Import more ok'd parts of ALAC encoder from GSoC repo.

Originally committed as revision 14820 to svn://svn.ffmpeg.org/ffmpeg/trunk

1 files changed, 125 insertions, 15 deletions

diff --git a/libavcodec/alacenc.c b/libavcodec/alacenc.c
index 98a3451597..0a802f1525 100644
--- a/libavcodec/alacenc.c
+++ b/libavcodec/alacenc.c
@@ -33,15 +33,52 @@
 
 #define ALAC_ESCAPE_CODE          0x1FF
 #define ALAC_MAX_LPC_ORDER        30
-
+#define DEFAULT_MAX_PRED_ORDER    6
+#define DEFAULT_MIN_PRED_ORDER    4
+#define ALAC_MAX_LPC_PRECISION    9
+#define ALAC_MAX_LPC_SHIFT        9
+
+typedef struct RiceContext {
+    int history_mult;
+    int initial_history;
+    int k_modifier;
+    int rice_modifier;
+} RiceContext;
+
+typedef struct LPCContext {
+    int lpc_order;
+    int lpc_coeff[ALAC_MAX_LPC_ORDER+1];
+    int lpc_quant;
+} LPCContext;
+
+typedef struct AlacEncodeContext {
+    int compression_level;
+    int max_coded_frame_size;
+    int write_sample_size;
+    int32_t sample_buf[MAX_CHANNELS][DEFAULT_FRAME_SIZE];
     int interlacing_shift;
     int interlacing_leftweight;
     PutBitContext pbctx;
+    RiceContext rc;
+    LPCContext lpc[MAX_CHANNELS];
     DSPContext dspctx;
     AVCodecContext *avctx;
 } AlacEncodeContext;
 
 
+static void init_sample_buffers(AlacEncodeContext *s, int16_t *input_samples)
+{
+    int ch, i;
+
+    for(ch=0;ch<s->avctx->channels;ch++) {
+        int16_t *sptr = input_samples + ch;
+        for(i=0;i<s->avctx->frame_size;i++) {
+            s->sample_buf[ch][i] = *sptr;
+            sptr += s->avctx->channels;
+        }
+    }
+}
+
 static void encode_scalar(AlacEncodeContext *s, int x, int k, int write_sample_size)
 {
     int divisor, q, r;
@@ -71,7 +108,7 @@ static void encode_scalar(AlacEncodeContext *s, int x, int k, int write_sample_s
 
 static void write_frame_header(AlacEncodeContext *s, int is_verbatim)
 {
-    put_bits(&s->pbctx, 3,  s->channels-1);                 // No. of channels -1
+    put_bits(&s->pbctx, 3,  s->avctx->channels-1);          // No. of channels -1
     put_bits(&s->pbctx, 16, 0);                             // Seems to be zero
     put_bits(&s->pbctx, 1,  1);                             // Sample count is in the header
     put_bits(&s->pbctx, 2,  0);                             // FIXME: Wasted bytes field
@@ -79,6 +116,38 @@ static void write_frame_header(AlacEncodeContext *s, int is_verbatim)
     put_bits(&s->pbctx, 32, s->avctx->frame_size);          // No. of samples in the frame
 }
 
+static int estimate_stereo_mode(int32_t *left_ch, int32_t *right_ch, int n)
+{
+    int i, best;
+    int32_t lt, rt;
+    uint64_t sum[4];
+    uint64_t score[4];
+
+    /* calculate sum of 2nd order residual for each channel */
+    sum[0] = sum[1] = sum[2] = sum[3] = 0;
+    for(i=2; i<n; i++) {
+        lt = left_ch[i] - 2*left_ch[i-1] + left_ch[i-2];
+        rt = right_ch[i] - 2*right_ch[i-1] + right_ch[i-2];
+        sum[2] += FFABS((lt + rt) >> 1);
+        sum[3] += FFABS(lt - rt);
+        sum[0] += FFABS(lt);
+        sum[1] += FFABS(rt);
+    }
+
+    /* calculate score for each mode */
+    score[0] = sum[0] + sum[1];
+    score[1] = sum[0] + sum[3];
+    score[2] = sum[1] + sum[3];
+    score[3] = sum[2] + sum[3];
+
+    /* return mode with lowest score */
+    best = 0;
+    for(i=1; i<4; i++) {
+        if(score[i] < score[best]) {
+            best = i;
+        }
+    }
+
 static void write_compressed_frame(AlacEncodeContext *s)
 {
     int i, j;
@@ -88,7 +157,7 @@ static void write_compressed_frame(AlacEncodeContext *s)
     put_bits(&s->pbctx, 8, s->interlacing_shift);
     put_bits(&s->pbctx, 8, s->interlacing_leftweight);
 
-    for(i=0;i<s->channels;i++) {
+    for(i=0;i<s->avctx->channels;i++) {
 
         calc_predictor_params(s, i);
 
@@ -105,7 +174,7 @@ static void write_compressed_frame(AlacEncodeContext *s)
 
     // apply lpc and entropy coding to audio samples
 
-    for(i=0;i<s->channels;i++) {
+    for(i=0;i<s->avctx->channels;i++) {
         alac_linear_predictor(s, i);
         alac_entropy_coder(s);
     }
@@ -118,8 +187,6 @@ static av_cold int alac_encode_init(AVCodecContext *avctx)
 
     avctx->frame_size      = DEFAULT_FRAME_SIZE;
     avctx->bits_per_sample = DEFAULT_SAMPLE_SIZE;
-    s->channels            = avctx->channels;
-    s->samplerate          = avctx->sample_rate;
 
     if(avctx->sample_fmt != SAMPLE_FMT_S16) {
         av_log(avctx, AV_LOG_ERROR, "only pcm_s16 input samples are supported\n");
@@ -139,18 +206,18 @@ static av_cold int alac_encode_init(AVCodecContext *avctx)
     s->rc.rice_modifier   = 4;
 
     s->max_coded_frame_size = (ALAC_FRAME_HEADER_SIZE + ALAC_FRAME_FOOTER_SIZE +
-                               avctx->frame_size*s->channels*avctx->bits_per_sample)>>3;
+                               avctx->frame_size*avctx->channels*avctx->bits_per_sample)>>3;
 
-    s->write_sample_size  = avctx->bits_per_sample + s->channels - 1; // FIXME: consider wasted_bytes
+    s->write_sample_size  = avctx->bits_per_sample + avctx->channels - 1; // FIXME: consider wasted_bytes
 
     AV_WB32(alac_extradata,    ALAC_EXTRADATA_SIZE);
     AV_WB32(alac_extradata+4,  MKBETAG('a','l','a','c'));
     AV_WB32(alac_extradata+12, avctx->frame_size);
     AV_WB8 (alac_extradata+17, avctx->bits_per_sample);
-    AV_WB8 (alac_extradata+21, s->channels);
+    AV_WB8 (alac_extradata+21, avctx->channels);
     AV_WB32(alac_extradata+24, s->max_coded_frame_size);
-    AV_WB32(alac_extradata+28, s->samplerate*s->channels*avctx->bits_per_sample); // average bitrate
-    AV_WB32(alac_extradata+32, s->samplerate);
+    AV_WB32(alac_extradata+28, avctx->sample_rate*avctx->channels*avctx->bits_per_sample); // average bitrate
+    AV_WB32(alac_extradata+32, avctx->sample_rate);
 
     // Set relevant extradata fields
     if(s->compression_level > 0) {
@@ -168,19 +235,62 @@ static av_cold int alac_encode_init(AVCodecContext *avctx)
     s->avctx = avctx;
     dsputil_init(&s->dspctx, avctx);
 
-    allocate_sample_buffers(s);
-
     return 0;
 }
 
-static av_cold int alac_encode_close(AVCodecContext *avctx)
+static int alac_encode_frame(AVCodecContext *avctx, uint8_t *frame,
+                             int buf_size, void *data)
 {
     AlacEncodeContext *s = avctx->priv_data;
+    PutBitContext *pb = &s->pbctx;
+    int i, out_bytes, verbatim_flag = 0;
 
+    if(avctx->frame_size > DEFAULT_FRAME_SIZE) {
+        av_log(avctx, AV_LOG_ERROR, "input frame size exceeded\n");
+        return -1;
+    }
+
+    if(buf_size < 2*s->max_coded_frame_size) {
+        av_log(avctx, AV_LOG_ERROR, "buffer size is too small\n");
+        return -1;
+    }
+
+    if((s->compression_level == 0) || verbatim_flag) {
+        // Verbatim mode
+        int16_t *samples = data;
+        write_frame_header(s, 1);
+        for(i=0; i<avctx->frame_size*avctx->channels; i++) {
+            put_sbits(pb, 16, *samples++);
+        }
+    } else {
+        init_sample_buffers(s, data);
+        write_frame_header(s, 0);
+        write_compressed_frame(s);
+    }
+
+    put_bits(pb, 3, 7);
+    flush_put_bits(pb);
+    out_bytes = put_bits_count(pb) >> 3;
+
+    if(out_bytes > s->max_coded_frame_size) {
+        /* frame too large. use verbatim mode */
+        if(verbatim_flag || (s->compression_level == 0)) {
+            /* still too large. must be an error. */
+            av_log(avctx, AV_LOG_ERROR, "error encoding frame\n");
+            return -1;
+        }
+        verbatim_flag = 1;
+        goto verbatim;
+    }
+
+    return out_bytes;
+}
+
+static av_cold int alac_encode_close(AVCodecContext *avctx)
+{
     av_freep(&avctx->extradata);
     avctx->extradata_size = 0;
     av_freep(&avctx->coded_frame);
-    free_sample_buffers(s);
     return 0;
 }