annotate wmaprodec.c @ 12197:fbf4d5b1b664 libavcodec

Remove FF_MM_SSE2/3 flags for CPUs where this is generally not faster than regular MMX code. Examples of this are the Core1 CPU. Instead, set a new flag, FF_MM_SSE2/3SLOW, which can be checked for particular SSE2/3 functions that have been checked specifically on such CPUs and are actually faster than their MMX counterparts. In addition, use this flag to enable particular VP8 and LPC SSE2 functions that are faster than their MMX counterparts. Based on a patch by Loren Merritt <lorenm AT u washington edu>.

/*
 * Wmapro compatible decoder
 * Copyright (c) 2007 Baptiste Coudurier, Benjamin Larsson, Ulion
 * Copyright (c) 2008 - 2009 Sascha Sommer, Benjamin Larsson
 *
 * 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
 * @brief wmapro decoder implementation
 * Wmapro is an MDCT based codec comparable to wma standard or AAC.
 * The decoding therefore consists of the following steps:
 * - bitstream decoding
 * - reconstruction of per-channel data
 * - rescaling and inverse quantization
 * - IMDCT
 * - windowing and overlapp-add
 *
 * The compressed wmapro bitstream is split into individual packets.
 * Every such packet contains one or more wma frames.
 * The compressed frames may have a variable length and frames may
 * cross packet boundaries.
 * Common to all wmapro frames is the number of samples that are stored in
 * a frame.
 * The number of samples and a few other decode flags are stored
 * as extradata that has to be passed to the decoder.
 *
 * The wmapro frames themselves are again split into a variable number of
 * subframes. Every subframe contains the data for 2^N time domain samples
 * where N varies between 7 and 12.
 *
 * Example wmapro bitstream (in samples):
 *
 * ||   packet 0           || packet 1 || packet 2      packets
 * ---------------------------------------------------
 * || frame 0      || frame 1       || frame 2    ||    frames
 * ---------------------------------------------------
 * ||   |      |   ||   |   |   |   ||            ||    subframes of channel 0
 * ---------------------------------------------------
 * ||      |   |   ||   |   |   |   ||            ||    subframes of channel 1
 * ---------------------------------------------------
 *
 * The frame layouts for the individual channels of a wma frame does not need
 * to be the same.
 *
 * However, if the offsets and lengths of several subframes of a frame are the
 * same, the subframes of the channels can be grouped.
 * Every group may then use special coding techniques like M/S stereo coding
 * to improve the compression ratio. These channel transformations do not
 * need to be applied to a whole subframe. Instead, they can also work on
 * individual scale factor bands (see below).
 * The coefficients that carry the audio signal in the frequency domain
 * are transmitted as huffman-coded vectors with 4, 2 and 1 elements.
 * In addition to that, the encoder can switch to a runlevel coding scheme
 * by transmitting subframe_length / 128 zero coefficients.
 *
 * Before the audio signal can be converted to the time domain, the
 * coefficients have to be rescaled and inverse quantized.
 * A subframe is therefore split into several scale factor bands that get
 * scaled individually.
 * Scale factors are submitted for every frame but they might be shared
 * between the subframes of a channel. Scale factors are initially DPCM-coded.
 * Once scale factors are shared, the differences are transmitted as runlevel
 * codes.
 * Every subframe length and offset combination in the frame layout shares a
 * common quantization factor that can be adjusted for every channel by a
 * modifier.
 * After the inverse quantization, the coefficients get processed by an IMDCT.
 * The resulting values are then windowed with a sine window and the first half
 * of the values are added to the second half of the output from the previous
 * subframe in order to reconstruct the output samples.
 */

#include "avcodec.h"
#include "internal.h"
#include "get_bits.h"
#include "put_bits.h"
#include "wmaprodata.h"
#include "dsputil.h"
#include "wma.h"

/** current decoder limitations */
#define WMAPRO_MAX_CHANNELS    8                             ///< max number of handled channels
#define MAX_SUBFRAMES  32                                    ///< max number of subframes per channel
#define MAX_BANDS      29                                    ///< max number of scale factor bands
#define MAX_FRAMESIZE  32768                                 ///< maximum compressed frame size

#define WMAPRO_BLOCK_MAX_BITS 12                                           ///< log2 of max block size
#define WMAPRO_BLOCK_MAX_SIZE (1 << WMAPRO_BLOCK_MAX_BITS)                 ///< maximum block size
#define WMAPRO_BLOCK_SIZES    (WMAPRO_BLOCK_MAX_BITS - BLOCK_MIN_BITS + 1) ///< possible block sizes


#define VLCBITS            9
#define SCALEVLCBITS       8
#define VEC4MAXDEPTH    ((HUFF_VEC4_MAXBITS+VLCBITS-1)/VLCBITS)
#define VEC2MAXDEPTH    ((HUFF_VEC2_MAXBITS+VLCBITS-1)/VLCBITS)
#define VEC1MAXDEPTH    ((HUFF_VEC1_MAXBITS+VLCBITS-1)/VLCBITS)
#define SCALEMAXDEPTH   ((HUFF_SCALE_MAXBITS+SCALEVLCBITS-1)/SCALEVLCBITS)
#define SCALERLMAXDEPTH ((HUFF_SCALE_RL_MAXBITS+VLCBITS-1)/VLCBITS)

static VLC              sf_vlc;           ///< scale factor DPCM vlc
static VLC              sf_rl_vlc;        ///< scale factor run length vlc
static VLC              vec4_vlc;         ///< 4 coefficients per symbol
static VLC              vec2_vlc;         ///< 2 coefficients per symbol
static VLC              vec1_vlc;         ///< 1 coefficient per symbol
static VLC              coef_vlc[2];      ///< coefficient run length vlc codes
static float            sin64[33];        ///< sinus table for decorrelation

/**
 * @brief frame specific decoder context for a single channel
 */
typedef struct {
    int16_t  prev_block_len;                          ///< length of the previous block
    uint8_t  transmit_coefs;
    uint8_t  num_subframes;
    uint16_t subframe_len[MAX_SUBFRAMES];             ///< subframe length in samples
    uint16_t subframe_offset[MAX_SUBFRAMES];          ///< subframe positions in the current frame
    uint8_t  cur_subframe;                            ///< current subframe number
    uint16_t decoded_samples;                         ///< number of already processed samples
    uint8_t  grouped;                                 ///< channel is part of a group
    int      quant_step;                              ///< quantization step for the current subframe
    int8_t   reuse_sf;                                ///< share scale factors between subframes
    int8_t   scale_factor_step;                       ///< scaling step for the current subframe
    int      max_scale_factor;                        ///< maximum scale factor for the current subframe
    int      saved_scale_factors[2][MAX_BANDS];       ///< resampled and (previously) transmitted scale factor values
    int8_t   scale_factor_idx;                        ///< index for the transmitted scale factor values (used for resampling)
    int*     scale_factors;                           ///< pointer to the scale factor values used for decoding
    uint8_t  table_idx;                               ///< index in sf_offsets for the scale factor reference block
    float*   coeffs;                                  ///< pointer to the subframe decode buffer
    DECLARE_ALIGNED(16, float, out)[WMAPRO_BLOCK_MAX_SIZE + WMAPRO_BLOCK_MAX_SIZE / 2]; ///< output buffer
} WMAProChannelCtx;

/**
 * @brief channel group for channel transformations
 */
typedef struct {
    uint8_t num_channels;                                     ///< number of channels in the group
    int8_t  transform;                                        ///< transform on / off
    int8_t  transform_band[MAX_BANDS];                        ///< controls if the transform is enabled for a certain band
    float   decorrelation_matrix[WMAPRO_MAX_CHANNELS*WMAPRO_MAX_CHANNELS];
    float*  channel_data[WMAPRO_MAX_CHANNELS];                ///< transformation coefficients
} WMAProChannelGrp;

/**
 * @brief main decoder context
 */
typedef struct WMAProDecodeCtx {
    /* generic decoder variables */
    AVCodecContext*  avctx;                         ///< codec context for av_log
    DSPContext       dsp;                           ///< accelerated DSP functions
    uint8_t          frame_data[MAX_FRAMESIZE +
                      FF_INPUT_BUFFER_PADDING_SIZE];///< compressed frame data
    PutBitContext    pb;                            ///< context for filling the frame_data buffer
    FFTContext       mdct_ctx[WMAPRO_BLOCK_SIZES];  ///< MDCT context per block size
    DECLARE_ALIGNED(16, float, tmp)[WMAPRO_BLOCK_MAX_SIZE]; ///< IMDCT output buffer
    float*           windows[WMAPRO_BLOCK_SIZES];   ///< windows for the different block sizes

    /* frame size dependent frame information (set during initialization) */
    uint32_t         decode_flags;                  ///< used compression features
    uint8_t          len_prefix;                    ///< frame is prefixed with its length
    uint8_t          dynamic_range_compression;     ///< frame contains DRC data
    uint8_t          bits_per_sample;               ///< integer audio sample size for the unscaled IMDCT output (used to scale to [-1.0, 1.0])
    uint16_t         samples_per_frame;             ///< number of samples to output
    uint16_t         log2_frame_size;
    int8_t           num_channels;                  ///< number of channels in the stream (same as AVCodecContext.num_channels)
    int8_t           lfe_channel;                   ///< lfe channel index
    uint8_t          max_num_subframes;
    uint8_t          subframe_len_bits;             ///< number of bits used for the subframe length
    uint8_t          max_subframe_len_bit;          ///< flag indicating that the subframe is of maximum size when the first subframe length bit is 1
    uint16_t         min_samples_per_subframe;
    int8_t           num_sfb[WMAPRO_BLOCK_SIZES];   ///< scale factor bands per block size
    int16_t          sfb_offsets[WMAPRO_BLOCK_SIZES][MAX_BANDS];                    ///< scale factor band offsets (multiples of 4)
    int8_t           sf_offsets[WMAPRO_BLOCK_SIZES][WMAPRO_BLOCK_SIZES][MAX_BANDS]; ///< scale factor resample matrix
    int16_t          subwoofer_cutoffs[WMAPRO_BLOCK_SIZES]; ///< subwoofer cutoff values

    /* packet decode state */
    GetBitContext    pgb;                           ///< bitstream reader context for the packet
    uint8_t          packet_offset;                 ///< frame offset in the packet
    uint8_t          packet_sequence_number;        ///< current packet number
    int              num_saved_bits;                ///< saved number of bits
    int              frame_offset;                  ///< frame offset in the bit reservoir
    int              subframe_offset;               ///< subframe offset in the bit reservoir
    uint8_t          packet_loss;                   ///< set in case of bitstream error
    uint8_t          packet_done;                   ///< set when a packet is fully decoded

    /* frame decode state */
    uint32_t         frame_num;                     ///< current frame number (not used for decoding)
    GetBitContext    gb;                            ///< bitstream reader context
    int              buf_bit_size;                  ///< buffer size in bits
    float*           samples;                       ///< current samplebuffer pointer
    float*           samples_end;                   ///< maximum samplebuffer pointer
    uint8_t          drc_gain;                      ///< gain for the DRC tool
    int8_t           skip_frame;                    ///< skip output step
    int8_t           parsed_all_subframes;          ///< all subframes decoded?

    /* subframe/block decode state */
    int16_t          subframe_len;                  ///< current subframe length
    int8_t           channels_for_cur_subframe;     ///< number of channels that contain the subframe
    int8_t           channel_indexes_for_cur_subframe[WMAPRO_MAX_CHANNELS];
    int8_t           num_bands;                     ///< number of scale factor bands
    int16_t*         cur_sfb_offsets;               ///< sfb offsets for the current block
    uint8_t          table_idx;                     ///< index for the num_sfb, sfb_offsets, sf_offsets and subwoofer_cutoffs tables
    int8_t           esc_len;                       ///< length of escaped coefficients

    uint8_t          num_chgroups;                  ///< number of channel groups
    WMAProChannelGrp chgroup[WMAPRO_MAX_CHANNELS];  ///< channel group information

    WMAProChannelCtx channel[WMAPRO_MAX_CHANNELS];  ///< per channel data
} WMAProDecodeCtx;


/**
 *@brief helper function to print the most important members of the context
 *@param s context
 */
static void av_cold dump_context(WMAProDecodeCtx *s)
{
#define PRINT(a, b)     av_log(s->avctx, AV_LOG_DEBUG, " %s = %d\n", a, b);
#define PRINT_HEX(a, b) av_log(s->avctx, AV_LOG_DEBUG, " %s = %x\n", a, b);

    PRINT("ed sample bit depth", s->bits_per_sample);
    PRINT_HEX("ed decode flags", s->decode_flags);
    PRINT("samples per frame",   s->samples_per_frame);
    PRINT("log2 frame size",     s->log2_frame_size);
    PRINT("max num subframes",   s->max_num_subframes);
    PRINT("len prefix",          s->len_prefix);
    PRINT("num channels",        s->num_channels);
}

/**
 *@brief Uninitialize the decoder and free all resources.
 *@param avctx codec context
 *@return 0 on success, < 0 otherwise
 */
static av_cold int decode_end(AVCodecContext *avctx)
{
    WMAProDecodeCtx *s = avctx->priv_data;
    int i;

    for (i = 0; i < WMAPRO_BLOCK_SIZES; i++)
        ff_mdct_end(&s->mdct_ctx[i]);

    return 0;
}

/**
 *@brief Initialize the decoder.
 *@param avctx codec context
 *@return 0 on success, -1 otherwise
 */
static av_cold int decode_init(AVCodecContext *avctx)
{
    WMAProDecodeCtx *s = avctx->priv_data;
    uint8_t *edata_ptr = avctx->extradata;
    unsigned int channel_mask;
    int i;
    int log2_max_num_subframes;
    int num_possible_block_sizes;

    s->avctx = avctx;
    dsputil_init(&s->dsp, avctx);
    init_put_bits(&s->pb, s->frame_data, MAX_FRAMESIZE);

    avctx->sample_fmt = SAMPLE_FMT_FLT;

    if (avctx->extradata_size >= 18) {
        s->decode_flags    = AV_RL16(edata_ptr+14);
        channel_mask       = AV_RL32(edata_ptr+2);
        s->bits_per_sample = AV_RL16(edata_ptr);
        /** dump the extradata */
        for (i = 0; i < avctx->extradata_size; i++)
            dprintf(avctx, "[%x] ", avctx->extradata[i]);
        dprintf(avctx, "\n");

    } else {
        av_log_ask_for_sample(avctx, "Unknown extradata size\n");
        return AVERROR_INVALIDDATA;
    }

    /** generic init */
    s->log2_frame_size = av_log2(avctx->block_align) + 4;

    /** frame info */
    s->skip_frame  = 1; /** skip first frame */
    s->packet_loss = 1;
    s->len_prefix  = (s->decode_flags & 0x40);

    if (!s->len_prefix) {
        av_log_ask_for_sample(avctx, "no length prefix\n");
        return AVERROR_INVALIDDATA;
    }

    /** get frame len */
    s->samples_per_frame = 1 << ff_wma_get_frame_len_bits(avctx->sample_rate,
                                                          3, s->decode_flags);

    /** init previous block len */
    for (i = 0; i < avctx->channels; i++)
        s->channel[i].prev_block_len = s->samples_per_frame;

    /** subframe info */
    log2_max_num_subframes       = ((s->decode_flags & 0x38) >> 3);
    s->max_num_subframes         = 1 << log2_max_num_subframes;
    if (s->max_num_subframes == 16)
        s->max_subframe_len_bit = 1;
    s->subframe_len_bits = av_log2(log2_max_num_subframes) + 1;

    num_possible_block_sizes     = log2_max_num_subframes + 1;
    s->min_samples_per_subframe  = s->samples_per_frame / s->max_num_subframes;
    s->dynamic_range_compression = (s->decode_flags & 0x80);

    if (s->max_num_subframes > MAX_SUBFRAMES) {
        av_log(avctx, AV_LOG_ERROR, "invalid number of subframes %i\n",
               s->max_num_subframes);
        return AVERROR_INVALIDDATA;
    }

    s->num_channels = avctx->channels;

    /** extract lfe channel position */
    s->lfe_channel = -1;

    if (channel_mask & 8) {
        unsigned int mask;
        for (mask = 1; mask < 16; mask <<= 1) {
            if (channel_mask & mask)
                ++s->lfe_channel;
        }
    }

    if (s->num_channels < 0) {
        av_log(avctx, AV_LOG_ERROR, "invalid number of channels %d\n", s->num_channels);
        return AVERROR_INVALIDDATA;
    } else if (s->num_channels > WMAPRO_MAX_CHANNELS) {
        av_log_ask_for_sample(avctx, "unsupported number of channels\n");
        return AVERROR_PATCHWELCOME;
    }

    INIT_VLC_STATIC(&sf_vlc, SCALEVLCBITS, HUFF_SCALE_SIZE,
                    scale_huffbits, 1, 1,
                    scale_huffcodes, 2, 2, 616);

    INIT_VLC_STATIC(&sf_rl_vlc, VLCBITS, HUFF_SCALE_RL_SIZE,
                    scale_rl_huffbits, 1, 1,
                    scale_rl_huffcodes, 4, 4, 1406);

    INIT_VLC_STATIC(&coef_vlc[0], VLCBITS, HUFF_COEF0_SIZE,
                    coef0_huffbits, 1, 1,
                    coef0_huffcodes, 4, 4, 2108);

    INIT_VLC_STATIC(&coef_vlc[1], VLCBITS, HUFF_COEF1_SIZE,
                    coef1_huffbits, 1, 1,
                    coef1_huffcodes, 4, 4, 3912);

    INIT_VLC_STATIC(&vec4_vlc, VLCBITS, HUFF_VEC4_SIZE,
                    vec4_huffbits, 1, 1,
                    vec4_huffcodes, 2, 2, 604);

    INIT_VLC_STATIC(&vec2_vlc, VLCBITS, HUFF_VEC2_SIZE,
                    vec2_huffbits, 1, 1,
                    vec2_huffcodes, 2, 2, 562);

    INIT_VLC_STATIC(&vec1_vlc, VLCBITS, HUFF_VEC1_SIZE,
                    vec1_huffbits, 1, 1,
                    vec1_huffcodes, 2, 2, 562);

    /** calculate number of scale factor bands and their offsets
        for every possible block size */
    for (i = 0; i < num_possible_block_sizes; i++) {
        int subframe_len = s->samples_per_frame >> i;
        int x;
        int band = 1;

        s->sfb_offsets[i][0] = 0;

        for (x = 0; x < MAX_BANDS-1 && s->sfb_offsets[i][band - 1] < subframe_len; x++) {
            int offset = (subframe_len * 2 * critical_freq[x])
                          / s->avctx->sample_rate + 2;
            offset &= ~3;
            if (offset > s->sfb_offsets[i][band - 1])
                s->sfb_offsets[i][band++] = offset;
        }
        s->sfb_offsets[i][band - 1] = subframe_len;
        s->num_sfb[i]               = band - 1;
    }


    /** Scale factors can be shared between blocks of different size
        as every block has a different scale factor band layout.
        The matrix sf_offsets is needed to find the correct scale factor.
     */

    for (i = 0; i < num_possible_block_sizes; i++) {
        int b;
        for (b = 0; b < s->num_sfb[i]; b++) {
            int x;
            int offset = ((s->sfb_offsets[i][b]
                           + s->sfb_offsets[i][b + 1] - 1) << i) >> 1;
            for (x = 0; x < num_possible_block_sizes; x++) {
                int v = 0;
                while (s->sfb_offsets[x][v + 1] << x < offset)
                    ++v;
                s->sf_offsets[i][x][b] = v;
            }
        }
    }

    /** init MDCT, FIXME: only init needed sizes */
    for (i = 0; i < WMAPRO_BLOCK_SIZES; i++)
        ff_mdct_init(&s->mdct_ctx[i], BLOCK_MIN_BITS+1+i, 1,
                     1.0 / (1 << (BLOCK_MIN_BITS + i - 1))
                     / (1 << (s->bits_per_sample - 1)));

    /** init MDCT windows: simple sinus window */
    for (i = 0; i < WMAPRO_BLOCK_SIZES; i++) {
        const int win_idx = WMAPRO_BLOCK_MAX_BITS - i;
        ff_init_ff_sine_windows(win_idx);
        s->windows[WMAPRO_BLOCK_SIZES - i - 1] = ff_sine_windows[win_idx];
    }

    /** calculate subwoofer cutoff values */
    for (i = 0; i < num_possible_block_sizes; i++) {
        int block_size = s->samples_per_frame >> i;
        int cutoff = (440*block_size + 3 * (s->avctx->sample_rate >> 1) - 1)
                     / s->avctx->sample_rate;
        s->subwoofer_cutoffs[i] = av_clip(cutoff, 4, block_size);
    }

    /** calculate sine values for the decorrelation matrix */
    for (i = 0; i < 33; i++)
        sin64[i] = sin(i*M_PI / 64.0);

    if (avctx->debug & FF_DEBUG_BITSTREAM)
        dump_context(s);

    avctx->channel_layout = channel_mask;
    return 0;
}

/**
 *@brief Decode the subframe length.
 *@param s context
 *@param offset sample offset in the frame
 *@return decoded subframe length on success, < 0 in case of an error
 */
static int decode_subframe_length(WMAProDecodeCtx *s, int offset)
{
    int frame_len_shift = 0;
    int subframe_len;

    /** no need to read from the bitstream when only one length is possible */
    if (offset == s->samples_per_frame - s->min_samples_per_subframe)
        return s->min_samples_per_subframe;

    /** 1 bit indicates if the subframe is of maximum length */
    if (s->max_subframe_len_bit) {
        if (get_bits1(&s->gb))
            frame_len_shift = 1 + get_bits(&s->gb, s->subframe_len_bits-1);
    } else
        frame_len_shift = get_bits(&s->gb, s->subframe_len_bits);

    subframe_len = s->samples_per_frame >> frame_len_shift;

    /** sanity check the length */
    if (subframe_len < s->min_samples_per_subframe ||
        subframe_len > s->samples_per_frame) {
        av_log(s->avctx, AV_LOG_ERROR, "broken frame: subframe_len %i\n",
               subframe_len);
        return AVERROR_INVALIDDATA;
    }
    return subframe_len;
}

/**
 *@brief Decode how the data in the frame is split into subframes.
 *       Every WMA frame contains the encoded data for a fixed number of
 *       samples per channel. The data for every channel might be split
 *       into several subframes. This function will reconstruct the list of
 *       subframes for every channel.
 *
 *       If the subframes are not evenly split, the algorithm estimates the
 *       channels with the lowest number of total samples.
 *       Afterwards, for each of these channels a bit is read from the
 *       bitstream that indicates if the channel contains a subframe with the
 *       next subframe size that is going to be read from the bitstream or not.
 *       If a channel contains such a subframe, the subframe size gets added to
 *       the channel's subframe list.
 *       The algorithm repeats these steps until the frame is properly divided
 *       between the individual channels.
 *
 *@param s context
 *@return 0 on success, < 0 in case of an error
 */
static int decode_tilehdr(WMAProDecodeCtx *s)
{
    uint16_t num_samples[WMAPRO_MAX_CHANNELS];        /** sum of samples for all currently known subframes of a channel */
    uint8_t  contains_subframe[WMAPRO_MAX_CHANNELS];  /** flag indicating if a channel contains the current subframe */
    int channels_for_cur_subframe = s->num_channels;  /** number of channels that contain the current subframe */
    int fixed_channel_layout = 0;                     /** flag indicating that all channels use the same subframe offsets and sizes */
    int min_channel_len = 0;                          /** smallest sum of samples (channels with this length will be processed first) */
    int c;

    /* Should never consume more than 3073 bits (256 iterations for the
     * while loop when always the minimum amount of 128 samples is substracted
     * from missing samples in the 8 channel case).
     * 1 + BLOCK_MAX_SIZE * MAX_CHANNELS / BLOCK_MIN_SIZE * (MAX_CHANNELS  + 4)
     */

    /** reset tiling information */
    for (c = 0; c < s->num_channels; c++)
        s->channel[c].num_subframes = 0;

    memset(num_samples, 0, sizeof(num_samples));

    if (s->max_num_subframes == 1 || get_bits1(&s->gb))
        fixed_channel_layout = 1;

    /** loop until the frame data is split between the subframes */
    do {
        int subframe_len;

        /** check which channels contain the subframe */
        for (c = 0; c < s->num_channels; c++) {
            if (num_samples[c] == min_channel_len) {
                if (fixed_channel_layout || channels_for_cur_subframe == 1 ||
                   (min_channel_len == s->samples_per_frame - s->min_samples_per_subframe))
                    contains_subframe[c] = 1;
                else
                    contains_subframe[c] = get_bits1(&s->gb);
            } else
                contains_subframe[c] = 0;
        }

        /** get subframe length, subframe_len == 0 is not allowed */
        if ((subframe_len = decode_subframe_length(s, min_channel_len)) <= 0)
            return AVERROR_INVALIDDATA;

        /** add subframes to the individual channels and find new min_channel_len */
        min_channel_len += subframe_len;
        for (c = 0; c < s->num_channels; c++) {
            WMAProChannelCtx* chan = &s->channel[c];

            if (contains_subframe[c]) {
                if (chan->num_subframes >= MAX_SUBFRAMES) {
                    av_log(s->avctx, AV_LOG_ERROR,
                           "broken frame: num subframes > 31\n");
                    return AVERROR_INVALIDDATA;
                }
                chan->subframe_len[chan->num_subframes] = subframe_len;
                num_samples[c] += subframe_len;
                ++chan->num_subframes;
                if (num_samples[c] > s->samples_per_frame) {
                    av_log(s->avctx, AV_LOG_ERROR, "broken frame: "
                           "channel len > samples_per_frame\n");
                    return AVERROR_INVALIDDATA;
                }
            } else if (num_samples[c] <= min_channel_len) {
                if (num_samples[c] < min_channel_len) {
                    channels_for_cur_subframe = 0;
                    min_channel_len = num_samples[c];
                }
                ++channels_for_cur_subframe;
            }
        }
    } while (min_channel_len < s->samples_per_frame);

    for (c = 0; c < s->num_channels; c++) {
        int i;
        int offset = 0;
        for (i = 0; i < s->channel[c].num_subframes; i++) {
            dprintf(s->avctx, "frame[%i] channel[%i] subframe[%i]"
                    " len %i\n", s->frame_num, c, i,
                    s->channel[c].subframe_len[i]);
            s->channel[c].subframe_offset[i] = offset;
            offset += s->channel[c].subframe_len[i];
        }
    }

    return 0;
}

/**
 *@brief Calculate a decorrelation matrix from the bitstream parameters.
 *@param s codec context
 *@param chgroup channel group for which the matrix needs to be calculated
 */
static void decode_decorrelation_matrix(WMAProDecodeCtx *s,
                                        WMAProChannelGrp *chgroup)
{
    int i;
    int offset = 0;
    int8_t rotation_offset[WMAPRO_MAX_CHANNELS * WMAPRO_MAX_CHANNELS];
    memset(chgroup->decorrelation_matrix, 0, s->num_channels *
           s->num_channels * sizeof(*chgroup->decorrelation_matrix));

    for (i = 0; i < chgroup->num_channels * (chgroup->num_channels - 1) >> 1; i++)
        rotation_offset[i] = get_bits(&s->gb, 6);

    for (i = 0; i < chgroup->num_channels; i++)
        chgroup->decorrelation_matrix[chgroup->num_channels * i + i] =
            get_bits1(&s->gb) ? 1.0 : -1.0;

    for (i = 1; i < chgroup->num_channels; i++) {
        int x;
        for (x = 0; x < i; x++) {
            int y;
            for (y = 0; y < i + 1; y++) {
                float v1 = chgroup->decorrelation_matrix[x * chgroup->num_channels + y];
                float v2 = chgroup->decorrelation_matrix[i * chgroup->num_channels + y];
                int n = rotation_offset[offset + x];
                float sinv;
                float cosv;

                if (n < 32) {
                    sinv = sin64[n];
                    cosv = sin64[32 - n];
                } else {
                    sinv =  sin64[64 -  n];
                    cosv = -sin64[n  - 32];
                }

                chgroup->decorrelation_matrix[y + x * chgroup->num_channels] =
                                               (v1 * sinv) - (v2 * cosv);
                chgroup->decorrelation_matrix[y + i * chgroup->num_channels] =
                                               (v1 * cosv) + (v2 * sinv);
            }
        }
        offset += i;
    }
}

/**
 *@brief Decode channel transformation parameters
 *@param s codec context
 *@return 0 in case of success, < 0 in case of bitstream errors
 */
static int decode_channel_transform(WMAProDecodeCtx* s)
{
    int i;
    /* should never consume more than 1921 bits for the 8 channel case
     * 1 + MAX_CHANNELS * (MAX_CHANNELS + 2 + 3 * MAX_CHANNELS * MAX_CHANNELS
     * + MAX_CHANNELS + MAX_BANDS + 1)
     */

    /** in the one channel case channel transforms are pointless */
    s->num_chgroups = 0;
    if (s->num_channels > 1) {
        int remaining_channels = s->channels_for_cur_subframe;

        if (get_bits1(&s->gb)) {
            av_log_ask_for_sample(s->avctx,
                                  "unsupported channel transform bit\n");
            return AVERROR_INVALIDDATA;
        }

        for (s->num_chgroups = 0; remaining_channels &&
             s->num_chgroups < s->channels_for_cur_subframe; s->num_chgroups++) {
            WMAProChannelGrp* chgroup = &s->chgroup[s->num_chgroups];
            float** channel_data = chgroup->channel_data;
            chgroup->num_channels = 0;
            chgroup->transform = 0;

            /** decode channel mask */
            if (remaining_channels > 2) {
                for (i = 0; i < s->channels_for_cur_subframe; i++) {
                    int channel_idx = s->channel_indexes_for_cur_subframe[i];
                    if (!s->channel[channel_idx].grouped
                        && get_bits1(&s->gb)) {
                        ++chgroup->num_channels;
                        s->channel[channel_idx].grouped = 1;
                        *channel_data++ = s->channel[channel_idx].coeffs;
                    }
                }
            } else {
                chgroup->num_channels = remaining_channels;
                for (i = 0; i < s->channels_for_cur_subframe; i++) {
                    int channel_idx = s->channel_indexes_for_cur_subframe[i];
                    if (!s->channel[channel_idx].grouped)
                        *channel_data++ = s->channel[channel_idx].coeffs;
                    s->channel[channel_idx].grouped = 1;
                }
            }

            /** decode transform type */
            if (chgroup->num_channels == 2) {
                if (get_bits1(&s->gb)) {
                    if (get_bits1(&s->gb)) {
                        av_log_ask_for_sample(s->avctx,
                                              "unsupported channel transform type\n");
                    }
                } else {
                    chgroup->transform = 1;
                    if (s->num_channels == 2) {
                        chgroup->decorrelation_matrix[0] =  1.0;
                        chgroup->decorrelation_matrix[1] = -1.0;
                        chgroup->decorrelation_matrix[2] =  1.0;
                        chgroup->decorrelation_matrix[3] =  1.0;
                    } else {
                        /** cos(pi/4) */
                        chgroup->decorrelation_matrix[0] =  0.70703125;
                        chgroup->decorrelation_matrix[1] = -0.70703125;
                        chgroup->decorrelation_matrix[2] =  0.70703125;
                        chgroup->decorrelation_matrix[3] =  0.70703125;
                    }
                }
            } else if (chgroup->num_channels > 2) {
                if (get_bits1(&s->gb)) {
                    chgroup->transform = 1;
                    if (get_bits1(&s->gb)) {
                        decode_decorrelation_matrix(s, chgroup);
                    } else {
                        /** FIXME: more than 6 coupled channels not supported */
                        if (chgroup->num_channels > 6) {
                            av_log_ask_for_sample(s->avctx,
                                                  "coupled channels > 6\n");
                        } else {
                            memcpy(chgroup->decorrelation_matrix,
                                   default_decorrelation[chgroup->num_channels],
                                   chgroup->num_channels * chgroup->num_channels *
                                   sizeof(*chgroup->decorrelation_matrix));
                        }
                    }
                }
            }

            /** decode transform on / off */
            if (chgroup->transform) {
                if (!get_bits1(&s->gb)) {
                    int i;
                    /** transform can be enabled for individual bands */
                    for (i = 0; i < s->num_bands; i++) {
                        chgroup->transform_band[i] = get_bits1(&s->gb);
                    }
                } else {
                    memset(chgroup->transform_band, 1, s->num_bands);
                }
            }
            remaining_channels -= chgroup->num_channels;
        }
    }
    return 0;
}

/**
 *@brief Extract the coefficients from the bitstream.
 *@param s codec context
 *@param c current channel number
 *@return 0 on success, < 0 in case of bitstream errors
 */
static int decode_coeffs(WMAProDecodeCtx *s, int c)
{
    /* Integers 0..15 as single-precision floats.  The table saves a
       costly int to float conversion, and storing the values as
       integers allows fast sign-flipping. */
    static const int fval_tab[16] = {
        0x00000000, 0x3f800000, 0x40000000, 0x40400000,
        0x40800000, 0x40a00000, 0x40c00000, 0x40e00000,
        0x41000000, 0x41100000, 0x41200000, 0x41300000,
        0x41400000, 0x41500000, 0x41600000, 0x41700000,
    };
    int vlctable;
    VLC* vlc;
    WMAProChannelCtx* ci = &s->channel[c];
    int rl_mode = 0;
    int cur_coeff = 0;
    int num_zeros = 0;
    const uint16_t* run;
    const float* level;

    dprintf(s->avctx, "decode coefficients for channel %i\n", c);

    vlctable = get_bits1(&s->gb);
    vlc = &coef_vlc[vlctable];

    if (vlctable) {
        run = coef1_run;
        level = coef1_level;
    } else {
        run = coef0_run;
        level = coef0_level;
    }

    /** decode vector coefficients (consumes up to 167 bits per iteration for
      4 vector coded large values) */
    while (!rl_mode && cur_coeff + 3 < s->subframe_len) {
        int vals[4];
        int i;
        unsigned int idx;

        idx = get_vlc2(&s->gb, vec4_vlc.table, VLCBITS, VEC4MAXDEPTH);

        if (idx == HUFF_VEC4_SIZE - 1) {
            for (i = 0; i < 4; i += 2) {
                idx = get_vlc2(&s->gb, vec2_vlc.table, VLCBITS, VEC2MAXDEPTH);
                if (idx == HUFF_VEC2_SIZE - 1) {
                    int v0, v1;
                    v0 = get_vlc2(&s->gb, vec1_vlc.table, VLCBITS, VEC1MAXDEPTH);
                    if (v0 == HUFF_VEC1_SIZE - 1)
                        v0 += ff_wma_get_large_val(&s->gb);
                    v1 = get_vlc2(&s->gb, vec1_vlc.table, VLCBITS, VEC1MAXDEPTH);
                    if (v1 == HUFF_VEC1_SIZE - 1)
                        v1 += ff_wma_get_large_val(&s->gb);
                    ((float*)vals)[i  ] = v0;
                    ((float*)vals)[i+1] = v1;
                } else {
                    vals[i]   = fval_tab[symbol_to_vec2[idx] >> 4 ];
                    vals[i+1] = fval_tab[symbol_to_vec2[idx] & 0xF];
                }
            }
        } else {
            vals[0] = fval_tab[ symbol_to_vec4[idx] >> 12      ];
            vals[1] = fval_tab[(symbol_to_vec4[idx] >> 8) & 0xF];
            vals[2] = fval_tab[(symbol_to_vec4[idx] >> 4) & 0xF];
            vals[3] = fval_tab[ symbol_to_vec4[idx]       & 0xF];
        }

        /** decode sign */
        for (i = 0; i < 4; i++) {
            if (vals[i]) {
                int sign = get_bits1(&s->gb) - 1;
                *(uint32_t*)&ci->coeffs[cur_coeff] = vals[i] ^ sign<<31;
                num_zeros = 0;
            } else {
                ci->coeffs[cur_coeff] = 0;
                /** switch to run level mode when subframe_len / 128 zeros
                    were found in a row */
                rl_mode |= (++num_zeros > s->subframe_len >> 8);
            }
            ++cur_coeff;
        }
    }

    /** decode run level coded coefficients */
    if (rl_mode) {
        memset(&ci->coeffs[cur_coeff], 0,
               sizeof(*ci->coeffs) * (s->subframe_len - cur_coeff));
        if (ff_wma_run_level_decode(s->avctx, &s->gb, vlc,
                                    level, run, 1, ci->coeffs,
                                    cur_coeff, s->subframe_len,
                                    s->subframe_len, s->esc_len, 0))
            return AVERROR_INVALIDDATA;
    }

    return 0;
}

/**
 *@brief Extract scale factors from the bitstream.
 *@param s codec context
 *@return 0 on success, < 0 in case of bitstream errors
 */
static int decode_scale_factors(WMAProDecodeCtx* s)
{
    int i;

    /** should never consume more than 5344 bits
     *  MAX_CHANNELS * (1 +  MAX_BANDS * 23)
     */

    for (i = 0; i < s->channels_for_cur_subframe; i++) {
        int c = s->channel_indexes_for_cur_subframe[i];
        int* sf;
        int* sf_end;
        s->channel[c].scale_factors = s->channel[c].saved_scale_factors[!s->channel[c].scale_factor_idx];
        sf_end = s->channel[c].scale_factors + s->num_bands;

        /** resample scale factors for the new block size
         *  as the scale factors might need to be resampled several times
         *  before some  new values are transmitted, a backup of the last
         *  transmitted scale factors is kept in saved_scale_factors
         */
        if (s->channel[c].reuse_sf) {
            const int8_t* sf_offsets = s->sf_offsets[s->table_idx][s->channel[c].table_idx];
            int b;
            for (b = 0; b < s->num_bands; b++)
                s->channel[c].scale_factors[b] =
                    s->channel[c].saved_scale_factors[s->channel[c].scale_factor_idx][*sf_offsets++];
        }

        if (!s->channel[c].cur_subframe || get_bits1(&s->gb)) {

            if (!s->channel[c].reuse_sf) {
                int val;
                /** decode DPCM coded scale factors */
                s->channel[c].scale_factor_step = get_bits(&s->gb, 2) + 1;
                val = 45 / s->channel[c].scale_factor_step;
                for (sf = s->channel[c].scale_factors; sf < sf_end; sf++) {
                    val += get_vlc2(&s->gb, sf_vlc.table, SCALEVLCBITS, SCALEMAXDEPTH) - 60;
                    *sf = val;
                }
            } else {
                int i;
                /** run level decode differences to the resampled factors */
                for (i = 0; i < s->num_bands; i++) {
                    int idx;
                    int skip;
                    int val;
                    int sign;

                    idx = get_vlc2(&s->gb, sf_rl_vlc.table, VLCBITS, SCALERLMAXDEPTH);

                    if (!idx) {
                        uint32_t code = get_bits(&s->gb, 14);
                        val  =  code >> 6;
                        sign = (code & 1) - 1;
                        skip = (code & 0x3f) >> 1;
                    } else if (idx == 1) {
                        break;
                    } else {
                        skip = scale_rl_run[idx];
                        val  = scale_rl_level[idx];
                        sign = get_bits1(&s->gb)-1;
                    }

                    i += skip;
                    if (i >= s->num_bands) {
                        av_log(s->avctx, AV_LOG_ERROR,
                               "invalid scale factor coding\n");
                        return AVERROR_INVALIDDATA;
                    }
                    s->channel[c].scale_factors[i] += (val ^ sign) - sign;
                }
            }
            /** swap buffers */
            s->channel[c].scale_factor_idx = !s->channel[c].scale_factor_idx;
            s->channel[c].table_idx = s->table_idx;
            s->channel[c].reuse_sf  = 1;
        }

        /** calculate new scale factor maximum */
        s->channel[c].max_scale_factor = s->channel[c].scale_factors[0];
        for (sf = s->channel[c].scale_factors + 1; sf < sf_end; sf++) {
            s->channel[c].max_scale_factor =
                FFMAX(s->channel[c].max_scale_factor, *sf);
        }

    }
    return 0;
}

/**
 *@brief Reconstruct the individual channel data.
 *@param s codec context
 */
static void inverse_channel_transform(WMAProDecodeCtx *s)
{
    int i;

    for (i = 0; i < s->num_chgroups; i++) {
        if (s->chgroup[i].transform) {
            float data[WMAPRO_MAX_CHANNELS];
            const int num_channels = s->chgroup[i].num_channels;
            float** ch_data = s->chgroup[i].channel_data;
            float** ch_end = ch_data + num_channels;
            const int8_t* tb = s->chgroup[i].transform_band;
            int16_t* sfb;

            /** multichannel decorrelation */
            for (sfb = s->cur_sfb_offsets;
                 sfb < s->cur_sfb_offsets + s->num_bands; sfb++) {
                int y;
                if (*tb++ == 1) {
                    /** multiply values with the decorrelation_matrix */
                    for (y = sfb[0]; y < FFMIN(sfb[1], s->subframe_len); y++) {
                        const float* mat = s->chgroup[i].decorrelation_matrix;
                        const float* data_end = data + num_channels;
                        float* data_ptr = data;
                        float** ch;

                        for (ch = ch_data; ch < ch_end; ch++)
                            *data_ptr++ = (*ch)[y];

                        for (ch = ch_data; ch < ch_end; ch++) {
                            float sum = 0;
                            data_ptr = data;
                            while (data_ptr < data_end)
                                sum += *data_ptr++ * *mat++;

                            (*ch)[y] = sum;
                        }
                    }
                } else if (s->num_channels == 2) {
                    int len = FFMIN(sfb[1], s->subframe_len) - sfb[0];
                    s->dsp.vector_fmul_scalar(ch_data[0] + sfb[0],
                                              ch_data[0] + sfb[0],
                                              181.0 / 128, len);
                    s->dsp.vector_fmul_scalar(ch_data[1] + sfb[0],
                                              ch_data[1] + sfb[0],
                                              181.0 / 128, len);
                }
            }
        }
    }
}

/**
 *@brief Apply sine window and reconstruct the output buffer.
 *@param s codec context
 */
static void wmapro_window(WMAProDecodeCtx *s)
{
    int i;
    for (i = 0; i < s->channels_for_cur_subframe; i++) {
        int c = s->channel_indexes_for_cur_subframe[i];
        float* window;
        int winlen = s->channel[c].prev_block_len;
        float* start = s->channel[c].coeffs - (winlen >> 1);

        if (s->subframe_len < winlen) {
            start += (winlen - s->subframe_len) >> 1;
            winlen = s->subframe_len;
        }

        window = s->windows[av_log2(winlen) - BLOCK_MIN_BITS];

        winlen >>= 1;

        s->dsp.vector_fmul_window(start, start, start + winlen,
                                  window, 0, winlen);

        s->channel[c].prev_block_len = s->subframe_len;
    }
}

/**
 *@brief Decode a single subframe (block).
 *@param s codec context
 *@return 0 on success, < 0 when decoding failed
 */
static int decode_subframe(WMAProDecodeCtx *s)
{
    int offset = s->samples_per_frame;
    int subframe_len = s->samples_per_frame;
    int i;
    int total_samples   = s->samples_per_frame * s->num_channels;
    int transmit_coeffs = 0;
    int cur_subwoofer_cutoff;

    s->subframe_offset = get_bits_count(&s->gb);

    /** reset channel context and find the next block offset and size
        == the next block of the channel with the smallest number of
        decoded samples
    */
    for (i = 0; i < s->num_channels; i++) {
        s->channel[i].grouped = 0;
        if (offset > s->channel[i].decoded_samples) {
            offset = s->channel[i].decoded_samples;
            subframe_len =
                s->channel[i].subframe_len[s->channel[i].cur_subframe];
        }
    }

    dprintf(s->avctx,
            "processing subframe with offset %i len %i\n", offset, subframe_len);

    /** get a list of all channels that contain the estimated block */
    s->channels_for_cur_subframe = 0;
    for (i = 0; i < s->num_channels; i++) {
        const int cur_subframe = s->channel[i].cur_subframe;
        /** substract already processed samples */
        total_samples -= s->channel[i].decoded_samples;

        /** and count if there are multiple subframes that match our profile */
        if (offset == s->channel[i].decoded_samples &&
            subframe_len == s->channel[i].subframe_len[cur_subframe]) {
            total_samples -= s->channel[i].subframe_len[cur_subframe];
            s->channel[i].decoded_samples +=
                s->channel[i].subframe_len[cur_subframe];
            s->channel_indexes_for_cur_subframe[s->channels_for_cur_subframe] = i;
            ++s->channels_for_cur_subframe;
        }
    }

    /** check if the frame will be complete after processing the
        estimated block */
    if (!total_samples)
        s->parsed_all_subframes = 1;


    dprintf(s->avctx, "subframe is part of %i channels\n",
            s->channels_for_cur_subframe);

    /** calculate number of scale factor bands and their offsets */
    s->table_idx         = av_log2(s->samples_per_frame/subframe_len);
    s->num_bands         = s->num_sfb[s->table_idx];
    s->cur_sfb_offsets   = s->sfb_offsets[s->table_idx];
    cur_subwoofer_cutoff = s->subwoofer_cutoffs[s->table_idx];

    /** configure the decoder for the current subframe */
    for (i = 0; i < s->channels_for_cur_subframe; i++) {
        int c = s->channel_indexes_for_cur_subframe[i];

        s->channel[c].coeffs = &s->channel[c].out[(s->samples_per_frame >> 1)
                                                  + offset];
    }

    s->subframe_len = subframe_len;
    s->esc_len = av_log2(s->subframe_len - 1) + 1;

    /** skip extended header if any */
    if (get_bits1(&s->gb)) {
        int num_fill_bits;
        if (!(num_fill_bits = get_bits(&s->gb, 2))) {
            int len = get_bits(&s->gb, 4);
            num_fill_bits = get_bits(&s->gb, len) + 1;
        }

        if (num_fill_bits >= 0) {
            if (get_bits_count(&s->gb) + num_fill_bits > s->num_saved_bits) {
                av_log(s->avctx, AV_LOG_ERROR, "invalid number of fill bits\n");
                return AVERROR_INVALIDDATA;
            }

            skip_bits_long(&s->gb, num_fill_bits);
        }
    }

    /** no idea for what the following bit is used */
    if (get_bits1(&s->gb)) {
        av_log_ask_for_sample(s->avctx, "reserved bit set\n");
        return AVERROR_INVALIDDATA;
    }


    if (decode_channel_transform(s) < 0)
        return AVERROR_INVALIDDATA;


    for (i = 0; i < s->channels_for_cur_subframe; i++) {
        int c = s->channel_indexes_for_cur_subframe[i];
        if ((s->channel[c].transmit_coefs = get_bits1(&s->gb)))
            transmit_coeffs = 1;
    }

    if (transmit_coeffs) {
        int step;
        int quant_step = 90 * s->bits_per_sample >> 4;
        if ((get_bits1(&s->gb))) {
            /** FIXME: might change run level mode decision */
            av_log_ask_for_sample(s->avctx, "unsupported quant step coding\n");
            return AVERROR_INVALIDDATA;
        }
        /** decode quantization step */
        step = get_sbits(&s->gb, 6);
        quant_step += step;
        if (step == -32 || step == 31) {
            const int sign = (step == 31) - 1;
            int quant = 0;
            while (get_bits_count(&s->gb) + 5 < s->num_saved_bits &&
                   (step = get_bits(&s->gb, 5)) == 31) {
                quant += 31;
            }
            quant_step += ((quant + step) ^ sign) - sign;
        }
        if (quant_step < 0) {
            av_log(s->avctx, AV_LOG_DEBUG, "negative quant step\n");
        }

        /** decode quantization step modifiers for every channel */

        if (s->channels_for_cur_subframe == 1) {
            s->channel[s->channel_indexes_for_cur_subframe[0]].quant_step = quant_step;
        } else {
            int modifier_len = get_bits(&s->gb, 3);
            for (i = 0; i < s->channels_for_cur_subframe; i++) {
                int c = s->channel_indexes_for_cur_subframe[i];
                s->channel[c].quant_step = quant_step;
                if (get_bits1(&s->gb)) {
                    if (modifier_len) {
                        s->channel[c].quant_step += get_bits(&s->gb, modifier_len) + 1;
                    } else
                        ++s->channel[c].quant_step;
                }
            }
        }

        /** decode scale factors */
        if (decode_scale_factors(s) < 0)
            return AVERROR_INVALIDDATA;
    }

    dprintf(s->avctx, "BITSTREAM: subframe header length was %i\n",
            get_bits_count(&s->gb) - s->subframe_offset);

    /** parse coefficients */
    for (i = 0; i < s->channels_for_cur_subframe; i++) {
        int c = s->channel_indexes_for_cur_subframe[i];
        if (s->channel[c].transmit_coefs &&
            get_bits_count(&s->gb) < s->num_saved_bits) {
            decode_coeffs(s, c);
        } else
            memset(s->channel[c].coeffs, 0,
                   sizeof(*s->channel[c].coeffs) * subframe_len);
    }

    dprintf(s->avctx, "BITSTREAM: subframe length was %i\n",
            get_bits_count(&s->gb) - s->subframe_offset);

    if (transmit_coeffs) {
        /** reconstruct the per channel data */
        inverse_channel_transform(s);
        for (i = 0; i < s->channels_for_cur_subframe; i++) {
            int c = s->channel_indexes_for_cur_subframe[i];
            const int* sf = s->channel[c].scale_factors;
            int b;

            if (c == s->lfe_channel)
                memset(&s->tmp[cur_subwoofer_cutoff], 0, sizeof(*s->tmp) *
                       (subframe_len - cur_subwoofer_cutoff));

            /** inverse quantization and rescaling */
            for (b = 0; b < s->num_bands; b++) {
                const int end = FFMIN(s->cur_sfb_offsets[b+1], s->subframe_len);
                const int exp = s->channel[c].quant_step -
                            (s->channel[c].max_scale_factor - *sf++) *
                            s->channel[c].scale_factor_step;
                const float quant = pow(10.0, exp / 20.0);
                int start = s->cur_sfb_offsets[b];
                s->dsp.vector_fmul_scalar(s->tmp + start,
                                          s->channel[c].coeffs + start,
                                          quant, end - start);
            }

            /** apply imdct (ff_imdct_half == DCTIV with reverse) */
            ff_imdct_half(&s->mdct_ctx[av_log2(subframe_len) - BLOCK_MIN_BITS],
                          s->channel[c].coeffs, s->tmp);
        }
    }

    /** window and overlapp-add */
    wmapro_window(s);

    /** handled one subframe */
    for (i = 0; i < s->channels_for_cur_subframe; i++) {
        int c = s->channel_indexes_for_cur_subframe[i];
        if (s->channel[c].cur_subframe >= s->channel[c].num_subframes) {
            av_log(s->avctx, AV_LOG_ERROR, "broken subframe\n");
            return AVERROR_INVALIDDATA;
        }
        ++s->channel[c].cur_subframe;
    }

    return 0;
}

/**
 *@brief Decode one WMA frame.
 *@param s codec context
 *@return 0 if the trailer bit indicates that this is the last frame,
 *        1 if there are additional frames
 */
static int decode_frame(WMAProDecodeCtx *s)
{
    GetBitContext* gb = &s->gb;
    int more_frames = 0;
    int len = 0;
    int i;

    /** check for potential output buffer overflow */
    if (s->num_channels * s->samples_per_frame > s->samples_end - s->samples) {
        /** return an error if no frame could be decoded at all */
        av_log(s->avctx, AV_LOG_ERROR,
               "not enough space for the output samples\n");
        s->packet_loss = 1;
        return 0;
    }

    /** get frame length */
    if (s->len_prefix)
        len = get_bits(gb, s->log2_frame_size);

    dprintf(s->avctx, "decoding frame with length %x\n", len);

    /** decode tile information */
    if (decode_tilehdr(s)) {
        s->packet_loss = 1;
        return 0;
    }

    /** read postproc transform */
    if (s->num_channels > 1 && get_bits1(gb)) {
        av_log_ask_for_sample(s->avctx, "Unsupported postproc transform found\n");
        s->packet_loss = 1;
        return 0;
    }

    /** read drc info */
    if (s->dynamic_range_compression) {
        s->drc_gain = get_bits(gb, 8);
        dprintf(s->avctx, "drc_gain %i\n", s->drc_gain);
    }

    /** no idea what these are for, might be the number of samples
        that need to be skipped at the beginning or end of a stream */
    if (get_bits1(gb)) {
        int skip;

        /** usually true for the first frame */
        if (get_bits1(gb)) {
            skip = get_bits(gb, av_log2(s->samples_per_frame * 2));
            dprintf(s->avctx, "start skip: %i\n", skip);
        }

        /** sometimes true for the last frame */
        if (get_bits1(gb)) {
            skip = get_bits(gb, av_log2(s->samples_per_frame * 2));
            dprintf(s->avctx, "end skip: %i\n", skip);
        }

    }

    dprintf(s->avctx, "BITSTREAM: frame header length was %i\n",
            get_bits_count(gb) - s->frame_offset);

    /** reset subframe states */
    s->parsed_all_subframes = 0;
    for (i = 0; i < s->num_channels; i++) {
        s->channel[i].decoded_samples = 0;
        s->channel[i].cur_subframe    = 0;
        s->channel[i].reuse_sf        = 0;
    }

    /** decode all subframes */
    while (!s->parsed_all_subframes) {
        if (decode_subframe(s) < 0) {
            s->packet_loss = 1;
            return 0;
        }
    }

    /** interleave samples and write them to the output buffer */
    for (i = 0; i < s->num_channels; i++) {
        float* ptr  = s->samples + i;
        int incr = s->num_channels;
        float* iptr = s->channel[i].out;
        float* iend = iptr + s->samples_per_frame;

        // FIXME should create/use a DSP function here
        while (iptr < iend) {
            *ptr = *iptr++;
            ptr += incr;
        }

        /** reuse second half of the IMDCT output for the next frame */
        memcpy(&s->channel[i].out[0],
               &s->channel[i].out[s->samples_per_frame],
               s->samples_per_frame * sizeof(*s->channel[i].out) >> 1);
    }

    if (s->skip_frame) {
        s->skip_frame = 0;
    } else
        s->samples += s->num_channels * s->samples_per_frame;

    if (len != (get_bits_count(gb) - s->frame_offset) + 2) {
        /** FIXME: not sure if this is always an error */
        av_log(s->avctx, AV_LOG_ERROR, "frame[%i] would have to skip %i bits\n",
               s->frame_num, len - (get_bits_count(gb) - s->frame_offset) - 1);
        s->packet_loss = 1;
        return 0;
    }

    /** skip the rest of the frame data */
    skip_bits_long(gb, len - (get_bits_count(gb) - s->frame_offset) - 1);

    /** decode trailer bit */
    more_frames = get_bits1(gb);

    ++s->frame_num;
    return more_frames;
}

/**
 *@brief Calculate remaining input buffer length.
 *@param s codec context
 *@param gb bitstream reader context
 *@return remaining size in bits
 */
static int remaining_bits(WMAProDecodeCtx *s, GetBitContext *gb)
{
    return s->buf_bit_size - get_bits_count(gb);
}

/**
 *@brief Fill the bit reservoir with a (partial) frame.
 *@param s codec context
 *@param gb bitstream reader context
 *@param len length of the partial frame
 *@param append decides wether to reset the buffer or not
 */
static void save_bits(WMAProDecodeCtx *s, GetBitContext* gb, int len,
                      int append)
{
    int buflen;

    /** when the frame data does not need to be concatenated, the input buffer
        is resetted and additional bits from the previous frame are copyed
        and skipped later so that a fast byte copy is possible */

    if (!append) {
        s->frame_offset = get_bits_count(gb) & 7;
        s->num_saved_bits = s->frame_offset;
        init_put_bits(&s->pb, s->frame_data, MAX_FRAMESIZE);
    }

    buflen = (s->num_saved_bits + len + 8) >> 3;

    if (len <= 0 || buflen > MAX_FRAMESIZE) {
        av_log_ask_for_sample(s->avctx, "input buffer too small\n");
        s->packet_loss = 1;
        return;
    }

    s->num_saved_bits += len;
    if (!append) {
        ff_copy_bits(&s->pb, gb->buffer + (get_bits_count(gb) >> 3),
                     s->num_saved_bits);
    } else {
        int align = 8 - (get_bits_count(gb) & 7);
        align = FFMIN(align, len);
        put_bits(&s->pb, align, get_bits(gb, align));
        len -= align;
        ff_copy_bits(&s->pb, gb->buffer + (get_bits_count(gb) >> 3), len);
    }
    skip_bits_long(gb, len);

    {
        PutBitContext tmp = s->pb;
        flush_put_bits(&tmp);
    }

    init_get_bits(&s->gb, s->frame_data, s->num_saved_bits);
    skip_bits(&s->gb, s->frame_offset);
}

/**
 *@brief Decode a single WMA packet.
 *@param avctx codec context
 *@param data the output buffer
 *@param data_size number of bytes that were written to the output buffer
 *@param avpkt input packet
 *@return number of bytes that were read from the input buffer
 */
static int decode_packet(AVCodecContext *avctx,
                         void *data, int *data_size, AVPacket* avpkt)
{
    WMAProDecodeCtx *s = avctx->priv_data;
    GetBitContext* gb  = &s->pgb;
    const uint8_t* buf = avpkt->data;
    int buf_size       = avpkt->size;
    int num_bits_prev_frame;
    int packet_sequence_number;

    s->samples       = data;
    s->samples_end   = (float*)((int8_t*)data + *data_size);
    *data_size = 0;

    if (s->packet_done || s->packet_loss) {
        s->packet_done = 0;
        s->buf_bit_size = buf_size << 3;

        /** sanity check for the buffer length */
        if (buf_size < avctx->block_align)
            return 0;

        buf_size = avctx->block_align;

        /** parse packet header */
        init_get_bits(gb, buf, s->buf_bit_size);
        packet_sequence_number = get_bits(gb, 4);
        skip_bits(gb, 2);

        /** get number of bits that need to be added to the previous frame */
        num_bits_prev_frame = get_bits(gb, s->log2_frame_size);
        dprintf(avctx, "packet[%d]: nbpf %x\n", avctx->frame_number,
                num_bits_prev_frame);

        /** check for packet loss */
        if (!s->packet_loss &&
            ((s->packet_sequence_number + 1) & 0xF) != packet_sequence_number) {
            s->packet_loss = 1;
            av_log(avctx, AV_LOG_ERROR, "Packet loss detected! seq %x vs %x\n",
                   s->packet_sequence_number, packet_sequence_number);
        }
        s->packet_sequence_number = packet_sequence_number;

        if (num_bits_prev_frame > 0) {
            /** append the previous frame data to the remaining data from the
                previous packet to create a full frame */
            save_bits(s, gb, num_bits_prev_frame, 1);
            dprintf(avctx, "accumulated %x bits of frame data\n",
                    s->num_saved_bits - s->frame_offset);

            /** decode the cross packet frame if it is valid */
            if (!s->packet_loss)
                decode_frame(s);
        } else if (s->num_saved_bits - s->frame_offset) {
            dprintf(avctx, "ignoring %x previously saved bits\n",
                    s->num_saved_bits - s->frame_offset);
        }

        s->packet_loss = 0;

    } else {
        int frame_size;
        s->buf_bit_size = avpkt->size << 3;
        init_get_bits(gb, avpkt->data, s->buf_bit_size);
        skip_bits(gb, s->packet_offset);
        if (remaining_bits(s, gb) > s->log2_frame_size &&
            (frame_size = show_bits(gb, s->log2_frame_size)) &&
            frame_size <= remaining_bits(s, gb)) {
            save_bits(s, gb, frame_size, 0);
            s->packet_done = !decode_frame(s);
        } else
            s->packet_done = 1;
    }

    if (s->packet_done && !s->packet_loss &&
        remaining_bits(s, gb) > 0) {
        /** save the rest of the data so that it can be decoded
            with the next packet */
        save_bits(s, gb, remaining_bits(s, gb), 0);
    }

    *data_size = (int8_t *)s->samples - (int8_t *)data;
    s->packet_offset = get_bits_count(gb) & 7;

    return (s->packet_loss) ? AVERROR_INVALIDDATA : get_bits_count(gb) >> 3;
}

/**
 *@brief Clear decoder buffers (for seeking).
 *@param avctx codec context
 */
static void flush(AVCodecContext *avctx)
{
    WMAProDecodeCtx *s = avctx->priv_data;
    int i;
    /** reset output buffer as a part of it is used during the windowing of a
        new frame */
    for (i = 0; i < s->num_channels; i++)
        memset(s->channel[i].out, 0, s->samples_per_frame *
               sizeof(*s->channel[i].out));
    s->packet_loss = 1;
}


/**
 *@brief wmapro decoder
 */
AVCodec wmapro_decoder = {
    "wmapro",
    AVMEDIA_TYPE_AUDIO,
    CODEC_ID_WMAPRO,
    sizeof(WMAProDecodeCtx),
    decode_init,
    NULL,
    decode_end,
    decode_packet,
    .capabilities = CODEC_CAP_SUBFRAMES,
    .flush= flush,
    .long_name = NULL_IF_CONFIG_SMALL("Windows Media Audio 9 Professional"),
};