Add some more wmapro decoder hunks

Originally committed as revision 19678 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavcodec/wmaprodec.c

+/*
+ * 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  libavcodec/wmaprodec.c
+ * @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.
+ */
+
 /**
  *@brief Uninitialize the decoder and free all resources.
  *@param avctx codec context
@@ -8,11 +96,6 @@ static av_cold int decode_end(AVCodecContext *avctx)
     WMA3DecodeContext *s = avctx->priv_data;
     int i;
 
-    av_freep(&s->num_sfb);
-    av_freep(&s->sfb_offsets);
-    av_freep(&s->subwoofer_cutoffs);
-    av_freep(&s->sf_offsets);
-
     for (i = 0 ; i < WMAPRO_BLOCK_SIZES ; i++)
         ff_mdct_end(&s->mdct_ctx[i]);