIDDATA; + + for (int i = 0; i < avctx->channels; i++) { + c->ch[i].chan_type = r[i]; + c->ch[i].count = c->base_band_count + ((r[i] != 2) ? c->stereo_band_count : 0); + c->ch[i].hfr_scale = &c->ch[i].scale_factors[c->base_band_count + c->stereo_band_count]; + if (c->ch[i].count > 128) + return AVERROR_INVALIDDATA; + } + + c->fdsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT); + if (!c->fdsp) + return AVERROR(ENOMEM); + + return av_tx_init(&c->tx_ctx, &c->tx_fn, AV_TX_FLOAT_MDCT, 1, 128, &scale, 0); +} + +static void run_imdct(HCAContext *c, ChannelContext *ch, int index, float *out) +{ + c->tx_fn(c->tx_ctx, ch->imdct_out, ch->imdct_in, sizeof(float)); + + c->fdsp->vector_fmul_window(out, ch->imdct_prev + (128 >> 1), + ch->imdct_out, window, 128 >> 1); + + memcpy(ch->imdct_prev, ch->imdct_out, 128 * sizeof(float)); +} + +static void apply_intensity_stereo(HCAContext *s, ChannelContext *ch1, ChannelContext *ch2, + int index, unsigned band_count, unsigned base_band_count, + unsigned stereo_band_count) +{ + float ratio_l = intensity_ratio_table[ch1->intensity[index]]; + float ratio_r = ratio_l - 2.0f; + float *c1 = &ch1->imdct_in[base_band_count]; + float *c2 = &ch2->imdct_in[base_band_count]; + + if (ch1->chan_type != 1 || !stereo_band_count) + return; + + for (int i = 0; i < band_count; i++) { + *(c2++) = *c1 * ratio_r; + *(c1++) *= ratio_l; + } +} + +static void reconstruct_hfr(HCAContext *s, ChannelContext *ch, + unsigned hfr_group_count, + unsigned bands_per_hfr_group, + unsigned start_band, unsigned total_band_count) +{ + if (ch->chan_type == 2 || !bands_per_hfr_group) + return; + + for (int i = 0, k = start_band, l = start_band - 1; i < hfr_group_count; i++){ + for (int j = 0; j < bands_per_hfr_group && k < total_band_count; j++, k++, l--){ + ch->imdct_in[k] = scale_conversion_table[ch->hfr_scale[i] - ch->scale_factors[l]] * ch->imdct_in[l]; + } + } + + ch->imdct_in[127] = 0; +} + +static void dequantize_coefficients(HCAContext *c, ChannelContext *ch) +{ + GetBitContext *gb = &c->gb; + + for (int i = 0; i < ch->count; i++) { + unsigned scale = ch->scale[i]; + int nb_bits = max_bits_table[scale]; + int value = get_bitsz(gb, nb_bits); + float factor; + + if (scale > 7) { + value = (1 - ((value & 1) << 1)) * (value >> 1); + if (!value) + skip_bits_long(gb, -1); + factor = value; + } else { + value += scale << 4; + skip_bits_long(gb, quant_spectrum_bits[value] - nb_bits); + factor = quant_spectrum_value[value]; + } + ch->imdct_in[i] = factor * ch->base[i]; + } + + memset(ch->imdct_in + ch->count, 0, sizeof(ch->imdct_in) - ch->count * sizeof(ch->imdct_in[0])); +} + +static void unpack(HCAContext *c, ChannelContext *ch, + unsigned hfr_group_count, + int packed_noise_level, + const uint8_t *ath) +{ + GetBitContext *gb = &c->gb; + int delta_bits = get_bits(gb, 3); + + if (delta_bits > 5) { + for (int i = 0; i < ch->count; i++) + ch->scale_factors[i] = get_bits(gb, 6); + } else if (delta_bits) { + int factor = get_bits(gb, 6); + int max_value = (1 << delta_bits) - 1; + int half_max = max_value >> 1; + + ch->scale_factors[0] = factor; + for (int i = 1; i < ch->count; i++){ + int delta = get_bits(gb, delta_bits); + + if (delta == max_value) { + factor = get_bits(gb, 6); + } else { + factor += delta - half_max; + } + + ch->scale_factors[i] = factor; + } + } else { + memset(ch->scale_factors, 0, 128); + } + + if (ch->chan_type == 2){ + ch->intensity[0] = get_bits(gb, 4); + if (ch->intensity[0] < 15) { + for (int i = 1; i < 8; i++) + ch->intensity[i] = get_bits(gb, 4); + } + } else { + for (int i = 0; i < hfr_group_count; i++) + ch->hfr_scale[i] = get_bits(gb, 6); + } + + for (int i = 0; i < ch->count; i++) { + int scale = ch->scale_factors[i]; + + if (scale) { + scale = c->ath[i] + ((packed_noise_level + i) >> 8) - ((scale * 5) >> 1) + 2; + scale = scale_table[av_clip(scale, 0, 58)]; + } + ch->scale[i] = scale; + } + + memset(ch->scale + ch->count, 0, sizeof(ch->scale) - ch->count); + + for (int i = 0; i < ch->count; i++) + ch->base[i] = dequantizer_scaling_table[ch->scale_factors[i]] * quant_step_size[ch->scale[i]]; +} + +static int decode_frame(AVCodecContext *avctx, void *data, + int *got_frame_ptr, AVPacket *avpkt) +{ + AVFrame *frame = data; + HCAContext *c = avctx->priv_data; + int ch, ret, packed_noise_level; + GetBitContext *gb = &c->gb; + float **samples; + + if (avctx->err_recognition & AV_EF_CRCCHECK) { + if (av_crc(c->crc_table, 0, avpkt->data, avpkt->size)) + return AVERROR_INVALIDDATA; + } + + if ((ret = init_get_bits8(gb, avpkt->data, avpkt->size)) < 0) + return ret; + + if (get_bits(gb, 16) != 0xFFFF) + return AVERROR_INVALIDDATA; + + frame->nb_samples = 1024; + if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) + return ret; + samples = (float **)frame->extended_data; + + packed_noise_level = (get_bits(gb, 9) << 8) - get_bits(gb, 7); + + for (ch = 0; ch < avctx->channels; ch++) + unpack(c, &c->ch[ch], c->hfr_group_count, packed_noise_level, c->ath); + + for (int i = 0; i < 8; i++) { + for (ch = 0; ch < avctx->channels; ch++) + dequantize_coefficients(c, &c->ch[ch]); + for (ch = 0; ch < avctx->channels; ch++) + reconstruct_hfr(c, &c->ch[ch], c->hfr_group_count, c->bands_per_hfr_group, + c->stereo_band_count + c->base_band_count, c->total_band_count); + for (ch = 0; ch < avctx->channels - 1; ch++) + apply_intensity_stereo(c, &c->ch[ch], &c->ch[ch+1], i, + c->total_band_count - c->base_band_count, + c->base_band_count, c->stereo_band_count); + for (ch = 0; ch < avctx->channels; ch++) + run_imdct(c, &c->ch[ch], i, samples[ch] + i * 128); + } + + *got_frame_ptr = 1; + + return avpkt->size; +} + +static av_cold int decode_close(AVCodecContext *avctx) +{ + HCAContext *c = avctx->priv_data; + + av_freep(&c->fdsp); + av_tx_uninit(&c->tx_ctx); + + return 0; +} + +AVCodec ff_hca_decoder = { + .name = "hca", + .long_name = NULL_IF_CONFIG_SMALL("CRI HCA"), + .type = AVMEDIA_TYPE_AUDIO, + .id = AV_CODEC_ID_HCA, + .priv_data_size = sizeof(HCAContext), + .init = decode_init, + .decode = decode_frame, + .close = decode_close, + .capabilities = AV_CODEC_CAP_DR1, + .sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP, + AV_SAMPLE_FMT_NONE }, +}; |