Add TXTH codec CP_YM [Marvel vs SF (SAT)]

doc/TXTH.md

@@ -157,6 +157,8 @@ as explained below, but often will use default values. Accepted codec strings:
 #   * For rare EA games [Harry Potter and the Chamber of Secrets (PC)]
 # - XA             CD-XA ADPCM (ISO 2048 mode1/data streams without subchannels)
 #   * For rare Saturn and PS2 games [Phantasy Star Collection (SAT), Fantavision (PS2), EA SAT videos]
+# - CP_YM          Capcom's Yamaha ADPCM
+#   * For rare Saturn games [Marvel Super Heroes vs Street Fighter (SAT)]
 codec = (codec string)
 ```
 
@@ -1052,6 +1054,15 @@ channels = 2
 num_samples = data_size
 ```
 
+#### Marvel Super Heroes vs Street Fighter (SAT) .ADP.txth
+```
+codec = CP_YM
+sample_rate = 24000
+channels = 2
+
+#loops are in MM.BIN, table at 0x80700 + id*4 - 0x06018B00
+```
+
 #### Sega Rally 3 (SAT) ALL_SOUND.txth
 ```
 codec             = PCM16LE

doc/USAGE.md

@@ -183,7 +183,7 @@ like foobar or Winamp don't react well to that, they may be renamed to these
 - `.aac` to `.laac` (tri-Ace games)
 - `.ac3` to `.lac3` (standard AC3)
 - `.aif` to `.laif` (standard Mac AIF, Asobo AIF, Ogg)
-- `.aiff/aifc` to `.laiffl/laifc` (standard Mac AIF)
+- `.aiff/aifc` to `.laiff/laifc` (standard Mac AIF)
 - `.asf` to `.lasf` (EA games, Argonaut ASF)
 - `.bin` to `.lbin` (various formats)
 - `.flac` to `.lflac` (standard FLAC)

src/coding/coding.h

@@ -172,6 +172,7 @@ int msadpcm_check_coefs(STREAMFILE* sf, off_t offset);
 
 /* yamaha_decoder */
 void decode_aica(VGMSTREAMCHANNEL* stream, sample_t* outbuf, int channelspacing, int32_t first_sample, int32_t samples_to_do, int channel, int is_stereo);
+void decode_cp_ym(VGMSTREAMCHANNEL* stream, sample_t* outbuf, int channelspacing, int32_t first_sample, int32_t samples_to_do, int channel, int is_stereo);
 void decode_aska(VGMSTREAMCHANNEL* stream, sample_t* outbuf, int channelspacing, int32_t first_sample, int32_t samples_to_do, int channel, size_t frame_size);
 void decode_nxap(VGMSTREAMCHANNEL* stream, sample_t* outbuf, int channelspacing, int32_t first_sample, int32_t samples_to_do);
 size_t yamaha_bytes_to_samples(size_t bytes, int channels);

src/coding/yamaha_decoder.c

@@ -12,6 +12,10 @@ static const int scale_step_adpcmb[16] = {
     57, 57, 57, 57, 77, 102, 128, 153,
 };
 
+static const int scale_step_capcom[8] = {
+    58982, 58982, 58982, 58982, 78643, 104858, 131072, 157286,
+};
+
 /* look-up for 'mul' IMA's sign*((code&7) * 2 + 1) for every code */
 static const int scale_delta[16] = {
       1,  3,  5,  7,  9, 11, 13, 15,
@@ -59,6 +63,31 @@ static void yamaha_aica_expand_nibble(VGMSTREAMCHANNEL * stream, off_t byte_offs
     *hist1 = sample;
 }
 
+/* Capcom's version of Yamaha expand */
+static void yamaha_capcom_expand_nibble(uint8_t byte, int shift, int32_t* hist1, int32_t* step_size, int16_t *out_sample) {
+    int code, ucode, delta, sample;
+    const int scale = 0x200; /* ADPCM state var, but seemingly fixed */
+
+    code =  (byte >> shift) & 0xf;
+    ucode = code & 0x7;
+    delta = (ucode * (*step_size)) >> 2; /* custom (SH2 CPU can't do odd shifts in one op, it seems) */
+    if (code & 8)
+        delta = -delta;
+    sample = *hist1 + delta;
+
+    sample = (short)sample; /* clamp not done, but output is always low-ish */
+
+    *step_size = ((*step_size) * scale_step_capcom[ucode]) >> 16;
+    if (*step_size < 0x80) *step_size = 0x80; /* unlike usual 0x7f */
+    else if (*step_size > 0x6000) *step_size = 0x6000;
+
+    *hist1 = sample;
+
+    /* OG code adds out sample, but seems to be always 0 (used for mono downmix?) */
+    sample = ((scale * sample) >> 8) /*+ *out_sample */;
+    *out_sample = sample;
+}
+
 
 /* info about Yamaha ADPCM as created by official yadpcm.acm (in 'Yamaha-ADPCM-ACM-Driver-100-j')
  * - possibly RIFF codec 0x20
@@ -107,6 +136,37 @@ void decode_aica(VGMSTREAMCHANNEL * stream, sample_t * outbuf, int channelspacin
     stream->adpcm_step_index = step_size;
 }
 
+/* Capcom/Saturn Yamaha ADPCM, reverse engineered from the exe (codec has no apparent name so CP_YM = Capcom Yamaha) */
+void decode_cp_ym(VGMSTREAMCHANNEL* stream, sample_t* outbuf, int channelspacing, int32_t first_sample, int32_t samples_to_do, int channel, int is_stereo) {
+    int i, sample_count = 0;
+    int16_t out_sample;
+    int32_t hist1 = stream->adpcm_history1_16;
+    int step_size = stream->adpcm_step_index;
+
+    /* no header (external setup), pre-clamp for wrong values */
+    if (step_size < 0x80) step_size = 0x80;
+    if (step_size > 0x6000) step_size = 0x6000;
+
+    for (i = first_sample; i < first_sample + samples_to_do; i++) {
+        uint8_t byte;
+        uint32_t offset = is_stereo ?
+                stream->offset + i :    /* stereo: one nibble per channel */
+                stream->offset + i/2;   /* mono: consecutive nibbles */
+        int shift = is_stereo ?
+                (!(channel&1) ? 0:4) :  /* even = low/L, odd = high/R */
+                (!(i&1) ? 0:4);         /* low nibble first */
+
+        byte = read_u8(offset, stream->streamfile);
+        yamaha_capcom_expand_nibble(byte, shift, &hist1, &step_size, &out_sample);
+        outbuf[sample_count] = out_sample;
+        sample_count += channelspacing;
+    }
+
+    stream->adpcm_history1_16 = hist1;
+    stream->adpcm_step_index = step_size;
+}
+
+
 /* tri-Ace Aska ADPCM, Yamaha ADPCM-B with headered frames (reversed from Android SO's .so)
  * implements table with if-else/switchs too but that's too goofy */
 void decode_aska(VGMSTREAMCHANNEL * stream, sample_t * outbuf, int channelspacing, int32_t first_sample, int32_t samples_to_do, int channel, size_t frame_size) {

src/decode.c

@@ -471,6 +471,7 @@ int get_vgmstream_samples_per_frame(VGMSTREAM* vgmstream) {
         case coding_WS: /* only works if output sample size is 8 bit, which always is for WS ADPCM */
             return vgmstream->ws_output_size;
         case coding_AICA:
+        case coding_CP_YM:
             return 1;
         case coding_AICA_int:
             return 2;
@@ -690,6 +691,7 @@ int get_vgmstream_frame_size(VGMSTREAM* vgmstream) {
             return vgmstream->current_block_size;
         case coding_AICA:
         case coding_AICA_int:
+        case coding_CP_YM:
             return 0x01;
         case coding_ASKA:
             return vgmstream->frame_size;
@@ -1331,6 +1333,15 @@ void decode_vgmstream(VGMSTREAM* vgmstream, int samples_written, int samples_to_
             }
             break;
         }
+        case coding_CP_YM: {
+            int is_stereo = (vgmstream->channels > 1);
+            for (ch = 0; ch < vgmstream->channels; ch++) {
+                decode_cp_ym(&vgmstream->ch[ch], buffer+ch,
+                        vgmstream->channels, vgmstream->samples_into_block, samples_to_do, ch,
+                        is_stereo);
+            }
+            break;
+        }
         case coding_ASKA:
             for (ch = 0; ch < vgmstream->channels; ch++) {
                 decode_aska(&vgmstream->ch[ch], buffer+ch,

src/formats.c

@@ -788,6 +788,7 @@ static const coding_info coding_info_list[] = {
         {coding_WS,                 "Westwood Studios VBR ADPCM"},
         {coding_AICA,               "Yamaha AICA 4-bit ADPCM"},
         {coding_AICA_int,           "Yamaha AICA 4-bit ADPCM (mono/interleave)"},
+        {coding_CP_YM,              "Capcom Yamaha 4-bit ADPCM"},
         {coding_ASKA,               "tri-Ace Aska 4-bit ADPCM"},
         {coding_NXAP,               "Nex NXAP 4-bit ADPCM"},
         {coding_TGC,                "Tiger Game.com 4-bit ADPCM"},

src/meta/txth.c

@@ -40,7 +40,8 @@ typedef enum {
     ASF = 30,           /* Argonaut ASF 4-bit ADPCM */
     EAXA = 31,          /* Electronic Arts EA-XA 4-bit ADPCM v1 */
     OKI4S = 32,         /* OKI ADPCM with 16-bit output (unlike OKI/VOX/Dialogic ADPCM's 12-bit) */
-    XA = 33,
+    XA,
+    CP_YM,
 
     UNKNOWN = 99,
 } txth_codec_t;
@@ -261,6 +262,7 @@ VGMSTREAM* init_vgmstream_txth(STREAMFILE* sf) {
         case ASF:        coding = coding_ASF; break;
         case EAXA:       coding = coding_EA_XA; break;
         case XA:         coding = coding_XA; break;
+        case CP_YM:      coding = coding_CP_YM; break;
         default:
             goto fail;
     }
@@ -373,6 +375,7 @@ VGMSTREAM* init_vgmstream_txth(STREAMFILE* sf) {
         case coding_OKI16:
         case coding_OKI4S:
         case coding_XA:
+        case coding_CP_YM:
             vgmstream->layout_type = layout_none;
             break;
 
@@ -945,6 +948,7 @@ static txth_codec_t parse_codec(txth_header* txth, const char* val) {
     else if (is_string(val,"ASF"))          return ASF;
     else if (is_string(val,"EAXA"))         return EAXA;
     else if (is_string(val,"XA"))           return XA;
+    else if (is_string(val,"CP_YM"))        return CP_YM;
     /* special handling */
     else if (is_string(val,"name_value"))   return txth->name_values[0];
     else if (is_string(val,"name_value1"))  return txth->name_values[0];
@@ -1582,6 +1586,16 @@ fail:
     return 0;
 }
 
+static void string_trim(char* str) {
+    int str_len = strlen(str);
+    int i;
+    for (i = str_len - 1; i >= 0; i--) {
+        if (str[i] != ' ')
+            break;
+        str[i] = '\0';
+    }
+}
+
 static int read_name_table_keyval(txth_header* txth, const char* line, char* key, char* val) {
     int ok;
     int subsong;
@@ -1594,8 +1608,10 @@ static int read_name_table_keyval(txth_header* txth, const char* line, char* key
         return 0;
 
     /* try "(name): (val))" */
+    
     ok = sscanf(line, " %[^\t#:] : %[^\t#\r\n] ", key, val);
     if (ok == 2) {
+        string_trim(key); /* otherwise includes end spaces before : */
         //;VGM_LOG("TXTH: name %s get\n", key);
         return 1;
     }
@@ -1658,15 +1674,7 @@ static int parse_name_table(txth_header* txth, char* name_list) {
         goto fail;
 
     /* trim name_list just in case */
-    {
-        int name_list_len = strlen(name_list);
-        int i;
-        for (i = name_list_len - 1; i >= 0; i--) {
-            if (name_list[i] != ' ')
-                break;
-            name_list[i] = '\0';
-        }
-    }
+    string_trim(name_list);
 
     //;VGM_LOG("TXTH: name_list='%s'\n", name_list);
 
@@ -2031,6 +2039,7 @@ static int get_bytes_to_samples(txth_header* txth, uint32_t bytes) {
         case DVI_IMA:
             return ima_bytes_to_samples(bytes, txth->channels);
         case AICA:
+        case CP_YM:
             return yamaha_bytes_to_samples(bytes, txth->channels);
         case PCFX:
         case OKI16:

src/vgmstream.h

@@ -138,6 +138,7 @@ typedef enum {
 
     coding_AICA,            /* Yamaha AICA ADPCM (stereo) */
     coding_AICA_int,        /* Yamaha AICA ADPCM (mono/interleave) */
+    coding_CP_YM,           /* Capcom's Yamaha ADPCM (stereo/mono) */
     coding_ASKA,            /* Aska ADPCM */
     coding_NXAP,            /* NXAP ADPCM */