bump version

Instead, we always use explicit/backward compatible SBR signaling by
ASC extension in case of m4a, which is not supported by FDK library
(so we do it on our side).
For LOAS, we use explicit hierarchical signaling.

ChangeLog

@@ -1,6 +1,26 @@
+2013-10-25  nu774 <honeycomb77@gmail.com>
+
+  * update ChangeLog [HEAD]
+
+  * bump version [v0.4.0]
+
+  * update README
+
+2013-10-24  nu774 <honeycomb77@gmail.com>
+
+  * caf input support
+
+  * refactor pcm io routines
+
+2013-10-23  nu774 <honeycomb77@gmail.com>
+
+  * cleanup metadata handling
+
+  * --tag-from-json: properly support number/total format in json track field
+
 2013-10-22  nu774 <honeycomb77@gmail.com>
 
-  * bump version [HEAD]
+  * bump version [v0.3.3]
 
   * fixed bogus sgpd written on --gapless-mode=1 and 2
 

src/aacenc.c

@@ -10,6 +10,7 @@
 #endif
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include "aacenc.h"
 
 int aacenc_is_sbr_active(const aacenc_param_t *params)
@@ -25,6 +26,68 @@ int aacenc_is_sbr_active(const aacenc_param_t *params)
     return 0;
 }
 
+static const unsigned aacenc_sampling_freq_tab[] = {
+    96000, 88200, 64000, 48000, 44100, 32000, 24000, 22050, 
+    16000, 12000, 11025, 8000, 7350, 0, 0, 0
+};
+
+static
+unsigned sampling_freq_index(unsigned rate)
+{
+    unsigned i;
+    for (i = 0; aacenc_sampling_freq_tab[i]; ++i)
+        if (aacenc_sampling_freq_tab[i] == rate)
+            return i;
+    return 0xf;
+}
+
+/*
+ * Append backward compatible SBR/PS signaling to implicit signaling ASC,
+ * if SBR/PS is present.
+ */
+int aacenc_mp4asc(const aacenc_param_t *params,
+                  const uint8_t *asc, uint32_t ascsize,
+                  uint8_t *outasc, uint32_t *outsize)
+{
+    unsigned asc_sfreq = aacenc_sampling_freq_tab[(asc[0]&0x7)<<1 |asc[1]>>7];
+
+    switch (params->profile) {
+    case AOT_SBR:
+    case AOT_PS:
+        if (*outsize < ascsize + 3)
+            return -1;
+        memcpy(outasc, asc, ascsize);
+        /* syncExtensionType:11 (value:0x2b7) */
+        outasc[ascsize+0] = 0x2b << 1;
+        outasc[ascsize+1] = 0x7 << 5;
+        /* extensionAudioObjectType:5 (value:5)*/
+        outasc[ascsize+1] |= 5;
+        /* sbrPresentFlag:1 (value:1) */
+        outasc[ascsize+2] = 0x80;
+        /* extensionSamplingFrequencyIndex:4 */
+        outasc[ascsize+2] |= sampling_freq_index(asc_sfreq << 1) << 3;
+        if (params->profile == AOT_SBR) {
+            *outsize = ascsize + 3;
+            break;
+        }
+        if (*outsize < ascsize + 5)
+            return -1;
+        /* syncExtensionType:11 (value:0x548) */
+        outasc[ascsize+2] |= 0x5;
+        outasc[ascsize+3] = 0x48;
+        /* psPresentFlag:1 (value:1) */
+        outasc[ascsize+4] = 0x80;
+        *outsize = ascsize + 5;
+        break;
+    default:
+        if (*outsize < ascsize)
+            return -1;
+        memcpy(outasc, asc, ascsize);
+        *outsize = ascsize;
+    }
+    return 0;
+}
+
 static
 int aacenc_channel_mode(const pcm_sample_description_t *format)
 {

src/aacenc.h

@@ -26,6 +26,10 @@ typedef struct aacenc_param_t {
 
 int aacenc_is_sbr_active(const aacenc_param_t *params);
 
+int aacenc_mp4asc(const aacenc_param_t *params,
+                  const uint8_t *asc, uint32_t ascsize,
+                  uint8_t *outasc, uint32_t *outsize);
+
 int aacenc_init(HANDLE_AACENCODER *encoder, const aacenc_param_t *params,
                 const pcm_sample_description_t *format,
                 AACENC_InfoStruct *info);

src/lpcm.c

@@ -13,8 +13,7 @@
 #include "lpcm.h"
 #include "m4af_endian.h"
 
-#ifdef _MSC_VER
+#if defined(_MSC_VER) && _MSC_VER < 1800
-#  define inline __inline
 #  ifdef _M_IX86
 inline int lrint(double x)
 {

src/main.c

@@ -133,10 +133,6 @@ PROGNAME " %s\n"
 "                                 0: Off\n"
 "                                 1: On(default)\n"
 " -L, --lowdelay-sbr            Enable ELD-SBR (AAC ELD only)\n"
-" -s, --sbr-signaling <n>       SBR signaling mode\n"
-"                                 0: Implicit, backward compatible(default)\n"
-"                                 1: Explicit SBR and implicit PS\n"
-"                                 2: Explicit hierarchical signaling\n"
 " -f, --transport-format <n>    Transport format\n"
 "                                 0: RAW (default, muxed into M4A)\n"
 "                                 1: ADIF\n"
@@ -153,7 +149,10 @@ PROGNAME " %s\n"
 "                                 0: iTunSMPB (default)\n"
 "                                 1: ISO standard (edts + sgpd)\n"
 "                                 2: Both\n"
-" --ignorelength                Ignore length of WAV header\n"
+" --include-sbr-delay           Count SBR decoder delay in encoder delay\n"
+"                               This is not iTunes compatible, but is default\n"
+"                               behavior of FDK library.\n"
+" -I, --ignorelength            Ignore length of WAV header\n"
 " -S, --silent                  Don't print progress messages\n"
 " --moov-before-mdat            Place moov box before mdat box on m4a output\n"
 "\n"
@@ -208,6 +207,7 @@ typedef struct aacenc_param_ex_t {
     char *output_filename;
     FILE *output_fp;
     unsigned gapless_mode;
+    unsigned include_sbr_delay;
     unsigned ignore_length;
     int silent;
     int moov_before_mdat;
@@ -230,6 +230,7 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
     int ch;
     unsigned n;
 
+#define OPT_INCLUDE_SBR_DELAY    M4AF_FOURCC('s','d','l','y')
 #define OPT_MOOV_BEFORE_MDAT     M4AF_FOURCC('m','o','o','v')
 #define OPT_RAW_CHANNELS         M4AF_FOURCC('r','c','h','n')
 #define OPT_RAW_RATE             M4AF_FOURCC('r','r','a','t')
@@ -247,12 +248,12 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
         { "bandwidth",        required_argument, 0, 'w' },
         { "afterburner",      required_argument, 0, 'a' },
         { "lowdelay-sbr",     no_argument,       0, 'L' },
-        { "sbr-signaling",    required_argument, 0, 's' },
         { "transport-format", required_argument, 0, 'f' },
         { "adts-crc-check",   no_argument,       0, 'C' },
         { "header-period",    required_argument, 0, 'P' },
 
         { "gapless-mode",     required_argument, 0, 'G' },
+        { "include-sbr-delay", no_argument,      0, OPT_INCLUDE_SBR_DELAY  },
         { "ignorelength",     no_argument,       0, 'I' },
         { "silent",           no_argument,       0, 'S' },
         { "moov-before-mdat", no_argument,       0, OPT_MOOV_BEFORE_MDAT   },
@@ -326,13 +327,6 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
         case 'L':
             params->lowdelay_sbr = 1;
             break;
-        case 's':
-            if (sscanf(optarg, "%u", &n) != 1 || n > 2) {
-                fprintf(stderr, "invalid arg for sbr-signaling\n");
-                return -1;
-            }
-            params->sbr_signaling = n;
-            break;
         case 'f':
             if (sscanf(optarg, "%u", &n) != 1) {
                 fprintf(stderr, "invalid arg for transport-format\n");
@@ -360,6 +354,9 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params)
             }
             params->gapless_mode = n;
             break;
+        case OPT_INCLUDE_SBR_DELAY:
+            params->include_sbr_delay = 1;
+            break;
         case 'I':
             params->ignore_length = 1;
             break;
@@ -669,7 +666,7 @@ pcm_reader_t *open_input(aacenc_param_ex_t *params)
         goto END;
     }
     io.cookie = params->input_fp;
-    if (fstat(fileno(io.cookie), &stb) == 0
+    if (fstat(fileno(params->input_fp), &stb) == 0
             && (stb.st_mode & S_IFMT) == S_IFREG)
         io.vtbl = &pcm_io_vtbl;
     else
@@ -736,6 +733,7 @@ int main(int argc, char **argv)
     const pcm_sample_description_t *sample_format;
     int downsampled_timescale = 0;
     int frame_count = 0;
+    int sbr_mode = 0;
 
     setlocale(LC_CTYPE, "");
     setbuf(stderr, 0);
@@ -748,6 +746,17 @@ int main(int argc, char **argv)
 
     sample_format = pcm_get_format(reader);
 
+    /*
+     * We use explicit/hierarchical signaling for LOAS.
+     * Other than that, we request implicit signaling to FDK library, then 
+     * append explicit/backward-compatible signaling to ASC in case of MP4FF.
+     *
+     * Explicit/backward-compatible signaling of SBR is the most recommended
+     * way in MPEG4 part3 spec, and seems the only way supported by iTunes.
+     * Since FDK library does not support it, we have to do it on our side.
+     */
+    params.sbr_signaling = (params.transport_format == TT_MP4_LOAS) ? 2 : 0;
+
     if (aacenc_init(&encoder, (aacenc_param_t*)&params, sample_format,
                     &aacinfo) < 0)
         goto END;
@@ -764,19 +773,21 @@ int main(int argc, char **argv)
         goto END;
     }
     handle_signals();
+    sbr_mode = aacenc_is_sbr_active((aacenc_param_t*)&params);
     if (!params.transport_format) {
         uint32_t scale;
+        uint8_t mp4asc[32];
+        uint32_t ascsize = sizeof(mp4asc);
         unsigned framelen = aacinfo.frameLength;
-        int sbr_mode = aacenc_is_sbr_active((aacenc_param_t*)&params);
+        if (sbr_mode)
-        int sig_mode = aacEncoder_GetParam(encoder, AACENC_SIGNALING_MODE);
-        if (sbr_mode && !sig_mode)
             downsampled_timescale = 1;
         scale = sample_format->sample_rate >> downsampled_timescale;
         if ((m4af = m4af_create(M4AF_CODEC_MP4A, scale, &m4af_io,
                                 params.output_fp)) < 0)
             goto END;
-        m4af_set_decoder_specific_info(m4af, 0, aacinfo.confBuf,
+        aacenc_mp4asc((aacenc_param_t*)&params, aacinfo.confBuf,
-                                       aacinfo.confSize);
+                      aacinfo.confSize, mp4asc, &ascsize);
+        m4af_set_decoder_specific_info(m4af, 0, mp4asc, ascsize);
         m4af_set_fixed_frame_duration(m4af, 0,
                                       framelen >> downsampled_timescale);
         m4af_set_vbr_mode(m4af, 0, params.bitrate_mode);
@@ -789,9 +800,15 @@ int main(int argc, char **argv)
         goto END;
     if (m4af) {
         uint32_t delay = aacinfo.encoderDelay;
+        uint32_t padding;
         int64_t frames_read = pcm_get_position(reader);
-        uint32_t padding = frame_count * aacinfo.frameLength
+
-                            - frames_read - aacinfo.encoderDelay;
+        if (sbr_mode && params.profile != AOT_ER_AAC_ELD &&
+            !params.include_sbr_delay)
+            delay -= 481 << 1;
+        if (sbr_mode && (delay & 1))
+            ++delay;
+        padding = frame_count * aacinfo.frameLength - frames_read - delay;
         m4af_set_priming(m4af, 0, delay >> downsampled_timescale,
                          padding >> downsampled_timescale);
         if (finalize_m4a(m4af, &params, encoder) < 0)

src/metadata.c

@@ -56,7 +56,8 @@ static tag_key_mapping_t tag_mapping_table[] = {
     { "grouping",                   M4AF_TAG_GROUPING            },
     { "itunescompilation",          M4AF_TAG_COMPILATION         },
     { "lyrics",                     M4AF_TAG_LYRICS              },
-    { "performer",                  M4AF_TAG_ARTIST              },
+    { "tempo",                      M4AF_TAG_TEMPO               },
+    { "recordeddate",               M4AF_TAG_DATE                },
     { "title",                      M4AF_TAG_TITLE               },
     { "titlesort",                  M4AF_FOURCC('s','o','n','m') },
     { "titlesortorder",             M4AF_FOURCC('s','o','n','m') },

version.h

@@ -1,4 +1,4 @@
 #ifndef VERSION_H
 #define VERSION_H
-const char *fdkaac_version = "0.4.0";
+const char *fdkaac_version = "0.4.1";
 #endif