annotate binkaudio.c @ 11352:6e0af2cfdcfe libavcodec

Do MC and IDCT in coding (hilbert) order This increases the slice size to 64 pixels, due to having to decode an entire chroma superblock row per slice. This can be up to 6% slower depending on clip and CPU, but is necessary for future optimizations that gain significantly more than was lost.

/*
 * Bink Audio decoder
 * Copyright (c) 2007-2010 Peter Ross ([e-mail])
 * Copyright (c) 2009 Daniel Verkamp ([e-mail])
 *
 * 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
 */

/**
 * @file libavcodec/binkaudio.c
 * Bink Audio decoder
 *
 * Technical details here:
 *  http://wiki.multimedia.cx/index.php?title=Bink_Audio
 */

#include "avcodec.h"
#define ALT_BITSTREAM_READER_LE
#include "get_bits.h"
#include "dsputil.h"
extern const uint16_t ff_wma_critical_freqs[25];

#define MAX_CHANNELS 2
#define BINK_BLOCK_MAX_SIZE (MAX_CHANNELS << 11)

typedef struct {
    AVCodecContext *avctx;
    GetBitContext gb;
    DSPContext dsp;
    int first;
    int channels;
    int frame_len;          ///< transform size (samples)
    int overlap_len;        ///< overlap size (samples)
    int block_size;
    int num_bands;
    unsigned int *bands;
    float root;
    DECLARE_ALIGNED_16(FFTSample, coeffs)[BINK_BLOCK_MAX_SIZE];
    DECLARE_ALIGNED_16(short, previous)[BINK_BLOCK_MAX_SIZE / 16];  ///< coeffs from previous audio block
    float *coeffs_ptr[MAX_CHANNELS]; ///< pointers to the coeffs arrays for float_to_int16_interleave
    union {
        RDFTContext rdft;
        DCTContext dct;
    } trans;
} BinkAudioContext;


static av_cold int decode_init(AVCodecContext *avctx)
{
    BinkAudioContext *s = avctx->priv_data;
    int sample_rate = avctx->sample_rate;
    int sample_rate_half;
    int i;
    int frame_len_bits;

    s->avctx = avctx;
    dsputil_init(&s->dsp, avctx);

    /* determine frame length */
    if (avctx->sample_rate < 22050) {
        frame_len_bits = 9;
    } else if (avctx->sample_rate < 44100) {
        frame_len_bits = 10;
    } else {
        frame_len_bits = 11;
    }
    s->frame_len = 1 << frame_len_bits;

    if (s->channels > MAX_CHANNELS) {
        av_log(s->avctx, AV_LOG_ERROR, "too many channels: %d\n", s->channels);
        return -1;
    }

    if (avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT) {
        // audio is already interleaved for the RDFT format variant
        sample_rate  *= avctx->channels;
        s->frame_len *= avctx->channels;
        s->channels = 1;
        if (avctx->channels == 2)
            frame_len_bits++;
    } else {
        s->channels = avctx->channels;
    }

    s->overlap_len   = s->frame_len / 16;
    s->block_size    = (s->frame_len - s->overlap_len) * s->channels;
    sample_rate_half = (sample_rate + 1) / 2;
    s->root          = 2.0 / sqrt(s->frame_len);

    /* calculate number of bands */
    for (s->num_bands = 1; s->num_bands < 25; s->num_bands++)
        if (sample_rate_half <= ff_wma_critical_freqs[s->num_bands - 1])
            break;

    s->bands = av_malloc((s->num_bands + 1) * sizeof(*s->bands));
    if (!s->bands)
        return AVERROR(ENOMEM);

    /* populate bands data */
    s->bands[0] = 1;
    for (i = 1; i < s->num_bands; i++)
        s->bands[i] = ff_wma_critical_freqs[i - 1] * (s->frame_len / 2) / sample_rate_half;
    s->bands[s->num_bands] = s->frame_len / 2;

    s->first = 1;
    avctx->sample_fmt = SAMPLE_FMT_S16;

    for (i = 0; i < s->channels; i++)
        s->coeffs_ptr[i] = s->coeffs + i * s->frame_len;

    if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT)
        ff_rdft_init(&s->trans.rdft, frame_len_bits, IRIDFT);
    else if (CONFIG_BINKAUDIO_DCT_DECODER)
        ff_dct_init(&s->trans.dct, frame_len_bits, 1);
    else
        return -1;

    return 0;
}

static float get_float(GetBitContext *gb)
{
    int power = get_bits(gb, 5);
    float f = ldexpf(get_bits_long(gb, 23), power - 23);
    if (get_bits1(gb))
        f = -f;
    return f;
}

static const uint8_t rle_length_tab[16] = {
    2, 3, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 15, 16, 32, 64
};

/**
 * Decode Bink Audio block
 * @param[out] out Output buffer (must contain s->block_size elements)
 */
static void decode_block(BinkAudioContext *s, short *out, int use_dct)
{
    int ch, i, j, k;
    float q, quant[25];
    int width, coeff;
    GetBitContext *gb = &s->gb;

    if (use_dct)
        skip_bits(gb, 2);

    for (ch = 0; ch < s->channels; ch++) {
        FFTSample *coeffs = s->coeffs_ptr[ch];
        q = 0.0f;
        coeffs[0] = get_float(gb) * s->root;
        coeffs[1] = get_float(gb) * s->root;

        for (i = 0; i < s->num_bands; i++) {
            /* constant is result of 0.066399999/log10(M_E) */
            int value = get_bits(gb, 8);
            quant[i] = expf(FFMIN(value, 95) * 0.15289164787221953823f) * s->root;
        }

        // find band (k)
        for (k = 0; s->bands[k] < 1; k++) {
            q = quant[k];
        }

        // parse coefficients
        i = 2;
        while (i < s->frame_len) {
            if (get_bits1(gb)) {
                j = i + rle_length_tab[get_bits(gb, 4)] * 8;
            } else {
                j = i + 8;
            }

            j = FFMIN(j, s->frame_len);

            width = get_bits(gb, 4);
            if (width == 0) {
                memset(coeffs + i, 0, (j - i) * sizeof(*coeffs));
                i = j;
                while (s->bands[k] * 2 < i)
                    q = quant[k++];
            } else {
                while (i < j) {
                    if (s->bands[k] * 2 == i)
                        q = quant[k++];
                    coeff = get_bits(gb, width);
                    if (coeff) {
                        if (get_bits1(gb))
                            coeffs[i] = -q * coeff;
                        else
                            coeffs[i] =  q * coeff;
                    } else {
                        coeffs[i] = 0.0f;
                    }
                    i++;
                }
            }
        }

        if (CONFIG_BINKAUDIO_DCT_DECODER && use_dct) {
            coeffs[0] /= 0.5;
            ff_dct_calc (&s->trans.dct,  coeffs);
            s->dsp.vector_fmul_scalar(coeffs, coeffs, s->frame_len / 2, s->frame_len);
        }
        else if (CONFIG_BINKAUDIO_RDFT_DECODER)
            ff_rdft_calc(&s->trans.rdft, coeffs);
    }

    s->dsp.float_to_int16_interleave(out, (const float **)s->coeffs_ptr, s->frame_len, s->channels);

    if (!s->first) {
        int count = s->overlap_len * s->channels;
        int shift = av_log2(count);
        for (i = 0; i < count; i++) {
            out[i] = (s->previous[i] * (count - i) + out[i] * i) >> shift;
        }
    }

    memcpy(s->previous, out + s->block_size,
           s->overlap_len * s->channels * sizeof(*out));

    s->first = 0;
}

static av_cold int decode_end(AVCodecContext *avctx)
{
    BinkAudioContext * s = avctx->priv_data;
    av_freep(&s->bands);
    if (CONFIG_BINKAUDIO_RDFT_DECODER && avctx->codec->id == CODEC_ID_BINKAUDIO_RDFT)
        ff_rdft_end(&s->trans.rdft);
    else if (CONFIG_BINKAUDIO_DCT_DECODER)
        ff_dct_end(&s->trans.dct);
    return 0;
}

static void get_bits_align32(GetBitContext *s)
{
    int n = (-get_bits_count(s)) & 31;
    if (n) skip_bits(s, n);
}

static int decode_frame(AVCodecContext *avctx,
                        void *data, int *data_size,
                        AVPacket *avpkt)
{
    BinkAudioContext *s = avctx->priv_data;
    const uint8_t *buf  = avpkt->data;
    int buf_size        = avpkt->size;
    short *samples      = data;
    short *samples_end  = (short*)((uint8_t*)data + *data_size);
    int reported_size;
    GetBitContext *gb = &s->gb;

    init_get_bits(gb, buf, buf_size * 8);

    reported_size = get_bits_long(gb, 32);
    while (get_bits_count(gb) / 8 < buf_size &&
           samples + s->block_size <= samples_end) {
        decode_block(s, samples, avctx->codec->id == CODEC_ID_BINKAUDIO_DCT);
        samples += s->block_size;
        get_bits_align32(gb);
    }

    *data_size = FFMIN(reported_size, (uint8_t*)samples - (uint8_t*)data);
    return buf_size;
}

AVCodec binkaudio_rdft_decoder = {
    "binkaudio_rdft",
    CODEC_TYPE_AUDIO,
    CODEC_ID_BINKAUDIO_RDFT,
    sizeof(BinkAudioContext),
    decode_init,
    NULL,
    decode_end,
    decode_frame,
    .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (RDFT)")
};

AVCodec binkaudio_dct_decoder = {
    "binkaudio_dct",
    CODEC_TYPE_AUDIO,
    CODEC_ID_BINKAUDIO_DCT,
    sizeof(BinkAudioContext),
    decode_init,
    NULL,
    decode_end,
    decode_frame,
    .long_name = NULL_IF_CONFIG_SMALL("Bink Audio (DCT)")
};