diff --git a/CMakeLists.txt b/CMakeLists.txt
index 3fc9b57..f9c2fa0 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -54,6 +54,12 @@ else()
add_compile_definitions(POMME_NO_SOUND_FORMATS)
endif()
+if (NOT(POMME_NO_MP3))
+ list(APPEND POMME_SOURCES ${POMME_SRCDIR}/SoundFormats/mp3.cpp)
+else()
+ add_compile_definitions(POMME_NO_MP3)
+endif()
+
if (NOT(POMME_NO_SOUND_MIXER))
list(APPEND POMME_SOURCES
${POMME_SRCDIR}/SoundMixer/ChannelImpl.cpp
diff --git a/LICENSE.md b/LICENSE.md
index 182a1dc..a21b2ce 100644
--- a/LICENSE.md
+++ b/LICENSE.md
@@ -168,3 +168,15 @@ by Tristan Brindle (Boost license 1.0):
> FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
> ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
> DEALINGS IN THE SOFTWARE.
+
+--------
+
+Portions copied from [lieff/minimp3](https://github.com/lieff/minimp3)
+(Creative Commons Zero v1.0 Universal license):
+- minimp3.h
+
+> To the extent possible under law, the author(s) have dedicated all copyright
+> and related and neighboring rights to this software to the public domain
+> worldwide.
+> This software is distributed without any warranty.
+> See .
diff --git a/src/PommeSound.h b/src/PommeSound.h
index bd61a89..22d9b79 100644
--- a/src/PommeSound.h
+++ b/src/PommeSound.h
@@ -32,6 +32,8 @@ namespace Pomme::Sound
int8_t baseNote;
uint32_t loopStart;
uint32_t loopEnd;
+
+ SndListHandle MakeStandaloneResource(char** dataOffsetOut = nullptr) const;
};
class Codec
@@ -101,7 +103,8 @@ namespace Pomme::Sound
void GetSoundInfoFromSndResource(Handle sndHandle, SampledSoundInfo& info);
- std::streampos GetSoundInfoFromAIFF(std::istream& input, SampledSoundInfo& info);
+ SndListHandle LoadAIFFAsResource(std::istream& input);
+ SndListHandle LoadMP3AsResource(std::istream& input);
std::unique_ptr GetCodec(uint32_t fourCC);
}
diff --git a/src/SoundFormats/AIFF.cpp b/src/SoundFormats/AIFF.cpp
index 504baec..fcc6248 100644
--- a/src/SoundFormats/AIFF.cpp
+++ b/src/SoundFormats/AIFF.cpp
@@ -88,9 +88,9 @@ static void ParseINST(Pomme::BigEndianIStream& f, Pomme::Sound::SampledSoundInfo
}
}
-std::streampos Pomme::Sound::GetSoundInfoFromAIFF(std::istream& input, SampledSoundInfo& info)
+static std::streampos GetSoundInfoFromAIFF(std::istream& input, Pomme::Sound::SampledSoundInfo& info)
{
- BigEndianIStream f(input);
+ Pomme::BigEndianIStream f(input);
AIFFAssert('FORM' == f.Read(), "AIFF: invalid FORM");
auto formSize = f.Read();
@@ -181,3 +181,16 @@ std::streampos Pomme::Sound::GetSoundInfoFromAIFF(std::istream& input, SampledSo
return sampledSoundDataOffset;
}
+SndListHandle Pomme::Sound::LoadAIFFAsResource(std::istream& stream)
+{
+ Pomme::Sound::SampledSoundInfo info = {};
+ std::streampos ssndStart = GetSoundInfoFromAIFF(stream, info);
+
+ char* dataOffset = nullptr;
+ SndListHandle h = info.MakeStandaloneResource(&dataOffset);
+
+ stream.seekg(ssndStart, std::ios::beg);
+ stream.read(dataOffset, info.compressedLength);
+
+ return h;
+}
diff --git a/src/SoundFormats/SoundFormats.cpp b/src/SoundFormats/SoundFormats.cpp
index 2054aeb..1a87e03 100644
--- a/src/SoundFormats/SoundFormats.cpp
+++ b/src/SoundFormats/SoundFormats.cpp
@@ -230,24 +230,61 @@ void Pomme::Sound::GetSoundInfoFromSndResource(Handle sndHandle, SampledSoundInf
GetSoundInfo(sndhdr, info);
}
+SndListHandle Pomme::Sound::SampledSoundInfo::MakeStandaloneResource(char** dataOffsetOut) const
+{
+ const char* data = dataStart;
+
+ SampledSoundInfo info = *this;
+
+ Handle h = NewHandleClear(2 + 4 + sizeof(info) + compressedLength);
+ Ptr p = *h;
+
+ memcpy(p, "poPOMM", 6); // "po": see kSoundResourceType_Pomme; "POMM": see GetSoundInfo
+ p += 6;
+
+ info.dataStart = p + sizeof(info);
+ memcpy(p, &info, sizeof(info));
+ p += sizeof(info);
+
+ if (data != nullptr)
+ {
+ memcpy(p, data, info.compressedLength);
+ }
+
+ if (dataOffsetOut != nullptr)
+ {
+ *dataOffsetOut = p;
+ }
+
+ return (SndListHandle) h;
+}
+
//-----------------------------------------------------------------------------
// Extension: load AIFF file as resource
SndListHandle Pomme_SndLoadFileAsResource(short fRefNum)
{
+ auto& spec = Pomme::Files::GetSpec(fRefNum);
auto& stream = Pomme::Files::GetStream(fRefNum);
- Pomme::Sound::SampledSoundInfo info = {};
- std::streampos ssndStart = Pomme::Sound::GetSoundInfoFromAIFF(stream, info);
+ std::string fileName(spec.cName);
+ std::transform(fileName.begin(), fileName.end(), fileName.begin(), tolower);
- stream.seekg(ssndStart, std::ios::beg);
+ // Guess media container from extension
+ if (fileName.ends_with(".aiff")
+ || fileName.ends_with(".aifc")
+ || fileName.ends_with(".aif"))
+ {
+ return LoadAIFFAsResource(stream);
+ }
+ else if (fileName.ends_with(".mp3"))
+ {
+#ifndef POMME_NO_MP3
+ return LoadMP3AsResource(stream);
+#endif
+ }
- Handle h = NewHandleClear(2 + 4 + sizeof(SampledSoundInfo) + info.compressedLength);
- memcpy(*h, "poPOMM", 6);
- memcpy(*h+6, &info, sizeof(SampledSoundInfo));
- stream.read(*h+6+sizeof(SampledSoundInfo), info.compressedLength);
-
- return (SndListHandle) h;
+ return nullptr;
}
//-----------------------------------------------------------------------------
diff --git a/src/SoundFormats/minimp3.h b/src/SoundFormats/minimp3.h
new file mode 100644
index 0000000..3220ae1
--- /dev/null
+++ b/src/SoundFormats/minimp3.h
@@ -0,0 +1,1865 @@
+#ifndef MINIMP3_H
+#define MINIMP3_H
+/*
+ https://github.com/lieff/minimp3
+ To the extent possible under law, the author(s) have dedicated all copyright and related and neighboring rights to this software to the public domain worldwide.
+ This software is distributed without any warranty.
+ See .
+*/
+#include
+
+#define MINIMP3_MAX_SAMPLES_PER_FRAME (1152*2)
+
+typedef struct
+{
+ int frame_bytes, frame_offset, channels, hz, layer, bitrate_kbps;
+} mp3dec_frame_info_t;
+
+typedef struct
+{
+ float mdct_overlap[2][9*32], qmf_state[15*2*32];
+ int reserv, free_format_bytes;
+ unsigned char header[4], reserv_buf[511];
+} mp3dec_t;
+
+#ifdef __cplusplus
+extern "C" {
+#endif /* __cplusplus */
+
+void mp3dec_init(mp3dec_t *dec);
+#ifndef MINIMP3_FLOAT_OUTPUT
+typedef int16_t mp3d_sample_t;
+#else /* MINIMP3_FLOAT_OUTPUT */
+typedef float mp3d_sample_t;
+void mp3dec_f32_to_s16(const float *in, int16_t *out, int num_samples);
+#endif /* MINIMP3_FLOAT_OUTPUT */
+int mp3dec_decode_frame(mp3dec_t *dec, const uint8_t *mp3, int mp3_bytes, mp3d_sample_t *pcm, mp3dec_frame_info_t *info);
+
+#ifdef __cplusplus
+}
+#endif /* __cplusplus */
+
+#endif /* MINIMP3_H */
+#if defined(MINIMP3_IMPLEMENTATION) && !defined(_MINIMP3_IMPLEMENTATION_GUARD)
+#define _MINIMP3_IMPLEMENTATION_GUARD
+
+#include
+#include
+
+#define MAX_FREE_FORMAT_FRAME_SIZE 2304 /* more than ISO spec's */
+#ifndef MAX_FRAME_SYNC_MATCHES
+#define MAX_FRAME_SYNC_MATCHES 10
+#endif /* MAX_FRAME_SYNC_MATCHES */
+
+#define MAX_L3_FRAME_PAYLOAD_BYTES MAX_FREE_FORMAT_FRAME_SIZE /* MUST be >= 320000/8/32000*1152 = 1440 */
+
+#define MAX_BITRESERVOIR_BYTES 511
+#define SHORT_BLOCK_TYPE 2
+#define STOP_BLOCK_TYPE 3
+#define MODE_MONO 3
+#define MODE_JOINT_STEREO 1
+#define HDR_SIZE 4
+#define HDR_IS_MONO(h) (((h[3]) & 0xC0) == 0xC0)
+#define HDR_IS_MS_STEREO(h) (((h[3]) & 0xE0) == 0x60)
+#define HDR_IS_FREE_FORMAT(h) (((h[2]) & 0xF0) == 0)
+#define HDR_IS_CRC(h) (!((h[1]) & 1))
+#define HDR_TEST_PADDING(h) ((h[2]) & 0x2)
+#define HDR_TEST_MPEG1(h) ((h[1]) & 0x8)
+#define HDR_TEST_NOT_MPEG25(h) ((h[1]) & 0x10)
+#define HDR_TEST_I_STEREO(h) ((h[3]) & 0x10)
+#define HDR_TEST_MS_STEREO(h) ((h[3]) & 0x20)
+#define HDR_GET_STEREO_MODE(h) (((h[3]) >> 6) & 3)
+#define HDR_GET_STEREO_MODE_EXT(h) (((h[3]) >> 4) & 3)
+#define HDR_GET_LAYER(h) (((h[1]) >> 1) & 3)
+#define HDR_GET_BITRATE(h) ((h[2]) >> 4)
+#define HDR_GET_SAMPLE_RATE(h) (((h[2]) >> 2) & 3)
+#define HDR_GET_MY_SAMPLE_RATE(h) (HDR_GET_SAMPLE_RATE(h) + (((h[1] >> 3) & 1) + ((h[1] >> 4) & 1))*3)
+#define HDR_IS_FRAME_576(h) ((h[1] & 14) == 2)
+#define HDR_IS_LAYER_1(h) ((h[1] & 6) == 6)
+
+#define BITS_DEQUANTIZER_OUT -1
+#define MAX_SCF (255 + BITS_DEQUANTIZER_OUT*4 - 210)
+#define MAX_SCFI ((MAX_SCF + 3) & ~3)
+
+#define MINIMP3_MIN(a, b) ((a) > (b) ? (b) : (a))
+#define MINIMP3_MAX(a, b) ((a) < (b) ? (b) : (a))
+
+#if !defined(MINIMP3_NO_SIMD)
+
+#if !defined(MINIMP3_ONLY_SIMD) && (defined(_M_X64) || defined(__x86_64__) || defined(__aarch64__) || defined(_M_ARM64))
+/* x64 always have SSE2, arm64 always have neon, no need for generic code */
+#define MINIMP3_ONLY_SIMD
+#endif /* SIMD checks... */
+
+#if (defined(_MSC_VER) && (defined(_M_IX86) || defined(_M_X64))) || ((defined(__i386__) || defined(__x86_64__)) && defined(__SSE2__))
+#if defined(_MSC_VER)
+#include
+#endif /* defined(_MSC_VER) */
+#include
+#define HAVE_SSE 1
+#define HAVE_SIMD 1
+#define VSTORE _mm_storeu_ps
+#define VLD _mm_loadu_ps
+#define VSET _mm_set1_ps
+#define VADD _mm_add_ps
+#define VSUB _mm_sub_ps
+#define VMUL _mm_mul_ps
+#define VMAC(a, x, y) _mm_add_ps(a, _mm_mul_ps(x, y))
+#define VMSB(a, x, y) _mm_sub_ps(a, _mm_mul_ps(x, y))
+#define VMUL_S(x, s) _mm_mul_ps(x, _mm_set1_ps(s))
+#define VREV(x) _mm_shuffle_ps(x, x, _MM_SHUFFLE(0, 1, 2, 3))
+typedef __m128 f4;
+#if defined(_MSC_VER) || defined(MINIMP3_ONLY_SIMD)
+#define minimp3_cpuid __cpuid
+#else /* defined(_MSC_VER) || defined(MINIMP3_ONLY_SIMD) */
+static __inline__ __attribute__((always_inline)) void minimp3_cpuid(int CPUInfo[], const int InfoType)
+{
+#if defined(__PIC__)
+ __asm__ __volatile__(
+#if defined(__x86_64__)
+ "push %%rbx\n"
+ "cpuid\n"
+ "xchgl %%ebx, %1\n"
+ "pop %%rbx\n"
+#else /* defined(__x86_64__) */
+ "xchgl %%ebx, %1\n"
+ "cpuid\n"
+ "xchgl %%ebx, %1\n"
+#endif /* defined(__x86_64__) */
+ : "=a" (CPUInfo[0]), "=r" (CPUInfo[1]), "=c" (CPUInfo[2]), "=d" (CPUInfo[3])
+ : "a" (InfoType));
+#else /* defined(__PIC__) */
+ __asm__ __volatile__(
+ "cpuid"
+ : "=a" (CPUInfo[0]), "=b" (CPUInfo[1]), "=c" (CPUInfo[2]), "=d" (CPUInfo[3])
+ : "a" (InfoType));
+#endif /* defined(__PIC__)*/
+}
+#endif /* defined(_MSC_VER) || defined(MINIMP3_ONLY_SIMD) */
+static int have_simd(void)
+{
+#ifdef MINIMP3_ONLY_SIMD
+ return 1;
+#else /* MINIMP3_ONLY_SIMD */
+ static int g_have_simd;
+ int CPUInfo[4];
+#ifdef MINIMP3_TEST
+ static int g_counter;
+ if (g_counter++ > 100)
+ return 0;
+#endif /* MINIMP3_TEST */
+ if (g_have_simd)
+ goto end;
+ minimp3_cpuid(CPUInfo, 0);
+ g_have_simd = 1;
+ if (CPUInfo[0] > 0)
+ {
+ minimp3_cpuid(CPUInfo, 1);
+ g_have_simd = (CPUInfo[3] & (1 << 26)) + 1; /* SSE2 */
+ }
+end:
+ return g_have_simd - 1;
+#endif /* MINIMP3_ONLY_SIMD */
+}
+#elif defined(__ARM_NEON) || defined(__aarch64__) || defined(_M_ARM64)
+#include
+#define HAVE_SSE 0
+#define HAVE_SIMD 1
+#define VSTORE vst1q_f32
+#define VLD vld1q_f32
+#define VSET vmovq_n_f32
+#define VADD vaddq_f32
+#define VSUB vsubq_f32
+#define VMUL vmulq_f32
+#define VMAC(a, x, y) vmlaq_f32(a, x, y)
+#define VMSB(a, x, y) vmlsq_f32(a, x, y)
+#define VMUL_S(x, s) vmulq_f32(x, vmovq_n_f32(s))
+#define VREV(x) vcombine_f32(vget_high_f32(vrev64q_f32(x)), vget_low_f32(vrev64q_f32(x)))
+typedef float32x4_t f4;
+static int have_simd()
+{ /* TODO: detect neon for !MINIMP3_ONLY_SIMD */
+ return 1;
+}
+#else /* SIMD checks... */
+#define HAVE_SSE 0
+#define HAVE_SIMD 0
+#ifdef MINIMP3_ONLY_SIMD
+#error MINIMP3_ONLY_SIMD used, but SSE/NEON not enabled
+#endif /* MINIMP3_ONLY_SIMD */
+#endif /* SIMD checks... */
+#else /* !defined(MINIMP3_NO_SIMD) */
+#define HAVE_SIMD 0
+#endif /* !defined(MINIMP3_NO_SIMD) */
+
+#if defined(__ARM_ARCH) && (__ARM_ARCH >= 6) && !defined(__aarch64__) && !defined(_M_ARM64)
+#define HAVE_ARMV6 1
+static __inline__ __attribute__((always_inline)) int32_t minimp3_clip_int16_arm(int32_t a)
+{
+ int32_t x = 0;
+ __asm__ ("ssat %0, #16, %1" : "=r"(x) : "r"(a));
+ return x;
+}
+#else
+#define HAVE_ARMV6 0
+#endif
+
+typedef struct
+{
+ const uint8_t *buf;
+ int pos, limit;
+} bs_t;
+
+typedef struct
+{
+ float scf[3*64];
+ uint8_t total_bands, stereo_bands, bitalloc[64], scfcod[64];
+} L12_scale_info;
+
+typedef struct
+{
+ uint8_t tab_offset, code_tab_width, band_count;
+} L12_subband_alloc_t;
+
+typedef struct
+{
+ const uint8_t *sfbtab;
+ uint16_t part_23_length, big_values, scalefac_compress;
+ uint8_t global_gain, block_type, mixed_block_flag, n_long_sfb, n_short_sfb;
+ uint8_t table_select[3], region_count[3], subblock_gain[3];
+ uint8_t preflag, scalefac_scale, count1_table, scfsi;
+} L3_gr_info_t;
+
+typedef struct
+{
+ bs_t bs;
+ uint8_t maindata[MAX_BITRESERVOIR_BYTES + MAX_L3_FRAME_PAYLOAD_BYTES];
+ L3_gr_info_t gr_info[4];
+ float grbuf[2][576], scf[40], syn[18 + 15][2*32];
+ uint8_t ist_pos[2][39];
+} mp3dec_scratch_t;
+
+static void bs_init(bs_t *bs, const uint8_t *data, int bytes)
+{
+ bs->buf = data;
+ bs->pos = 0;
+ bs->limit = bytes*8;
+}
+
+static uint32_t get_bits(bs_t *bs, int n)
+{
+ uint32_t next, cache = 0, s = bs->pos & 7;
+ int shl = n + s;
+ const uint8_t *p = bs->buf + (bs->pos >> 3);
+ if ((bs->pos += n) > bs->limit)
+ return 0;
+ next = *p++ & (255 >> s);
+ while ((shl -= 8) > 0)
+ {
+ cache |= next << shl;
+ next = *p++;
+ }
+ return cache | (next >> -shl);
+}
+
+static int hdr_valid(const uint8_t *h)
+{
+ return h[0] == 0xff &&
+ ((h[1] & 0xF0) == 0xf0 || (h[1] & 0xFE) == 0xe2) &&
+ (HDR_GET_LAYER(h) != 0) &&
+ (HDR_GET_BITRATE(h) != 15) &&
+ (HDR_GET_SAMPLE_RATE(h) != 3);
+}
+
+static int hdr_compare(const uint8_t *h1, const uint8_t *h2)
+{
+ return hdr_valid(h2) &&
+ ((h1[1] ^ h2[1]) & 0xFE) == 0 &&
+ ((h1[2] ^ h2[2]) & 0x0C) == 0 &&
+ !(HDR_IS_FREE_FORMAT(h1) ^ HDR_IS_FREE_FORMAT(h2));
+}
+
+static unsigned hdr_bitrate_kbps(const uint8_t *h)
+{
+ static const uint8_t halfrate[2][3][15] = {
+ { { 0,4,8,12,16,20,24,28,32,40,48,56,64,72,80 }, { 0,4,8,12,16,20,24,28,32,40,48,56,64,72,80 }, { 0,16,24,28,32,40,48,56,64,72,80,88,96,112,128 } },
+ { { 0,16,20,24,28,32,40,48,56,64,80,96,112,128,160 }, { 0,16,24,28,32,40,48,56,64,80,96,112,128,160,192 }, { 0,16,32,48,64,80,96,112,128,144,160,176,192,208,224 } },
+ };
+ return 2*halfrate[!!HDR_TEST_MPEG1(h)][HDR_GET_LAYER(h) - 1][HDR_GET_BITRATE(h)];
+}
+
+static unsigned hdr_sample_rate_hz(const uint8_t *h)
+{
+ static const unsigned g_hz[3] = { 44100, 48000, 32000 };
+ return g_hz[HDR_GET_SAMPLE_RATE(h)] >> (int)!HDR_TEST_MPEG1(h) >> (int)!HDR_TEST_NOT_MPEG25(h);
+}
+
+static unsigned hdr_frame_samples(const uint8_t *h)
+{
+ return HDR_IS_LAYER_1(h) ? 384 : (1152 >> (int)HDR_IS_FRAME_576(h));
+}
+
+static int hdr_frame_bytes(const uint8_t *h, int free_format_size)
+{
+ int frame_bytes = hdr_frame_samples(h)*hdr_bitrate_kbps(h)*125/hdr_sample_rate_hz(h);
+ if (HDR_IS_LAYER_1(h))
+ {
+ frame_bytes &= ~3; /* slot align */
+ }
+ return frame_bytes ? frame_bytes : free_format_size;
+}
+
+static int hdr_padding(const uint8_t *h)
+{
+ return HDR_TEST_PADDING(h) ? (HDR_IS_LAYER_1(h) ? 4 : 1) : 0;
+}
+
+#ifndef MINIMP3_ONLY_MP3
+static const L12_subband_alloc_t *L12_subband_alloc_table(const uint8_t *hdr, L12_scale_info *sci)
+{
+ const L12_subband_alloc_t *alloc;
+ int mode = HDR_GET_STEREO_MODE(hdr);
+ int nbands, stereo_bands = (mode == MODE_MONO) ? 0 : (mode == MODE_JOINT_STEREO) ? (HDR_GET_STEREO_MODE_EXT(hdr) << 2) + 4 : 32;
+
+ if (HDR_IS_LAYER_1(hdr))
+ {
+ static const L12_subband_alloc_t g_alloc_L1[] = { { 76, 4, 32 } };
+ alloc = g_alloc_L1;
+ nbands = 32;
+ } else if (!HDR_TEST_MPEG1(hdr))
+ {
+ static const L12_subband_alloc_t g_alloc_L2M2[] = { { 60, 4, 4 }, { 44, 3, 7 }, { 44, 2, 19 } };
+ alloc = g_alloc_L2M2;
+ nbands = 30;
+ } else
+ {
+ static const L12_subband_alloc_t g_alloc_L2M1[] = { { 0, 4, 3 }, { 16, 4, 8 }, { 32, 3, 12 }, { 40, 2, 7 } };
+ int sample_rate_idx = HDR_GET_SAMPLE_RATE(hdr);
+ unsigned kbps = hdr_bitrate_kbps(hdr) >> (int)(mode != MODE_MONO);
+ if (!kbps) /* free-format */
+ {
+ kbps = 192;
+ }
+
+ alloc = g_alloc_L2M1;
+ nbands = 27;
+ if (kbps < 56)
+ {
+ static const L12_subband_alloc_t g_alloc_L2M1_lowrate[] = { { 44, 4, 2 }, { 44, 3, 10 } };
+ alloc = g_alloc_L2M1_lowrate;
+ nbands = sample_rate_idx == 2 ? 12 : 8;
+ } else if (kbps >= 96 && sample_rate_idx != 1)
+ {
+ nbands = 30;
+ }
+ }
+
+ sci->total_bands = (uint8_t)nbands;
+ sci->stereo_bands = (uint8_t)MINIMP3_MIN(stereo_bands, nbands);
+
+ return alloc;
+}
+
+static void L12_read_scalefactors(bs_t *bs, uint8_t *pba, uint8_t *scfcod, int bands, float *scf)
+{
+ static const float g_deq_L12[18*3] = {
+#define DQ(x) 9.53674316e-07f/x, 7.56931807e-07f/x, 6.00777173e-07f/x
+ DQ(3),DQ(7),DQ(15),DQ(31),DQ(63),DQ(127),DQ(255),DQ(511),DQ(1023),DQ(2047),DQ(4095),DQ(8191),DQ(16383),DQ(32767),DQ(65535),DQ(3),DQ(5),DQ(9)
+ };
+ int i, m;
+ for (i = 0; i < bands; i++)
+ {
+ float s = 0;
+ int ba = *pba++;
+ int mask = ba ? 4 + ((19 >> scfcod[i]) & 3) : 0;
+ for (m = 4; m; m >>= 1)
+ {
+ if (mask & m)
+ {
+ int b = get_bits(bs, 6);
+ s = g_deq_L12[ba*3 - 6 + b % 3]*(1 << 21 >> b/3);
+ }
+ *scf++ = s;
+ }
+ }
+}
+
+static void L12_read_scale_info(const uint8_t *hdr, bs_t *bs, L12_scale_info *sci)
+{
+ static const uint8_t g_bitalloc_code_tab[] = {
+ 0,17, 3, 4, 5,6,7, 8,9,10,11,12,13,14,15,16,
+ 0,17,18, 3,19,4,5, 6,7, 8, 9,10,11,12,13,16,
+ 0,17,18, 3,19,4,5,16,
+ 0,17,18,16,
+ 0,17,18,19, 4,5,6, 7,8, 9,10,11,12,13,14,15,
+ 0,17,18, 3,19,4,5, 6,7, 8, 9,10,11,12,13,14,
+ 0, 2, 3, 4, 5,6,7, 8,9,10,11,12,13,14,15,16
+ };
+ const L12_subband_alloc_t *subband_alloc = L12_subband_alloc_table(hdr, sci);
+
+ int i, k = 0, ba_bits = 0;
+ const uint8_t *ba_code_tab = g_bitalloc_code_tab;
+
+ for (i = 0; i < sci->total_bands; i++)
+ {
+ uint8_t ba;
+ if (i == k)
+ {
+ k += subband_alloc->band_count;
+ ba_bits = subband_alloc->code_tab_width;
+ ba_code_tab = g_bitalloc_code_tab + subband_alloc->tab_offset;
+ subband_alloc++;
+ }
+ ba = ba_code_tab[get_bits(bs, ba_bits)];
+ sci->bitalloc[2*i] = ba;
+ if (i < sci->stereo_bands)
+ {
+ ba = ba_code_tab[get_bits(bs, ba_bits)];
+ }
+ sci->bitalloc[2*i + 1] = sci->stereo_bands ? ba : 0;
+ }
+
+ for (i = 0; i < 2*sci->total_bands; i++)
+ {
+ sci->scfcod[i] = sci->bitalloc[i] ? HDR_IS_LAYER_1(hdr) ? 2 : get_bits(bs, 2) : 6;
+ }
+
+ L12_read_scalefactors(bs, sci->bitalloc, sci->scfcod, sci->total_bands*2, sci->scf);
+
+ for (i = sci->stereo_bands; i < sci->total_bands; i++)
+ {
+ sci->bitalloc[2*i + 1] = 0;
+ }
+}
+
+static int L12_dequantize_granule(float *grbuf, bs_t *bs, L12_scale_info *sci, int group_size)
+{
+ int i, j, k, choff = 576;
+ for (j = 0; j < 4; j++)
+ {
+ float *dst = grbuf + group_size*j;
+ for (i = 0; i < 2*sci->total_bands; i++)
+ {
+ int ba = sci->bitalloc[i];
+ if (ba != 0)
+ {
+ if (ba < 17)
+ {
+ int half = (1 << (ba - 1)) - 1;
+ for (k = 0; k < group_size; k++)
+ {
+ dst[k] = (float)((int)get_bits(bs, ba) - half);
+ }
+ } else
+ {
+ unsigned mod = (2 << (ba - 17)) + 1; /* 3, 5, 9 */
+ unsigned code = get_bits(bs, mod + 2 - (mod >> 3)); /* 5, 7, 10 */
+ for (k = 0; k < group_size; k++, code /= mod)
+ {
+ dst[k] = (float)((int)(code % mod - mod/2));
+ }
+ }
+ }
+ dst += choff;
+ choff = 18 - choff;
+ }
+ }
+ return group_size*4;
+}
+
+static void L12_apply_scf_384(L12_scale_info *sci, const float *scf, float *dst)
+{
+ int i, k;
+ memcpy(dst + 576 + sci->stereo_bands*18, dst + sci->stereo_bands*18, (sci->total_bands - sci->stereo_bands)*18*sizeof(float));
+ for (i = 0; i < sci->total_bands; i++, dst += 18, scf += 6)
+ {
+ for (k = 0; k < 12; k++)
+ {
+ dst[k + 0] *= scf[0];
+ dst[k + 576] *= scf[3];
+ }
+ }
+}
+#endif /* MINIMP3_ONLY_MP3 */
+
+static int L3_read_side_info(bs_t *bs, L3_gr_info_t *gr, const uint8_t *hdr)
+{
+ static const uint8_t g_scf_long[8][23] = {
+ { 6,6,6,6,6,6,8,10,12,14,16,20,24,28,32,38,46,52,60,68,58,54,0 },
+ { 12,12,12,12,12,12,16,20,24,28,32,40,48,56,64,76,90,2,2,2,2,2,0 },
+ { 6,6,6,6,6,6,8,10,12,14,16,20,24,28,32,38,46,52,60,68,58,54,0 },
+ { 6,6,6,6,6,6,8,10,12,14,16,18,22,26,32,38,46,54,62,70,76,36,0 },
+ { 6,6,6,6,6,6,8,10,12,14,16,20,24,28,32,38,46,52,60,68,58,54,0 },
+ { 4,4,4,4,4,4,6,6,8,8,10,12,16,20,24,28,34,42,50,54,76,158,0 },
+ { 4,4,4,4,4,4,6,6,6,8,10,12,16,18,22,28,34,40,46,54,54,192,0 },
+ { 4,4,4,4,4,4,6,6,8,10,12,16,20,24,30,38,46,56,68,84,102,26,0 }
+ };
+ static const uint8_t g_scf_short[8][40] = {
+ { 4,4,4,4,4,4,4,4,4,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,24,24,24,30,30,30,40,40,40,18,18,18,0 },
+ { 8,8,8,8,8,8,8,8,8,12,12,12,16,16,16,20,20,20,24,24,24,28,28,28,36,36,36,2,2,2,2,2,2,2,2,2,26,26,26,0 },
+ { 4,4,4,4,4,4,4,4,4,6,6,6,6,6,6,8,8,8,10,10,10,14,14,14,18,18,18,26,26,26,32,32,32,42,42,42,18,18,18,0 },
+ { 4,4,4,4,4,4,4,4,4,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,24,24,24,32,32,32,44,44,44,12,12,12,0 },
+ { 4,4,4,4,4,4,4,4,4,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,24,24,24,30,30,30,40,40,40,18,18,18,0 },
+ { 4,4,4,4,4,4,4,4,4,4,4,4,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,22,22,22,30,30,30,56,56,56,0 },
+ { 4,4,4,4,4,4,4,4,4,4,4,4,6,6,6,6,6,6,10,10,10,12,12,12,14,14,14,16,16,16,20,20,20,26,26,26,66,66,66,0 },
+ { 4,4,4,4,4,4,4,4,4,4,4,4,6,6,6,8,8,8,12,12,12,16,16,16,20,20,20,26,26,26,34,34,34,42,42,42,12,12,12,0 }
+ };
+ static const uint8_t g_scf_mixed[8][40] = {
+ { 6,6,6,6,6,6,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,24,24,24,30,30,30,40,40,40,18,18,18,0 },
+ { 12,12,12,4,4,4,8,8,8,12,12,12,16,16,16,20,20,20,24,24,24,28,28,28,36,36,36,2,2,2,2,2,2,2,2,2,26,26,26,0 },
+ { 6,6,6,6,6,6,6,6,6,6,6,6,8,8,8,10,10,10,14,14,14,18,18,18,26,26,26,32,32,32,42,42,42,18,18,18,0 },
+ { 6,6,6,6,6,6,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,24,24,24,32,32,32,44,44,44,12,12,12,0 },
+ { 6,6,6,6,6,6,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,24,24,24,30,30,30,40,40,40,18,18,18,0 },
+ { 4,4,4,4,4,4,6,6,4,4,4,6,6,6,8,8,8,10,10,10,12,12,12,14,14,14,18,18,18,22,22,22,30,30,30,56,56,56,0 },
+ { 4,4,4,4,4,4,6,6,4,4,4,6,6,6,6,6,6,10,10,10,12,12,12,14,14,14,16,16,16,20,20,20,26,26,26,66,66,66,0 },
+ { 4,4,4,4,4,4,6,6,4,4,4,6,6,6,8,8,8,12,12,12,16,16,16,20,20,20,26,26,26,34,34,34,42,42,42,12,12,12,0 }
+ };
+
+ unsigned tables, scfsi = 0;
+ int main_data_begin, part_23_sum = 0;
+ int sr_idx = HDR_GET_MY_SAMPLE_RATE(hdr); sr_idx -= (sr_idx != 0);
+ int gr_count = HDR_IS_MONO(hdr) ? 1 : 2;
+
+ if (HDR_TEST_MPEG1(hdr))
+ {
+ gr_count *= 2;
+ main_data_begin = get_bits(bs, 9);
+ scfsi = get_bits(bs, 7 + gr_count);
+ } else
+ {
+ main_data_begin = get_bits(bs, 8 + gr_count) >> gr_count;
+ }
+
+ do
+ {
+ if (HDR_IS_MONO(hdr))
+ {
+ scfsi <<= 4;
+ }
+ gr->part_23_length = (uint16_t)get_bits(bs, 12);
+ part_23_sum += gr->part_23_length;
+ gr->big_values = (uint16_t)get_bits(bs, 9);
+ if (gr->big_values > 288)
+ {
+ return -1;
+ }
+ gr->global_gain = (uint8_t)get_bits(bs, 8);
+ gr->scalefac_compress = (uint16_t)get_bits(bs, HDR_TEST_MPEG1(hdr) ? 4 : 9);
+ gr->sfbtab = g_scf_long[sr_idx];
+ gr->n_long_sfb = 22;
+ gr->n_short_sfb = 0;
+ if (get_bits(bs, 1))
+ {
+ gr->block_type = (uint8_t)get_bits(bs, 2);
+ if (!gr->block_type)
+ {
+ return -1;
+ }
+ gr->mixed_block_flag = (uint8_t)get_bits(bs, 1);
+ gr->region_count[0] = 7;
+ gr->region_count[1] = 255;
+ if (gr->block_type == SHORT_BLOCK_TYPE)
+ {
+ scfsi &= 0x0F0F;
+ if (!gr->mixed_block_flag)
+ {
+ gr->region_count[0] = 8;
+ gr->sfbtab = g_scf_short[sr_idx];
+ gr->n_long_sfb = 0;
+ gr->n_short_sfb = 39;
+ } else
+ {
+ gr->sfbtab = g_scf_mixed[sr_idx];
+ gr->n_long_sfb = HDR_TEST_MPEG1(hdr) ? 8 : 6;
+ gr->n_short_sfb = 30;
+ }
+ }
+ tables = get_bits(bs, 10);
+ tables <<= 5;
+ gr->subblock_gain[0] = (uint8_t)get_bits(bs, 3);
+ gr->subblock_gain[1] = (uint8_t)get_bits(bs, 3);
+ gr->subblock_gain[2] = (uint8_t)get_bits(bs, 3);
+ } else
+ {
+ gr->block_type = 0;
+ gr->mixed_block_flag = 0;
+ tables = get_bits(bs, 15);
+ gr->region_count[0] = (uint8_t)get_bits(bs, 4);
+ gr->region_count[1] = (uint8_t)get_bits(bs, 3);
+ gr->region_count[2] = 255;
+ }
+ gr->table_select[0] = (uint8_t)(tables >> 10);
+ gr->table_select[1] = (uint8_t)((tables >> 5) & 31);
+ gr->table_select[2] = (uint8_t)((tables) & 31);
+ gr->preflag = HDR_TEST_MPEG1(hdr) ? get_bits(bs, 1) : (gr->scalefac_compress >= 500);
+ gr->scalefac_scale = (uint8_t)get_bits(bs, 1);
+ gr->count1_table = (uint8_t)get_bits(bs, 1);
+ gr->scfsi = (uint8_t)((scfsi >> 12) & 15);
+ scfsi <<= 4;
+ gr++;
+ } while(--gr_count);
+
+ if (part_23_sum + bs->pos > bs->limit + main_data_begin*8)
+ {
+ return -1;
+ }
+
+ return main_data_begin;
+}
+
+static void L3_read_scalefactors(uint8_t *scf, uint8_t *ist_pos, const uint8_t *scf_size, const uint8_t *scf_count, bs_t *bitbuf, int scfsi)
+{
+ int i, k;
+ for (i = 0; i < 4 && scf_count[i]; i++, scfsi *= 2)
+ {
+ int cnt = scf_count[i];
+ if (scfsi & 8)
+ {
+ memcpy(scf, ist_pos, cnt);
+ } else
+ {
+ int bits = scf_size[i];
+ if (!bits)
+ {
+ memset(scf, 0, cnt);
+ memset(ist_pos, 0, cnt);
+ } else
+ {
+ int max_scf = (scfsi < 0) ? (1 << bits) - 1 : -1;
+ for (k = 0; k < cnt; k++)
+ {
+ int s = get_bits(bitbuf, bits);
+ ist_pos[k] = (s == max_scf ? -1 : s);
+ scf[k] = s;
+ }
+ }
+ }
+ ist_pos += cnt;
+ scf += cnt;
+ }
+ scf[0] = scf[1] = scf[2] = 0;
+}
+
+static float L3_ldexp_q2(float y, int exp_q2)
+{
+ static const float g_expfrac[4] = { 9.31322575e-10f,7.83145814e-10f,6.58544508e-10f,5.53767716e-10f };
+ int e;
+ do
+ {
+ e = MINIMP3_MIN(30*4, exp_q2);
+ y *= g_expfrac[e & 3]*(1 << 30 >> (e >> 2));
+ } while ((exp_q2 -= e) > 0);
+ return y;
+}
+
+static void L3_decode_scalefactors(const uint8_t *hdr, uint8_t *ist_pos, bs_t *bs, const L3_gr_info_t *gr, float *scf, int ch)
+{
+ static const uint8_t g_scf_partitions[3][28] = {
+ { 6,5,5, 5,6,5,5,5,6,5, 7,3,11,10,0,0, 7, 7, 7,0, 6, 6,6,3, 8, 8,5,0 },
+ { 8,9,6,12,6,9,9,9,6,9,12,6,15,18,0,0, 6,15,12,0, 6,12,9,6, 6,18,9,0 },
+ { 9,9,6,12,9,9,9,9,9,9,12,6,18,18,0,0,12,12,12,0,12, 9,9,6,15,12,9,0 }
+ };
+ const uint8_t *scf_partition = g_scf_partitions[!!gr->n_short_sfb + !gr->n_long_sfb];
+ uint8_t scf_size[4], iscf[40];
+ int i, scf_shift = gr->scalefac_scale + 1, gain_exp, scfsi = gr->scfsi;
+ float gain;
+
+ if (HDR_TEST_MPEG1(hdr))
+ {
+ static const uint8_t g_scfc_decode[16] = { 0,1,2,3, 12,5,6,7, 9,10,11,13, 14,15,18,19 };
+ int part = g_scfc_decode[gr->scalefac_compress];
+ scf_size[1] = scf_size[0] = (uint8_t)(part >> 2);
+ scf_size[3] = scf_size[2] = (uint8_t)(part & 3);
+ } else
+ {
+ static const uint8_t g_mod[6*4] = { 5,5,4,4,5,5,4,1,4,3,1,1,5,6,6,1,4,4,4,1,4,3,1,1 };
+ int k, modprod, sfc, ist = HDR_TEST_I_STEREO(hdr) && ch;
+ sfc = gr->scalefac_compress >> ist;
+ for (k = ist*3*4; sfc >= 0; sfc -= modprod, k += 4)
+ {
+ for (modprod = 1, i = 3; i >= 0; i--)
+ {
+ scf_size[i] = (uint8_t)(sfc / modprod % g_mod[k + i]);
+ modprod *= g_mod[k + i];
+ }
+ }
+ scf_partition += k;
+ scfsi = -16;
+ }
+ L3_read_scalefactors(iscf, ist_pos, scf_size, scf_partition, bs, scfsi);
+
+ if (gr->n_short_sfb)
+ {
+ int sh = 3 - scf_shift;
+ for (i = 0; i < gr->n_short_sfb; i += 3)
+ {
+ iscf[gr->n_long_sfb + i + 0] += gr->subblock_gain[0] << sh;
+ iscf[gr->n_long_sfb + i + 1] += gr->subblock_gain[1] << sh;
+ iscf[gr->n_long_sfb + i + 2] += gr->subblock_gain[2] << sh;
+ }
+ } else if (gr->preflag)
+ {
+ static const uint8_t g_preamp[10] = { 1,1,1,1,2,2,3,3,3,2 };
+ for (i = 0; i < 10; i++)
+ {
+ iscf[11 + i] += g_preamp[i];
+ }
+ }
+
+ gain_exp = gr->global_gain + BITS_DEQUANTIZER_OUT*4 - 210 - (HDR_IS_MS_STEREO(hdr) ? 2 : 0);
+ gain = L3_ldexp_q2(1 << (MAX_SCFI/4), MAX_SCFI - gain_exp);
+ for (i = 0; i < (int)(gr->n_long_sfb + gr->n_short_sfb); i++)
+ {
+ scf[i] = L3_ldexp_q2(gain, iscf[i] << scf_shift);
+ }
+}
+
+static const float g_pow43[129 + 16] = {
+ 0,-1,-2.519842f,-4.326749f,-6.349604f,-8.549880f,-10.902724f,-13.390518f,-16.000000f,-18.720754f,-21.544347f,-24.463781f,-27.473142f,-30.567351f,-33.741992f,-36.993181f,
+ 0,1,2.519842f,4.326749f,6.349604f,8.549880f,10.902724f,13.390518f,16.000000f,18.720754f,21.544347f,24.463781f,27.473142f,30.567351f,33.741992f,36.993181f,40.317474f,43.711787f,47.173345f,50.699631f,54.288352f,57.937408f,61.644865f,65.408941f,69.227979f,73.100443f,77.024898f,81.000000f,85.024491f,89.097188f,93.216975f,97.382800f,101.593667f,105.848633f,110.146801f,114.487321f,118.869381f,123.292209f,127.755065f,132.257246f,136.798076f,141.376907f,145.993119f,150.646117f,155.335327f,160.060199f,164.820202f,169.614826f,174.443577f,179.305980f,184.201575f,189.129918f,194.090580f,199.083145f,204.107210f,209.162385f,214.248292f,219.364564f,224.510845f,229.686789f,234.892058f,240.126328f,245.389280f,250.680604f,256.000000f,261.347174f,266.721841f,272.123723f,277.552547f,283.008049f,288.489971f,293.998060f,299.532071f,305.091761f,310.676898f,316.287249f,321.922592f,327.582707f,333.267377f,338.976394f,344.709550f,350.466646f,356.247482f,362.051866f,367.879608f,373.730522f,379.604427f,385.501143f,391.420496f,397.362314f,403.326427f,409.312672f,415.320884f,421.350905f,427.402579f,433.475750f,439.570269f,445.685987f,451.822757f,457.980436f,464.158883f,470.357960f,476.577530f,482.817459f,489.077615f,495.357868f,501.658090f,507.978156f,514.317941f,520.677324f,527.056184f,533.454404f,539.871867f,546.308458f,552.764065f,559.238575f,565.731879f,572.243870f,578.774440f,585.323483f,591.890898f,598.476581f,605.080431f,611.702349f,618.342238f,625.000000f,631.675540f,638.368763f,645.079578f
+};
+
+static float L3_pow_43(int x)
+{
+ float frac;
+ int sign, mult = 256;
+
+ if (x < 129)
+ {
+ return g_pow43[16 + x];
+ }
+
+ if (x < 1024)
+ {
+ mult = 16;
+ x <<= 3;
+ }
+
+ sign = 2*x & 64;
+ frac = (float)((x & 63) - sign) / ((x & ~63) + sign);
+ return g_pow43[16 + ((x + sign) >> 6)]*(1.f + frac*((4.f/3) + frac*(2.f/9)))*mult;
+}
+
+static void L3_huffman(float *dst, bs_t *bs, const L3_gr_info_t *gr_info, const float *scf, int layer3gr_limit)
+{
+ static const int16_t tabs[] = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 785,785,785,785,784,784,784,784,513,513,513,513,513,513,513,513,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,
+ -255,1313,1298,1282,785,785,785,785,784,784,784,784,769,769,769,769,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,290,288,
+ -255,1313,1298,1282,769,769,769,769,529,529,529,529,529,529,529,529,528,528,528,528,528,528,528,528,512,512,512,512,512,512,512,512,290,288,
+ -253,-318,-351,-367,785,785,785,785,784,784,784,784,769,769,769,769,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,819,818,547,547,275,275,275,275,561,560,515,546,289,274,288,258,
+ -254,-287,1329,1299,1314,1312,1057,1057,1042,1042,1026,1026,784,784,784,784,529,529,529,529,529,529,529,529,769,769,769,769,768,768,768,768,563,560,306,306,291,259,
+ -252,-413,-477,-542,1298,-575,1041,1041,784,784,784,784,769,769,769,769,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,-383,-399,1107,1092,1106,1061,849,849,789,789,1104,1091,773,773,1076,1075,341,340,325,309,834,804,577,577,532,532,516,516,832,818,803,816,561,561,531,531,515,546,289,289,288,258,
+ -252,-429,-493,-559,1057,1057,1042,1042,529,529,529,529,529,529,529,529,784,784,784,784,769,769,769,769,512,512,512,512,512,512,512,512,-382,1077,-415,1106,1061,1104,849,849,789,789,1091,1076,1029,1075,834,834,597,581,340,340,339,324,804,833,532,532,832,772,818,803,817,787,816,771,290,290,290,290,288,258,
+ -253,-349,-414,-447,-463,1329,1299,-479,1314,1312,1057,1057,1042,1042,1026,1026,785,785,785,785,784,784,784,784,769,769,769,769,768,768,768,768,-319,851,821,-335,836,850,805,849,341,340,325,336,533,533,579,579,564,564,773,832,578,548,563,516,321,276,306,291,304,259,
+ -251,-572,-733,-830,-863,-879,1041,1041,784,784,784,784,769,769,769,769,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,-511,-527,-543,1396,1351,1381,1366,1395,1335,1380,-559,1334,1138,1138,1063,1063,1350,1392,1031,1031,1062,1062,1364,1363,1120,1120,1333,1348,881,881,881,881,375,374,359,373,343,358,341,325,791,791,1123,1122,-703,1105,1045,-719,865,865,790,790,774,774,1104,1029,338,293,323,308,-799,-815,833,788,772,818,803,816,322,292,307,320,561,531,515,546,289,274,288,258,
+ -251,-525,-605,-685,-765,-831,-846,1298,1057,1057,1312,1282,785,785,785,785,784,784,784,784,769,769,769,769,512,512,512,512,512,512,512,512,1399,1398,1383,1367,1382,1396,1351,-511,1381,1366,1139,1139,1079,1079,1124,1124,1364,1349,1363,1333,882,882,882,882,807,807,807,807,1094,1094,1136,1136,373,341,535,535,881,775,867,822,774,-591,324,338,-671,849,550,550,866,864,609,609,293,336,534,534,789,835,773,-751,834,804,308,307,833,788,832,772,562,562,547,547,305,275,560,515,290,290,
+ -252,-397,-477,-557,-622,-653,-719,-735,-750,1329,1299,1314,1057,1057,1042,1042,1312,1282,1024,1024,785,785,785,785,784,784,784,784,769,769,769,769,-383,1127,1141,1111,1126,1140,1095,1110,869,869,883,883,1079,1109,882,882,375,374,807,868,838,881,791,-463,867,822,368,263,852,837,836,-543,610,610,550,550,352,336,534,534,865,774,851,821,850,805,593,533,579,564,773,832,578,578,548,548,577,577,307,276,306,291,516,560,259,259,
+ -250,-2107,-2507,-2764,-2909,-2974,-3007,-3023,1041,1041,1040,1040,769,769,769,769,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,-767,-1052,-1213,-1277,-1358,-1405,-1469,-1535,-1550,-1582,-1614,-1647,-1662,-1694,-1726,-1759,-1774,-1807,-1822,-1854,-1886,1565,-1919,-1935,-1951,-1967,1731,1730,1580,1717,-1983,1729,1564,-1999,1548,-2015,-2031,1715,1595,-2047,1714,-2063,1610,-2079,1609,-2095,1323,1323,1457,1457,1307,1307,1712,1547,1641,1700,1699,1594,1685,1625,1442,1442,1322,1322,-780,-973,-910,1279,1278,1277,1262,1276,1261,1275,1215,1260,1229,-959,974,974,989,989,-943,735,478,478,495,463,506,414,-1039,1003,958,1017,927,942,987,957,431,476,1272,1167,1228,-1183,1256,-1199,895,895,941,941,1242,1227,1212,1135,1014,1014,490,489,503,487,910,1013,985,925,863,894,970,955,1012,847,-1343,831,755,755,984,909,428,366,754,559,-1391,752,486,457,924,997,698,698,983,893,740,740,908,877,739,739,667,667,953,938,497,287,271,271,683,606,590,712,726,574,302,302,738,736,481,286,526,725,605,711,636,724,696,651,589,681,666,710,364,467,573,695,466,466,301,465,379,379,709,604,665,679,316,316,634,633,436,436,464,269,424,394,452,332,438,363,347,408,393,448,331,422,362,407,392,421,346,406,391,376,375,359,1441,1306,-2367,1290,-2383,1337,-2399,-2415,1426,1321,-2431,1411,1336,-2447,-2463,-2479,1169,1169,1049,1049,1424,1289,1412,1352,1319,-2495,1154,1154,1064,1064,1153,1153,416,390,360,404,403,389,344,374,373,343,358,372,327,357,342,311,356,326,1395,1394,1137,1137,1047,1047,1365,1392,1287,1379,1334,1364,1349,1378,1318,1363,792,792,792,792,1152,1152,1032,1032,1121,1121,1046,1046,1120,1120,1030,1030,-2895,1106,1061,1104,849,849,789,789,1091,1076,1029,1090,1060,1075,833,833,309,324,532,532,832,772,818,803,561,561,531,560,515,546,289,274,288,258,
+ -250,-1179,-1579,-1836,-1996,-2124,-2253,-2333,-2413,-2477,-2542,-2574,-2607,-2622,-2655,1314,1313,1298,1312,1282,785,785,785,785,1040,1040,1025,1025,768,768,768,768,-766,-798,-830,-862,-895,-911,-927,-943,-959,-975,-991,-1007,-1023,-1039,-1055,-1070,1724,1647,-1103,-1119,1631,1767,1662,1738,1708,1723,-1135,1780,1615,1779,1599,1677,1646,1778,1583,-1151,1777,1567,1737,1692,1765,1722,1707,1630,1751,1661,1764,1614,1736,1676,1763,1750,1645,1598,1721,1691,1762,1706,1582,1761,1566,-1167,1749,1629,767,766,751,765,494,494,735,764,719,749,734,763,447,447,748,718,477,506,431,491,446,476,461,505,415,430,475,445,504,399,460,489,414,503,383,474,429,459,502,502,746,752,488,398,501,473,413,472,486,271,480,270,-1439,-1455,1357,-1471,-1487,-1503,1341,1325,-1519,1489,1463,1403,1309,-1535,1372,1448,1418,1476,1356,1462,1387,-1551,1475,1340,1447,1402,1386,-1567,1068,1068,1474,1461,455,380,468,440,395,425,410,454,364,467,466,464,453,269,409,448,268,432,1371,1473,1432,1417,1308,1460,1355,1446,1459,1431,1083,1083,1401,1416,1458,1445,1067,1067,1370,1457,1051,1051,1291,1430,1385,1444,1354,1415,1400,1443,1082,1082,1173,1113,1186,1066,1185,1050,-1967,1158,1128,1172,1097,1171,1081,-1983,1157,1112,416,266,375,400,1170,1142,1127,1065,793,793,1169,1033,1156,1096,1141,1111,1155,1080,1126,1140,898,898,808,808,897,897,792,792,1095,1152,1032,1125,1110,1139,1079,1124,882,807,838,881,853,791,-2319,867,368,263,822,852,837,866,806,865,-2399,851,352,262,534,534,821,836,594,594,549,549,593,593,533,533,848,773,579,579,564,578,548,563,276,276,577,576,306,291,516,560,305,305,275,259,
+ -251,-892,-2058,-2620,-2828,-2957,-3023,-3039,1041,1041,1040,1040,769,769,769,769,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,256,-511,-527,-543,-559,1530,-575,-591,1528,1527,1407,1526,1391,1023,1023,1023,1023,1525,1375,1268,1268,1103,1103,1087,1087,1039,1039,1523,-604,815,815,815,815,510,495,509,479,508,463,507,447,431,505,415,399,-734,-782,1262,-815,1259,1244,-831,1258,1228,-847,-863,1196,-879,1253,987,987,748,-767,493,493,462,477,414,414,686,669,478,446,461,445,474,429,487,458,412,471,1266,1264,1009,1009,799,799,-1019,-1276,-1452,-1581,-1677,-1757,-1821,-1886,-1933,-1997,1257,1257,1483,1468,1512,1422,1497,1406,1467,1496,1421,1510,1134,1134,1225,1225,1466,1451,1374,1405,1252,1252,1358,1480,1164,1164,1251,1251,1238,1238,1389,1465,-1407,1054,1101,-1423,1207,-1439,830,830,1248,1038,1237,1117,1223,1148,1236,1208,411,426,395,410,379,269,1193,1222,1132,1235,1221,1116,976,976,1192,1162,1177,1220,1131,1191,963,963,-1647,961,780,-1663,558,558,994,993,437,408,393,407,829,978,813,797,947,-1743,721,721,377,392,844,950,828,890,706,706,812,859,796,960,948,843,934,874,571,571,-1919,690,555,689,421,346,539,539,944,779,918,873,932,842,903,888,570,570,931,917,674,674,-2575,1562,-2591,1609,-2607,1654,1322,1322,1441,1441,1696,1546,1683,1593,1669,1624,1426,1426,1321,1321,1639,1680,1425,1425,1305,1305,1545,1668,1608,1623,1667,1592,1638,1666,1320,1320,1652,1607,1409,1409,1304,1304,1288,1288,1664,1637,1395,1395,1335,1335,1622,1636,1394,1394,1319,1319,1606,1621,1392,1392,1137,1137,1137,1137,345,390,360,375,404,373,1047,-2751,-2767,-2783,1062,1121,1046,-2799,1077,-2815,1106,1061,789,789,1105,1104,263,355,310,340,325,354,352,262,339,324,1091,1076,1029,1090,1060,1075,833,833,788,788,1088,1028,818,818,803,803,561,561,531,531,816,771,546,546,289,274,288,258,
+ -253,-317,-381,-446,-478,-509,1279,1279,-811,-1179,-1451,-1756,-1900,-2028,-2189,-2253,-2333,-2414,-2445,-2511,-2526,1313,1298,-2559,1041,1041,1040,1040,1025,1025,1024,1024,1022,1007,1021,991,1020,975,1019,959,687,687,1018,1017,671,671,655,655,1016,1015,639,639,758,758,623,623,757,607,756,591,755,575,754,559,543,543,1009,783,-575,-621,-685,-749,496,-590,750,749,734,748,974,989,1003,958,988,973,1002,942,987,957,972,1001,926,986,941,971,956,1000,910,985,925,999,894,970,-1071,-1087,-1102,1390,-1135,1436,1509,1451,1374,-1151,1405,1358,1480,1420,-1167,1507,1494,1389,1342,1465,1435,1450,1326,1505,1310,1493,1373,1479,1404,1492,1464,1419,428,443,472,397,736,526,464,464,486,457,442,471,484,482,1357,1449,1434,1478,1388,1491,1341,1490,1325,1489,1463,1403,1309,1477,1372,1448,1418,1433,1476,1356,1462,1387,-1439,1475,1340,1447,1402,1474,1324,1461,1371,1473,269,448,1432,1417,1308,1460,-1711,1459,-1727,1441,1099,1099,1446,1386,1431,1401,-1743,1289,1083,1083,1160,1160,1458,1445,1067,1067,1370,1457,1307,1430,1129,1129,1098,1098,268,432,267,416,266,400,-1887,1144,1187,1082,1173,1113,1186,1066,1050,1158,1128,1143,1172,1097,1171,1081,420,391,1157,1112,1170,1142,1127,1065,1169,1049,1156,1096,1141,1111,1155,1080,1126,1154,1064,1153,1140,1095,1048,-2159,1125,1110,1137,-2175,823,823,1139,1138,807,807,384,264,368,263,868,838,853,791,867,822,852,837,866,806,865,790,-2319,851,821,836,352,262,850,805,849,-2399,533,533,835,820,336,261,578,548,563,577,532,532,832,772,562,562,547,547,305,275,560,515,290,290,288,258 };
+ static const uint8_t tab32[] = { 130,162,193,209,44,28,76,140,9,9,9,9,9,9,9,9,190,254,222,238,126,94,157,157,109,61,173,205 };
+ static const uint8_t tab33[] = { 252,236,220,204,188,172,156,140,124,108,92,76,60,44,28,12 };
+ static const int16_t tabindex[2*16] = { 0,32,64,98,0,132,180,218,292,364,426,538,648,746,0,1126,1460,1460,1460,1460,1460,1460,1460,1460,1842,1842,1842,1842,1842,1842,1842,1842 };
+ static const uint8_t g_linbits[] = { 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,2,3,4,6,8,10,13,4,5,6,7,8,9,11,13 };
+
+#define PEEK_BITS(n) (bs_cache >> (32 - n))
+#define FLUSH_BITS(n) { bs_cache <<= (n); bs_sh += (n); }
+#define CHECK_BITS while (bs_sh >= 0) { bs_cache |= (uint32_t)*bs_next_ptr++ << bs_sh; bs_sh -= 8; }
+#define BSPOS ((bs_next_ptr - bs->buf)*8 - 24 + bs_sh)
+
+ float one = 0.0f;
+ int ireg = 0, big_val_cnt = gr_info->big_values;
+ const uint8_t *sfb = gr_info->sfbtab;
+ const uint8_t *bs_next_ptr = bs->buf + bs->pos/8;
+ uint32_t bs_cache = (((bs_next_ptr[0]*256u + bs_next_ptr[1])*256u + bs_next_ptr[2])*256u + bs_next_ptr[3]) << (bs->pos & 7);
+ int pairs_to_decode, np, bs_sh = (bs->pos & 7) - 8;
+ bs_next_ptr += 4;
+
+ while (big_val_cnt > 0)
+ {
+ int tab_num = gr_info->table_select[ireg];
+ int sfb_cnt = gr_info->region_count[ireg++];
+ const int16_t *codebook = tabs + tabindex[tab_num];
+ int linbits = g_linbits[tab_num];
+ if (linbits)
+ {
+ do
+ {
+ np = *sfb++ / 2;
+ pairs_to_decode = MINIMP3_MIN(big_val_cnt, np);
+ one = *scf++;
+ do
+ {
+ int j, w = 5;
+ int leaf = codebook[PEEK_BITS(w)];
+ while (leaf < 0)
+ {
+ FLUSH_BITS(w);
+ w = leaf & 7;
+ leaf = codebook[PEEK_BITS(w) - (leaf >> 3)];
+ }
+ FLUSH_BITS(leaf >> 8);
+
+ for (j = 0; j < 2; j++, dst++, leaf >>= 4)
+ {
+ int lsb = leaf & 0x0F;
+ if (lsb == 15)
+ {
+ lsb += PEEK_BITS(linbits);
+ FLUSH_BITS(linbits);
+ CHECK_BITS;
+ *dst = one*L3_pow_43(lsb)*((int32_t)bs_cache < 0 ? -1: 1);
+ } else
+ {
+ *dst = g_pow43[16 + lsb - 16*(bs_cache >> 31)]*one;
+ }
+ FLUSH_BITS(lsb ? 1 : 0);
+ }
+ CHECK_BITS;
+ } while (--pairs_to_decode);
+ } while ((big_val_cnt -= np) > 0 && --sfb_cnt >= 0);
+ } else
+ {
+ do
+ {
+ np = *sfb++ / 2;
+ pairs_to_decode = MINIMP3_MIN(big_val_cnt, np);
+ one = *scf++;
+ do
+ {
+ int j, w = 5;
+ int leaf = codebook[PEEK_BITS(w)];
+ while (leaf < 0)
+ {
+ FLUSH_BITS(w);
+ w = leaf & 7;
+ leaf = codebook[PEEK_BITS(w) - (leaf >> 3)];
+ }
+ FLUSH_BITS(leaf >> 8);
+
+ for (j = 0; j < 2; j++, dst++, leaf >>= 4)
+ {
+ int lsb = leaf & 0x0F;
+ *dst = g_pow43[16 + lsb - 16*(bs_cache >> 31)]*one;
+ FLUSH_BITS(lsb ? 1 : 0);
+ }
+ CHECK_BITS;
+ } while (--pairs_to_decode);
+ } while ((big_val_cnt -= np) > 0 && --sfb_cnt >= 0);
+ }
+ }
+
+ for (np = 1 - big_val_cnt;; dst += 4)
+ {
+ const uint8_t *codebook_count1 = (gr_info->count1_table) ? tab33 : tab32;
+ int leaf = codebook_count1[PEEK_BITS(4)];
+ if (!(leaf & 8))
+ {
+ leaf = codebook_count1[(leaf >> 3) + (bs_cache << 4 >> (32 - (leaf & 3)))];
+ }
+ FLUSH_BITS(leaf & 7);
+ if (BSPOS > layer3gr_limit)
+ {
+ break;
+ }
+#define RELOAD_SCALEFACTOR if (!--np) { np = *sfb++/2; if (!np) break; one = *scf++; }
+#define DEQ_COUNT1(s) if (leaf & (128 >> s)) { dst[s] = ((int32_t)bs_cache < 0) ? -one : one; FLUSH_BITS(1) }
+ RELOAD_SCALEFACTOR;
+ DEQ_COUNT1(0);
+ DEQ_COUNT1(1);
+ RELOAD_SCALEFACTOR;
+ DEQ_COUNT1(2);
+ DEQ_COUNT1(3);
+ CHECK_BITS;
+ }
+
+ bs->pos = layer3gr_limit;
+}
+
+static void L3_midside_stereo(float *left, int n)
+{
+ int i = 0;
+ float *right = left + 576;
+#if HAVE_SIMD
+ if (have_simd())
+ {
+ for (; i < n - 3; i += 4)
+ {
+ f4 vl = VLD(left + i);
+ f4 vr = VLD(right + i);
+ VSTORE(left + i, VADD(vl, vr));
+ VSTORE(right + i, VSUB(vl, vr));
+ }
+#ifdef __GNUC__
+ /* Workaround for spurious -Waggressive-loop-optimizations warning from gcc.
+ * For more info see: https://github.com/lieff/minimp3/issues/88
+ */
+ if (__builtin_constant_p(n % 4 == 0) && n % 4 == 0)
+ return;
+#endif
+ }
+#endif /* HAVE_SIMD */
+ for (; i < n; i++)
+ {
+ float a = left[i];
+ float b = right[i];
+ left[i] = a + b;
+ right[i] = a - b;
+ }
+}
+
+static void L3_intensity_stereo_band(float *left, int n, float kl, float kr)
+{
+ int i;
+ for (i = 0; i < n; i++)
+ {
+ left[i + 576] = left[i]*kr;
+ left[i] = left[i]*kl;
+ }
+}
+
+static void L3_stereo_top_band(const float *right, const uint8_t *sfb, int nbands, int max_band[3])
+{
+ int i, k;
+
+ max_band[0] = max_band[1] = max_band[2] = -1;
+
+ for (i = 0; i < nbands; i++)
+ {
+ for (k = 0; k < sfb[i]; k += 2)
+ {
+ if (right[k] != 0 || right[k + 1] != 0)
+ {
+ max_band[i % 3] = i;
+ break;
+ }
+ }
+ right += sfb[i];
+ }
+}
+
+static void L3_stereo_process(float *left, const uint8_t *ist_pos, const uint8_t *sfb, const uint8_t *hdr, int max_band[3], int mpeg2_sh)
+{
+ static const float g_pan[7*2] = { 0,1,0.21132487f,0.78867513f,0.36602540f,0.63397460f,0.5f,0.5f,0.63397460f,0.36602540f,0.78867513f,0.21132487f,1,0 };
+ unsigned i, max_pos = HDR_TEST_MPEG1(hdr) ? 7 : 64;
+
+ for (i = 0; sfb[i]; i++)
+ {
+ unsigned ipos = ist_pos[i];
+ if ((int)i > max_band[i % 3] && ipos < max_pos)
+ {
+ float kl, kr, s = HDR_TEST_MS_STEREO(hdr) ? 1.41421356f : 1;
+ if (HDR_TEST_MPEG1(hdr))
+ {
+ kl = g_pan[2*ipos];
+ kr = g_pan[2*ipos + 1];
+ } else
+ {
+ kl = 1;
+ kr = L3_ldexp_q2(1, (ipos + 1) >> 1 << mpeg2_sh);
+ if (ipos & 1)
+ {
+ kl = kr;
+ kr = 1;
+ }
+ }
+ L3_intensity_stereo_band(left, sfb[i], kl*s, kr*s);
+ } else if (HDR_TEST_MS_STEREO(hdr))
+ {
+ L3_midside_stereo(left, sfb[i]);
+ }
+ left += sfb[i];
+ }
+}
+
+static void L3_intensity_stereo(float *left, uint8_t *ist_pos, const L3_gr_info_t *gr, const uint8_t *hdr)
+{
+ int max_band[3], n_sfb = gr->n_long_sfb + gr->n_short_sfb;
+ int i, max_blocks = gr->n_short_sfb ? 3 : 1;
+
+ L3_stereo_top_band(left + 576, gr->sfbtab, n_sfb, max_band);
+ if (gr->n_long_sfb)
+ {
+ max_band[0] = max_band[1] = max_band[2] = MINIMP3_MAX(MINIMP3_MAX(max_band[0], max_band[1]), max_band[2]);
+ }
+ for (i = 0; i < max_blocks; i++)
+ {
+ int default_pos = HDR_TEST_MPEG1(hdr) ? 3 : 0;
+ int itop = n_sfb - max_blocks + i;
+ int prev = itop - max_blocks;
+ ist_pos[itop] = max_band[i] >= prev ? default_pos : ist_pos[prev];
+ }
+ L3_stereo_process(left, ist_pos, gr->sfbtab, hdr, max_band, gr[1].scalefac_compress & 1);
+}
+
+static void L3_reorder(float *grbuf, float *scratch, const uint8_t *sfb)
+{
+ int i, len;
+ float *src = grbuf, *dst = scratch;
+
+ for (;0 != (len = *sfb); sfb += 3, src += 2*len)
+ {
+ for (i = 0; i < len; i++, src++)
+ {
+ *dst++ = src[0*len];
+ *dst++ = src[1*len];
+ *dst++ = src[2*len];
+ }
+ }
+ memcpy(grbuf, scratch, (dst - scratch)*sizeof(float));
+}
+
+static void L3_antialias(float *grbuf, int nbands)
+{
+ static const float g_aa[2][8] = {
+ {0.85749293f,0.88174200f,0.94962865f,0.98331459f,0.99551782f,0.99916056f,0.99989920f,0.99999316f},
+ {0.51449576f,0.47173197f,0.31337745f,0.18191320f,0.09457419f,0.04096558f,0.01419856f,0.00369997f}
+ };
+
+ for (; nbands > 0; nbands--, grbuf += 18)
+ {
+ int i = 0;
+#if HAVE_SIMD
+ if (have_simd()) for (; i < 8; i += 4)
+ {
+ f4 vu = VLD(grbuf + 18 + i);
+ f4 vd = VLD(grbuf + 14 - i);
+ f4 vc0 = VLD(g_aa[0] + i);
+ f4 vc1 = VLD(g_aa[1] + i);
+ vd = VREV(vd);
+ VSTORE(grbuf + 18 + i, VSUB(VMUL(vu, vc0), VMUL(vd, vc1)));
+ vd = VADD(VMUL(vu, vc1), VMUL(vd, vc0));
+ VSTORE(grbuf + 14 - i, VREV(vd));
+ }
+#endif /* HAVE_SIMD */
+#ifndef MINIMP3_ONLY_SIMD
+ for(; i < 8; i++)
+ {
+ float u = grbuf[18 + i];
+ float d = grbuf[17 - i];
+ grbuf[18 + i] = u*g_aa[0][i] - d*g_aa[1][i];
+ grbuf[17 - i] = u*g_aa[1][i] + d*g_aa[0][i];
+ }
+#endif /* MINIMP3_ONLY_SIMD */
+ }
+}
+
+static void L3_dct3_9(float *y)
+{
+ float s0, s1, s2, s3, s4, s5, s6, s7, s8, t0, t2, t4;
+
+ s0 = y[0]; s2 = y[2]; s4 = y[4]; s6 = y[6]; s8 = y[8];
+ t0 = s0 + s6*0.5f;
+ s0 -= s6;
+ t4 = (s4 + s2)*0.93969262f;
+ t2 = (s8 + s2)*0.76604444f;
+ s6 = (s4 - s8)*0.17364818f;
+ s4 += s8 - s2;
+
+ s2 = s0 - s4*0.5f;
+ y[4] = s4 + s0;
+ s8 = t0 - t2 + s6;
+ s0 = t0 - t4 + t2;
+ s4 = t0 + t4 - s6;
+
+ s1 = y[1]; s3 = y[3]; s5 = y[5]; s7 = y[7];
+
+ s3 *= 0.86602540f;
+ t0 = (s5 + s1)*0.98480775f;
+ t4 = (s5 - s7)*0.34202014f;
+ t2 = (s1 + s7)*0.64278761f;
+ s1 = (s1 - s5 - s7)*0.86602540f;
+
+ s5 = t0 - s3 - t2;
+ s7 = t4 - s3 - t0;
+ s3 = t4 + s3 - t2;
+
+ y[0] = s4 - s7;
+ y[1] = s2 + s1;
+ y[2] = s0 - s3;
+ y[3] = s8 + s5;
+ y[5] = s8 - s5;
+ y[6] = s0 + s3;
+ y[7] = s2 - s1;
+ y[8] = s4 + s7;
+}
+
+static void L3_imdct36(float *grbuf, float *overlap, const float *window, int nbands)
+{
+ int i, j;
+ static const float g_twid9[18] = {
+ 0.73727734f,0.79335334f,0.84339145f,0.88701083f,0.92387953f,0.95371695f,0.97629601f,0.99144486f,0.99904822f,0.67559021f,0.60876143f,0.53729961f,0.46174861f,0.38268343f,0.30070580f,0.21643961f,0.13052619f,0.04361938f
+ };
+
+ for (j = 0; j < nbands; j++, grbuf += 18, overlap += 9)
+ {
+ float co[9], si[9];
+ co[0] = -grbuf[0];
+ si[0] = grbuf[17];
+ for (i = 0; i < 4; i++)
+ {
+ si[8 - 2*i] = grbuf[4*i + 1] - grbuf[4*i + 2];
+ co[1 + 2*i] = grbuf[4*i + 1] + grbuf[4*i + 2];
+ si[7 - 2*i] = grbuf[4*i + 4] - grbuf[4*i + 3];
+ co[2 + 2*i] = -(grbuf[4*i + 3] + grbuf[4*i + 4]);
+ }
+ L3_dct3_9(co);
+ L3_dct3_9(si);
+
+ si[1] = -si[1];
+ si[3] = -si[3];
+ si[5] = -si[5];
+ si[7] = -si[7];
+
+ i = 0;
+
+#if HAVE_SIMD
+ if (have_simd()) for (; i < 8; i += 4)
+ {
+ f4 vovl = VLD(overlap + i);
+ f4 vc = VLD(co + i);
+ f4 vs = VLD(si + i);
+ f4 vr0 = VLD(g_twid9 + i);
+ f4 vr1 = VLD(g_twid9 + 9 + i);
+ f4 vw0 = VLD(window + i);
+ f4 vw1 = VLD(window + 9 + i);
+ f4 vsum = VADD(VMUL(vc, vr1), VMUL(vs, vr0));
+ VSTORE(overlap + i, VSUB(VMUL(vc, vr0), VMUL(vs, vr1)));
+ VSTORE(grbuf + i, VSUB(VMUL(vovl, vw0), VMUL(vsum, vw1)));
+ vsum = VADD(VMUL(vovl, vw1), VMUL(vsum, vw0));
+ VSTORE(grbuf + 14 - i, VREV(vsum));
+ }
+#endif /* HAVE_SIMD */
+ for (; i < 9; i++)
+ {
+ float ovl = overlap[i];
+ float sum = co[i]*g_twid9[9 + i] + si[i]*g_twid9[0 + i];
+ overlap[i] = co[i]*g_twid9[0 + i] - si[i]*g_twid9[9 + i];
+ grbuf[i] = ovl*window[0 + i] - sum*window[9 + i];
+ grbuf[17 - i] = ovl*window[9 + i] + sum*window[0 + i];
+ }
+ }
+}
+
+static void L3_idct3(float x0, float x1, float x2, float *dst)
+{
+ float m1 = x1*0.86602540f;
+ float a1 = x0 - x2*0.5f;
+ dst[1] = x0 + x2;
+ dst[0] = a1 + m1;
+ dst[2] = a1 - m1;
+}
+
+static void L3_imdct12(float *x, float *dst, float *overlap)
+{
+ static const float g_twid3[6] = { 0.79335334f,0.92387953f,0.99144486f, 0.60876143f,0.38268343f,0.13052619f };
+ float co[3], si[3];
+ int i;
+
+ L3_idct3(-x[0], x[6] + x[3], x[12] + x[9], co);
+ L3_idct3(x[15], x[12] - x[9], x[6] - x[3], si);
+ si[1] = -si[1];
+
+ for (i = 0; i < 3; i++)
+ {
+ float ovl = overlap[i];
+ float sum = co[i]*g_twid3[3 + i] + si[i]*g_twid3[0 + i];
+ overlap[i] = co[i]*g_twid3[0 + i] - si[i]*g_twid3[3 + i];
+ dst[i] = ovl*g_twid3[2 - i] - sum*g_twid3[5 - i];
+ dst[5 - i] = ovl*g_twid3[5 - i] + sum*g_twid3[2 - i];
+ }
+}
+
+static void L3_imdct_short(float *grbuf, float *overlap, int nbands)
+{
+ for (;nbands > 0; nbands--, overlap += 9, grbuf += 18)
+ {
+ float tmp[18];
+ memcpy(tmp, grbuf, sizeof(tmp));
+ memcpy(grbuf, overlap, 6*sizeof(float));
+ L3_imdct12(tmp, grbuf + 6, overlap + 6);
+ L3_imdct12(tmp + 1, grbuf + 12, overlap + 6);
+ L3_imdct12(tmp + 2, overlap, overlap + 6);
+ }
+}
+
+static void L3_change_sign(float *grbuf)
+{
+ int b, i;
+ for (b = 0, grbuf += 18; b < 32; b += 2, grbuf += 36)
+ for (i = 1; i < 18; i += 2)
+ grbuf[i] = -grbuf[i];
+}
+
+static void L3_imdct_gr(float *grbuf, float *overlap, unsigned block_type, unsigned n_long_bands)
+{
+ static const float g_mdct_window[2][18] = {
+ { 0.99904822f,0.99144486f,0.97629601f,0.95371695f,0.92387953f,0.88701083f,0.84339145f,0.79335334f,0.73727734f,0.04361938f,0.13052619f,0.21643961f,0.30070580f,0.38268343f,0.46174861f,0.53729961f,0.60876143f,0.67559021f },
+ { 1,1,1,1,1,1,0.99144486f,0.92387953f,0.79335334f,0,0,0,0,0,0,0.13052619f,0.38268343f,0.60876143f }
+ };
+ if (n_long_bands)
+ {
+ L3_imdct36(grbuf, overlap, g_mdct_window[0], n_long_bands);
+ grbuf += 18*n_long_bands;
+ overlap += 9*n_long_bands;
+ }
+ if (block_type == SHORT_BLOCK_TYPE)
+ L3_imdct_short(grbuf, overlap, 32 - n_long_bands);
+ else
+ L3_imdct36(grbuf, overlap, g_mdct_window[block_type == STOP_BLOCK_TYPE], 32 - n_long_bands);
+}
+
+static void L3_save_reservoir(mp3dec_t *h, mp3dec_scratch_t *s)
+{
+ int pos = (s->bs.pos + 7)/8u;
+ int remains = s->bs.limit/8u - pos;
+ if (remains > MAX_BITRESERVOIR_BYTES)
+ {
+ pos += remains - MAX_BITRESERVOIR_BYTES;
+ remains = MAX_BITRESERVOIR_BYTES;
+ }
+ if (remains > 0)
+ {
+ memmove(h->reserv_buf, s->maindata + pos, remains);
+ }
+ h->reserv = remains;
+}
+
+static int L3_restore_reservoir(mp3dec_t *h, bs_t *bs, mp3dec_scratch_t *s, int main_data_begin)
+{
+ int frame_bytes = (bs->limit - bs->pos)/8;
+ int bytes_have = MINIMP3_MIN(h->reserv, main_data_begin);
+ memcpy(s->maindata, h->reserv_buf + MINIMP3_MAX(0, h->reserv - main_data_begin), MINIMP3_MIN(h->reserv, main_data_begin));
+ memcpy(s->maindata + bytes_have, bs->buf + bs->pos/8, frame_bytes);
+ bs_init(&s->bs, s->maindata, bytes_have + frame_bytes);
+ return h->reserv >= main_data_begin;
+}
+
+static void L3_decode(mp3dec_t *h, mp3dec_scratch_t *s, L3_gr_info_t *gr_info, int nch)
+{
+ int ch;
+
+ for (ch = 0; ch < nch; ch++)
+ {
+ int layer3gr_limit = s->bs.pos + gr_info[ch].part_23_length;
+ L3_decode_scalefactors(h->header, s->ist_pos[ch], &s->bs, gr_info + ch, s->scf, ch);
+ L3_huffman(s->grbuf[ch], &s->bs, gr_info + ch, s->scf, layer3gr_limit);
+ }
+
+ if (HDR_TEST_I_STEREO(h->header))
+ {
+ L3_intensity_stereo(s->grbuf[0], s->ist_pos[1], gr_info, h->header);
+ } else if (HDR_IS_MS_STEREO(h->header))
+ {
+ L3_midside_stereo(s->grbuf[0], 576);
+ }
+
+ for (ch = 0; ch < nch; ch++, gr_info++)
+ {
+ int aa_bands = 31;
+ int n_long_bands = (gr_info->mixed_block_flag ? 2 : 0) << (int)(HDR_GET_MY_SAMPLE_RATE(h->header) == 2);
+
+ if (gr_info->n_short_sfb)
+ {
+ aa_bands = n_long_bands - 1;
+ L3_reorder(s->grbuf[ch] + n_long_bands*18, s->syn[0], gr_info->sfbtab + gr_info->n_long_sfb);
+ }
+
+ L3_antialias(s->grbuf[ch], aa_bands);
+ L3_imdct_gr(s->grbuf[ch], h->mdct_overlap[ch], gr_info->block_type, n_long_bands);
+ L3_change_sign(s->grbuf[ch]);
+ }
+}
+
+static void mp3d_DCT_II(float *grbuf, int n)
+{
+ static const float g_sec[24] = {
+ 10.19000816f,0.50060302f,0.50241929f,3.40760851f,0.50547093f,0.52249861f,2.05778098f,0.51544732f,0.56694406f,1.48416460f,0.53104258f,0.64682180f,1.16943991f,0.55310392f,0.78815460f,0.97256821f,0.58293498f,1.06067765f,0.83934963f,0.62250412f,1.72244716f,0.74453628f,0.67480832f,5.10114861f
+ };
+ int i, k = 0;
+#if HAVE_SIMD
+ if (have_simd()) for (; k < n; k += 4)
+ {
+ f4 t[4][8], *x;
+ float *y = grbuf + k;
+
+ for (x = t[0], i = 0; i < 8; i++, x++)
+ {
+ f4 x0 = VLD(&y[i*18]);
+ f4 x1 = VLD(&y[(15 - i)*18]);
+ f4 x2 = VLD(&y[(16 + i)*18]);
+ f4 x3 = VLD(&y[(31 - i)*18]);
+ f4 t0 = VADD(x0, x3);
+ f4 t1 = VADD(x1, x2);
+ f4 t2 = VMUL_S(VSUB(x1, x2), g_sec[3*i + 0]);
+ f4 t3 = VMUL_S(VSUB(x0, x3), g_sec[3*i + 1]);
+ x[0] = VADD(t0, t1);
+ x[8] = VMUL_S(VSUB(t0, t1), g_sec[3*i + 2]);
+ x[16] = VADD(t3, t2);
+ x[24] = VMUL_S(VSUB(t3, t2), g_sec[3*i + 2]);
+ }
+ for (x = t[0], i = 0; i < 4; i++, x += 8)
+ {
+ f4 x0 = x[0], x1 = x[1], x2 = x[2], x3 = x[3], x4 = x[4], x5 = x[5], x6 = x[6], x7 = x[7], xt;
+ xt = VSUB(x0, x7); x0 = VADD(x0, x7);
+ x7 = VSUB(x1, x6); x1 = VADD(x1, x6);
+ x6 = VSUB(x2, x5); x2 = VADD(x2, x5);
+ x5 = VSUB(x3, x4); x3 = VADD(x3, x4);
+ x4 = VSUB(x0, x3); x0 = VADD(x0, x3);
+ x3 = VSUB(x1, x2); x1 = VADD(x1, x2);
+ x[0] = VADD(x0, x1);
+ x[4] = VMUL_S(VSUB(x0, x1), 0.70710677f);
+ x5 = VADD(x5, x6);
+ x6 = VMUL_S(VADD(x6, x7), 0.70710677f);
+ x7 = VADD(x7, xt);
+ x3 = VMUL_S(VADD(x3, x4), 0.70710677f);
+ x5 = VSUB(x5, VMUL_S(x7, 0.198912367f)); /* rotate by PI/8 */
+ x7 = VADD(x7, VMUL_S(x5, 0.382683432f));
+ x5 = VSUB(x5, VMUL_S(x7, 0.198912367f));
+ x0 = VSUB(xt, x6); xt = VADD(xt, x6);
+ x[1] = VMUL_S(VADD(xt, x7), 0.50979561f);
+ x[2] = VMUL_S(VADD(x4, x3), 0.54119611f);
+ x[3] = VMUL_S(VSUB(x0, x5), 0.60134488f);
+ x[5] = VMUL_S(VADD(x0, x5), 0.89997619f);
+ x[6] = VMUL_S(VSUB(x4, x3), 1.30656302f);
+ x[7] = VMUL_S(VSUB(xt, x7), 2.56291556f);
+ }
+
+ if (k > n - 3)
+ {
+#if HAVE_SSE
+#define VSAVE2(i, v) _mm_storel_pi((__m64 *)(void*)&y[i*18], v)
+#else /* HAVE_SSE */
+#define VSAVE2(i, v) vst1_f32((float32_t *)&y[i*18], vget_low_f32(v))
+#endif /* HAVE_SSE */
+ for (i = 0; i < 7; i++, y += 4*18)
+ {
+ f4 s = VADD(t[3][i], t[3][i + 1]);
+ VSAVE2(0, t[0][i]);
+ VSAVE2(1, VADD(t[2][i], s));
+ VSAVE2(2, VADD(t[1][i], t[1][i + 1]));
+ VSAVE2(3, VADD(t[2][1 + i], s));
+ }
+ VSAVE2(0, t[0][7]);
+ VSAVE2(1, VADD(t[2][7], t[3][7]));
+ VSAVE2(2, t[1][7]);
+ VSAVE2(3, t[3][7]);
+ } else
+ {
+#define VSAVE4(i, v) VSTORE(&y[i*18], v)
+ for (i = 0; i < 7; i++, y += 4*18)
+ {
+ f4 s = VADD(t[3][i], t[3][i + 1]);
+ VSAVE4(0, t[0][i]);
+ VSAVE4(1, VADD(t[2][i], s));
+ VSAVE4(2, VADD(t[1][i], t[1][i + 1]));
+ VSAVE4(3, VADD(t[2][1 + i], s));
+ }
+ VSAVE4(0, t[0][7]);
+ VSAVE4(1, VADD(t[2][7], t[3][7]));
+ VSAVE4(2, t[1][7]);
+ VSAVE4(3, t[3][7]);
+ }
+ } else
+#endif /* HAVE_SIMD */
+#ifdef MINIMP3_ONLY_SIMD
+ {} /* for HAVE_SIMD=1, MINIMP3_ONLY_SIMD=1 case we do not need non-intrinsic "else" branch */
+#else /* MINIMP3_ONLY_SIMD */
+ for (; k < n; k++)
+ {
+ float t[4][8], *x, *y = grbuf + k;
+
+ for (x = t[0], i = 0; i < 8; i++, x++)
+ {
+ float x0 = y[i*18];
+ float x1 = y[(15 - i)*18];
+ float x2 = y[(16 + i)*18];
+ float x3 = y[(31 - i)*18];
+ float t0 = x0 + x3;
+ float t1 = x1 + x2;
+ float t2 = (x1 - x2)*g_sec[3*i + 0];
+ float t3 = (x0 - x3)*g_sec[3*i + 1];
+ x[0] = t0 + t1;
+ x[8] = (t0 - t1)*g_sec[3*i + 2];
+ x[16] = t3 + t2;
+ x[24] = (t3 - t2)*g_sec[3*i + 2];
+ }
+ for (x = t[0], i = 0; i < 4; i++, x += 8)
+ {
+ float x0 = x[0], x1 = x[1], x2 = x[2], x3 = x[3], x4 = x[4], x5 = x[5], x6 = x[6], x7 = x[7], xt;
+ xt = x0 - x7; x0 += x7;
+ x7 = x1 - x6; x1 += x6;
+ x6 = x2 - x5; x2 += x5;
+ x5 = x3 - x4; x3 += x4;
+ x4 = x0 - x3; x0 += x3;
+ x3 = x1 - x2; x1 += x2;
+ x[0] = x0 + x1;
+ x[4] = (x0 - x1)*0.70710677f;
+ x5 = x5 + x6;
+ x6 = (x6 + x7)*0.70710677f;
+ x7 = x7 + xt;
+ x3 = (x3 + x4)*0.70710677f;
+ x5 -= x7*0.198912367f; /* rotate by PI/8 */
+ x7 += x5*0.382683432f;
+ x5 -= x7*0.198912367f;
+ x0 = xt - x6; xt += x6;
+ x[1] = (xt + x7)*0.50979561f;
+ x[2] = (x4 + x3)*0.54119611f;
+ x[3] = (x0 - x5)*0.60134488f;
+ x[5] = (x0 + x5)*0.89997619f;
+ x[6] = (x4 - x3)*1.30656302f;
+ x[7] = (xt - x7)*2.56291556f;
+
+ }
+ for (i = 0; i < 7; i++, y += 4*18)
+ {
+ y[0*18] = t[0][i];
+ y[1*18] = t[2][i] + t[3][i] + t[3][i + 1];
+ y[2*18] = t[1][i] + t[1][i + 1];
+ y[3*18] = t[2][i + 1] + t[3][i] + t[3][i + 1];
+ }
+ y[0*18] = t[0][7];
+ y[1*18] = t[2][7] + t[3][7];
+ y[2*18] = t[1][7];
+ y[3*18] = t[3][7];
+ }
+#endif /* MINIMP3_ONLY_SIMD */
+}
+
+#ifndef MINIMP3_FLOAT_OUTPUT
+static int16_t mp3d_scale_pcm(float sample)
+{
+#if HAVE_ARMV6
+ int32_t s32 = (int32_t)(sample + .5f);
+ s32 -= (s32 < 0);
+ int16_t s = (int16_t)minimp3_clip_int16_arm(s32);
+#else
+ if (sample >= 32766.5) return (int16_t) 32767;
+ if (sample <= -32767.5) return (int16_t)-32768;
+ int16_t s = (int16_t)(sample + .5f);
+ s -= (s < 0); /* away from zero, to be compliant */
+#endif
+ return s;
+}
+#else /* MINIMP3_FLOAT_OUTPUT */
+static float mp3d_scale_pcm(float sample)
+{
+ return sample*(1.f/32768.f);
+}
+#endif /* MINIMP3_FLOAT_OUTPUT */
+
+static void mp3d_synth_pair(mp3d_sample_t *pcm, int nch, const float *z)
+{
+ float a;
+ a = (z[14*64] - z[ 0]) * 29;
+ a += (z[ 1*64] + z[13*64]) * 213;
+ a += (z[12*64] - z[ 2*64]) * 459;
+ a += (z[ 3*64] + z[11*64]) * 2037;
+ a += (z[10*64] - z[ 4*64]) * 5153;
+ a += (z[ 5*64] + z[ 9*64]) * 6574;
+ a += (z[ 8*64] - z[ 6*64]) * 37489;
+ a += z[ 7*64] * 75038;
+ pcm[0] = mp3d_scale_pcm(a);
+
+ z += 2;
+ a = z[14*64] * 104;
+ a += z[12*64] * 1567;
+ a += z[10*64] * 9727;
+ a += z[ 8*64] * 64019;
+ a += z[ 6*64] * -9975;
+ a += z[ 4*64] * -45;
+ a += z[ 2*64] * 146;
+ a += z[ 0*64] * -5;
+ pcm[16*nch] = mp3d_scale_pcm(a);
+}
+
+static void mp3d_synth(float *xl, mp3d_sample_t *dstl, int nch, float *lins)
+{
+ int i;
+ float *xr = xl + 576*(nch - 1);
+ mp3d_sample_t *dstr = dstl + (nch - 1);
+
+ static const float g_win[] = {
+ -1,26,-31,208,218,401,-519,2063,2000,4788,-5517,7134,5959,35640,-39336,74992,
+ -1,24,-35,202,222,347,-581,2080,1952,4425,-5879,7640,5288,33791,-41176,74856,
+ -1,21,-38,196,225,294,-645,2087,1893,4063,-6237,8092,4561,31947,-43006,74630,
+ -1,19,-41,190,227,244,-711,2085,1822,3705,-6589,8492,3776,30112,-44821,74313,
+ -1,17,-45,183,228,197,-779,2075,1739,3351,-6935,8840,2935,28289,-46617,73908,
+ -1,16,-49,176,228,153,-848,2057,1644,3004,-7271,9139,2037,26482,-48390,73415,
+ -2,14,-53,169,227,111,-919,2032,1535,2663,-7597,9389,1082,24694,-50137,72835,
+ -2,13,-58,161,224,72,-991,2001,1414,2330,-7910,9592,70,22929,-51853,72169,
+ -2,11,-63,154,221,36,-1064,1962,1280,2006,-8209,9750,-998,21189,-53534,71420,
+ -2,10,-68,147,215,2,-1137,1919,1131,1692,-8491,9863,-2122,19478,-55178,70590,
+ -3,9,-73,139,208,-29,-1210,1870,970,1388,-8755,9935,-3300,17799,-56778,69679,
+ -3,8,-79,132,200,-57,-1283,1817,794,1095,-8998,9966,-4533,16155,-58333,68692,
+ -4,7,-85,125,189,-83,-1356,1759,605,814,-9219,9959,-5818,14548,-59838,67629,
+ -4,7,-91,117,177,-106,-1428,1698,402,545,-9416,9916,-7154,12980,-61289,66494,
+ -5,6,-97,111,163,-127,-1498,1634,185,288,-9585,9838,-8540,11455,-62684,65290
+ };
+ float *zlin = lins + 15*64;
+ const float *w = g_win;
+
+ zlin[4*15] = xl[18*16];
+ zlin[4*15 + 1] = xr[18*16];
+ zlin[4*15 + 2] = xl[0];
+ zlin[4*15 + 3] = xr[0];
+
+ zlin[4*31] = xl[1 + 18*16];
+ zlin[4*31 + 1] = xr[1 + 18*16];
+ zlin[4*31 + 2] = xl[1];
+ zlin[4*31 + 3] = xr[1];
+
+ mp3d_synth_pair(dstr, nch, lins + 4*15 + 1);
+ mp3d_synth_pair(dstr + 32*nch, nch, lins + 4*15 + 64 + 1);
+ mp3d_synth_pair(dstl, nch, lins + 4*15);
+ mp3d_synth_pair(dstl + 32*nch, nch, lins + 4*15 + 64);
+
+#if HAVE_SIMD
+ if (have_simd()) for (i = 14; i >= 0; i--)
+ {
+#define VLOAD(k) f4 w0 = VSET(*w++); f4 w1 = VSET(*w++); f4 vz = VLD(&zlin[4*i - 64*k]); f4 vy = VLD(&zlin[4*i - 64*(15 - k)]);
+#define V0(k) { VLOAD(k) b = VADD(VMUL(vz, w1), VMUL(vy, w0)) ; a = VSUB(VMUL(vz, w0), VMUL(vy, w1)); }
+#define V1(k) { VLOAD(k) b = VADD(b, VADD(VMUL(vz, w1), VMUL(vy, w0))); a = VADD(a, VSUB(VMUL(vz, w0), VMUL(vy, w1))); }
+#define V2(k) { VLOAD(k) b = VADD(b, VADD(VMUL(vz, w1), VMUL(vy, w0))); a = VADD(a, VSUB(VMUL(vy, w1), VMUL(vz, w0))); }
+ f4 a, b;
+ zlin[4*i] = xl[18*(31 - i)];
+ zlin[4*i + 1] = xr[18*(31 - i)];
+ zlin[4*i + 2] = xl[1 + 18*(31 - i)];
+ zlin[4*i + 3] = xr[1 + 18*(31 - i)];
+ zlin[4*i + 64] = xl[1 + 18*(1 + i)];
+ zlin[4*i + 64 + 1] = xr[1 + 18*(1 + i)];
+ zlin[4*i - 64 + 2] = xl[18*(1 + i)];
+ zlin[4*i - 64 + 3] = xr[18*(1 + i)];
+
+ V0(0) V2(1) V1(2) V2(3) V1(4) V2(5) V1(6) V2(7)
+
+ {
+#ifndef MINIMP3_FLOAT_OUTPUT
+#if HAVE_SSE
+ static const f4 g_max = { 32767.0f, 32767.0f, 32767.0f, 32767.0f };
+ static const f4 g_min = { -32768.0f, -32768.0f, -32768.0f, -32768.0f };
+ __m128i pcm8 = _mm_packs_epi32(_mm_cvtps_epi32(_mm_max_ps(_mm_min_ps(a, g_max), g_min)),
+ _mm_cvtps_epi32(_mm_max_ps(_mm_min_ps(b, g_max), g_min)));
+ dstr[(15 - i)*nch] = _mm_extract_epi16(pcm8, 1);
+ dstr[(17 + i)*nch] = _mm_extract_epi16(pcm8, 5);
+ dstl[(15 - i)*nch] = _mm_extract_epi16(pcm8, 0);
+ dstl[(17 + i)*nch] = _mm_extract_epi16(pcm8, 4);
+ dstr[(47 - i)*nch] = _mm_extract_epi16(pcm8, 3);
+ dstr[(49 + i)*nch] = _mm_extract_epi16(pcm8, 7);
+ dstl[(47 - i)*nch] = _mm_extract_epi16(pcm8, 2);
+ dstl[(49 + i)*nch] = _mm_extract_epi16(pcm8, 6);
+#else /* HAVE_SSE */
+ int16x4_t pcma, pcmb;
+ a = VADD(a, VSET(0.5f));
+ b = VADD(b, VSET(0.5f));
+ pcma = vqmovn_s32(vqaddq_s32(vcvtq_s32_f32(a), vreinterpretq_s32_u32(vcltq_f32(a, VSET(0)))));
+ pcmb = vqmovn_s32(vqaddq_s32(vcvtq_s32_f32(b), vreinterpretq_s32_u32(vcltq_f32(b, VSET(0)))));
+ vst1_lane_s16(dstr + (15 - i)*nch, pcma, 1);
+ vst1_lane_s16(dstr + (17 + i)*nch, pcmb, 1);
+ vst1_lane_s16(dstl + (15 - i)*nch, pcma, 0);
+ vst1_lane_s16(dstl + (17 + i)*nch, pcmb, 0);
+ vst1_lane_s16(dstr + (47 - i)*nch, pcma, 3);
+ vst1_lane_s16(dstr + (49 + i)*nch, pcmb, 3);
+ vst1_lane_s16(dstl + (47 - i)*nch, pcma, 2);
+ vst1_lane_s16(dstl + (49 + i)*nch, pcmb, 2);
+#endif /* HAVE_SSE */
+
+#else /* MINIMP3_FLOAT_OUTPUT */
+
+ static const f4 g_scale = { 1.0f/32768.0f, 1.0f/32768.0f, 1.0f/32768.0f, 1.0f/32768.0f };
+ a = VMUL(a, g_scale);
+ b = VMUL(b, g_scale);
+#if HAVE_SSE
+ _mm_store_ss(dstr + (15 - i)*nch, _mm_shuffle_ps(a, a, _MM_SHUFFLE(1, 1, 1, 1)));
+ _mm_store_ss(dstr + (17 + i)*nch, _mm_shuffle_ps(b, b, _MM_SHUFFLE(1, 1, 1, 1)));
+ _mm_store_ss(dstl + (15 - i)*nch, _mm_shuffle_ps(a, a, _MM_SHUFFLE(0, 0, 0, 0)));
+ _mm_store_ss(dstl + (17 + i)*nch, _mm_shuffle_ps(b, b, _MM_SHUFFLE(0, 0, 0, 0)));
+ _mm_store_ss(dstr + (47 - i)*nch, _mm_shuffle_ps(a, a, _MM_SHUFFLE(3, 3, 3, 3)));
+ _mm_store_ss(dstr + (49 + i)*nch, _mm_shuffle_ps(b, b, _MM_SHUFFLE(3, 3, 3, 3)));
+ _mm_store_ss(dstl + (47 - i)*nch, _mm_shuffle_ps(a, a, _MM_SHUFFLE(2, 2, 2, 2)));
+ _mm_store_ss(dstl + (49 + i)*nch, _mm_shuffle_ps(b, b, _MM_SHUFFLE(2, 2, 2, 2)));
+#else /* HAVE_SSE */
+ vst1q_lane_f32(dstr + (15 - i)*nch, a, 1);
+ vst1q_lane_f32(dstr + (17 + i)*nch, b, 1);
+ vst1q_lane_f32(dstl + (15 - i)*nch, a, 0);
+ vst1q_lane_f32(dstl + (17 + i)*nch, b, 0);
+ vst1q_lane_f32(dstr + (47 - i)*nch, a, 3);
+ vst1q_lane_f32(dstr + (49 + i)*nch, b, 3);
+ vst1q_lane_f32(dstl + (47 - i)*nch, a, 2);
+ vst1q_lane_f32(dstl + (49 + i)*nch, b, 2);
+#endif /* HAVE_SSE */
+#endif /* MINIMP3_FLOAT_OUTPUT */
+ }
+ } else
+#endif /* HAVE_SIMD */
+#ifdef MINIMP3_ONLY_SIMD
+ {} /* for HAVE_SIMD=1, MINIMP3_ONLY_SIMD=1 case we do not need non-intrinsic "else" branch */
+#else /* MINIMP3_ONLY_SIMD */
+ for (i = 14; i >= 0; i--)
+ {
+#define LOAD(k) float w0 = *w++; float w1 = *w++; float *vz = &zlin[4*i - k*64]; float *vy = &zlin[4*i - (15 - k)*64];
+#define S0(k) { int j; LOAD(k); for (j = 0; j < 4; j++) b[j] = vz[j]*w1 + vy[j]*w0, a[j] = vz[j]*w0 - vy[j]*w1; }
+#define S1(k) { int j; LOAD(k); for (j = 0; j < 4; j++) b[j] += vz[j]*w1 + vy[j]*w0, a[j] += vz[j]*w0 - vy[j]*w1; }
+#define S2(k) { int j; LOAD(k); for (j = 0; j < 4; j++) b[j] += vz[j]*w1 + vy[j]*w0, a[j] += vy[j]*w1 - vz[j]*w0; }
+ float a[4], b[4];
+
+ zlin[4*i] = xl[18*(31 - i)];
+ zlin[4*i + 1] = xr[18*(31 - i)];
+ zlin[4*i + 2] = xl[1 + 18*(31 - i)];
+ zlin[4*i + 3] = xr[1 + 18*(31 - i)];
+ zlin[4*(i + 16)] = xl[1 + 18*(1 + i)];
+ zlin[4*(i + 16) + 1] = xr[1 + 18*(1 + i)];
+ zlin[4*(i - 16) + 2] = xl[18*(1 + i)];
+ zlin[4*(i - 16) + 3] = xr[18*(1 + i)];
+
+ S0(0) S2(1) S1(2) S2(3) S1(4) S2(5) S1(6) S2(7)
+
+ dstr[(15 - i)*nch] = mp3d_scale_pcm(a[1]);
+ dstr[(17 + i)*nch] = mp3d_scale_pcm(b[1]);
+ dstl[(15 - i)*nch] = mp3d_scale_pcm(a[0]);
+ dstl[(17 + i)*nch] = mp3d_scale_pcm(b[0]);
+ dstr[(47 - i)*nch] = mp3d_scale_pcm(a[3]);
+ dstr[(49 + i)*nch] = mp3d_scale_pcm(b[3]);
+ dstl[(47 - i)*nch] = mp3d_scale_pcm(a[2]);
+ dstl[(49 + i)*nch] = mp3d_scale_pcm(b[2]);
+ }
+#endif /* MINIMP3_ONLY_SIMD */
+}
+
+static void mp3d_synth_granule(float *qmf_state, float *grbuf, int nbands, int nch, mp3d_sample_t *pcm, float *lins)
+{
+ int i;
+ for (i = 0; i < nch; i++)
+ {
+ mp3d_DCT_II(grbuf + 576*i, nbands);
+ }
+
+ memcpy(lins, qmf_state, sizeof(float)*15*64);
+
+ for (i = 0; i < nbands; i += 2)
+ {
+ mp3d_synth(grbuf + i, pcm + 32*nch*i, nch, lins + i*64);
+ }
+#ifndef MINIMP3_NONSTANDARD_BUT_LOGICAL
+ if (nch == 1)
+ {
+ for (i = 0; i < 15*64; i += 2)
+ {
+ qmf_state[i] = lins[nbands*64 + i];
+ }
+ } else
+#endif /* MINIMP3_NONSTANDARD_BUT_LOGICAL */
+ {
+ memcpy(qmf_state, lins + nbands*64, sizeof(float)*15*64);
+ }
+}
+
+static int mp3d_match_frame(const uint8_t *hdr, int mp3_bytes, int frame_bytes)
+{
+ int i, nmatch;
+ for (i = 0, nmatch = 0; nmatch < MAX_FRAME_SYNC_MATCHES; nmatch++)
+ {
+ i += hdr_frame_bytes(hdr + i, frame_bytes) + hdr_padding(hdr + i);
+ if (i + HDR_SIZE > mp3_bytes)
+ return nmatch > 0;
+ if (!hdr_compare(hdr, hdr + i))
+ return 0;
+ }
+ return 1;
+}
+
+static int mp3d_find_frame(const uint8_t *mp3, int mp3_bytes, int *free_format_bytes, int *ptr_frame_bytes)
+{
+ int i, k;
+ for (i = 0; i < mp3_bytes - HDR_SIZE; i++, mp3++)
+ {
+ if (hdr_valid(mp3))
+ {
+ int frame_bytes = hdr_frame_bytes(mp3, *free_format_bytes);
+ int frame_and_padding = frame_bytes + hdr_padding(mp3);
+
+ for (k = HDR_SIZE; !frame_bytes && k < MAX_FREE_FORMAT_FRAME_SIZE && i + 2*k < mp3_bytes - HDR_SIZE; k++)
+ {
+ if (hdr_compare(mp3, mp3 + k))
+ {
+ int fb = k - hdr_padding(mp3);
+ int nextfb = fb + hdr_padding(mp3 + k);
+ if (i + k + nextfb + HDR_SIZE > mp3_bytes || !hdr_compare(mp3, mp3 + k + nextfb))
+ continue;
+ frame_and_padding = k;
+ frame_bytes = fb;
+ *free_format_bytes = fb;
+ }
+ }
+ if ((frame_bytes && i + frame_and_padding <= mp3_bytes &&
+ mp3d_match_frame(mp3, mp3_bytes - i, frame_bytes)) ||
+ (!i && frame_and_padding == mp3_bytes))
+ {
+ *ptr_frame_bytes = frame_and_padding;
+ return i;
+ }
+ *free_format_bytes = 0;
+ }
+ }
+ *ptr_frame_bytes = 0;
+ return mp3_bytes;
+}
+
+void mp3dec_init(mp3dec_t *dec)
+{
+ dec->header[0] = 0;
+}
+
+int mp3dec_decode_frame(mp3dec_t *dec, const uint8_t *mp3, int mp3_bytes, mp3d_sample_t *pcm, mp3dec_frame_info_t *info)
+{
+ int i = 0, igr, frame_size = 0, success = 1;
+ const uint8_t *hdr;
+ bs_t bs_frame[1];
+ mp3dec_scratch_t scratch;
+
+ if (mp3_bytes > 4 && dec->header[0] == 0xff && hdr_compare(dec->header, mp3))
+ {
+ frame_size = hdr_frame_bytes(mp3, dec->free_format_bytes) + hdr_padding(mp3);
+ if (frame_size != mp3_bytes && (frame_size + HDR_SIZE > mp3_bytes || !hdr_compare(mp3, mp3 + frame_size)))
+ {
+ frame_size = 0;
+ }
+ }
+ if (!frame_size)
+ {
+ memset(dec, 0, sizeof(mp3dec_t));
+ i = mp3d_find_frame(mp3, mp3_bytes, &dec->free_format_bytes, &frame_size);
+ if (!frame_size || i + frame_size > mp3_bytes)
+ {
+ info->frame_bytes = i;
+ return 0;
+ }
+ }
+
+ hdr = mp3 + i;
+ memcpy(dec->header, hdr, HDR_SIZE);
+ info->frame_bytes = i + frame_size;
+ info->frame_offset = i;
+ info->channels = HDR_IS_MONO(hdr) ? 1 : 2;
+ info->hz = hdr_sample_rate_hz(hdr);
+ info->layer = 4 - HDR_GET_LAYER(hdr);
+ info->bitrate_kbps = hdr_bitrate_kbps(hdr);
+
+ if (!pcm)
+ {
+ return hdr_frame_samples(hdr);
+ }
+
+ bs_init(bs_frame, hdr + HDR_SIZE, frame_size - HDR_SIZE);
+ if (HDR_IS_CRC(hdr))
+ {
+ get_bits(bs_frame, 16);
+ }
+
+ if (info->layer == 3)
+ {
+ int main_data_begin = L3_read_side_info(bs_frame, scratch.gr_info, hdr);
+ if (main_data_begin < 0 || bs_frame->pos > bs_frame->limit)
+ {
+ mp3dec_init(dec);
+ return 0;
+ }
+ success = L3_restore_reservoir(dec, bs_frame, &scratch, main_data_begin);
+ if (success)
+ {
+ for (igr = 0; igr < (HDR_TEST_MPEG1(hdr) ? 2 : 1); igr++, pcm += 576*info->channels)
+ {
+ memset(scratch.grbuf[0], 0, 576*2*sizeof(float));
+ L3_decode(dec, &scratch, scratch.gr_info + igr*info->channels, info->channels);
+ mp3d_synth_granule(dec->qmf_state, scratch.grbuf[0], 18, info->channels, pcm, scratch.syn[0]);
+ }
+ }
+ L3_save_reservoir(dec, &scratch);
+ } else
+ {
+#ifdef MINIMP3_ONLY_MP3
+ return 0;
+#else /* MINIMP3_ONLY_MP3 */
+ L12_scale_info sci[1];
+ L12_read_scale_info(hdr, bs_frame, sci);
+
+ memset(scratch.grbuf[0], 0, 576*2*sizeof(float));
+ for (i = 0, igr = 0; igr < 3; igr++)
+ {
+ if (12 == (i += L12_dequantize_granule(scratch.grbuf[0] + i, bs_frame, sci, info->layer | 1)))
+ {
+ i = 0;
+ L12_apply_scf_384(sci, sci->scf + igr, scratch.grbuf[0]);
+ mp3d_synth_granule(dec->qmf_state, scratch.grbuf[0], 12, info->channels, pcm, scratch.syn[0]);
+ memset(scratch.grbuf[0], 0, 576*2*sizeof(float));
+ pcm += 384*info->channels;
+ }
+ if (bs_frame->pos > bs_frame->limit)
+ {
+ mp3dec_init(dec);
+ return 0;
+ }
+ }
+#endif /* MINIMP3_ONLY_MP3 */
+ }
+ return success*hdr_frame_samples(dec->header);
+}
+
+#ifdef MINIMP3_FLOAT_OUTPUT
+void mp3dec_f32_to_s16(const float *in, int16_t *out, int num_samples)
+{
+ int i = 0;
+#if HAVE_SIMD
+ int aligned_count = num_samples & ~7;
+ for(; i < aligned_count; i += 8)
+ {
+ static const f4 g_scale = { 32768.0f, 32768.0f, 32768.0f, 32768.0f };
+ f4 a = VMUL(VLD(&in[i ]), g_scale);
+ f4 b = VMUL(VLD(&in[i+4]), g_scale);
+#if HAVE_SSE
+ static const f4 g_max = { 32767.0f, 32767.0f, 32767.0f, 32767.0f };
+ static const f4 g_min = { -32768.0f, -32768.0f, -32768.0f, -32768.0f };
+ __m128i pcm8 = _mm_packs_epi32(_mm_cvtps_epi32(_mm_max_ps(_mm_min_ps(a, g_max), g_min)),
+ _mm_cvtps_epi32(_mm_max_ps(_mm_min_ps(b, g_max), g_min)));
+ out[i ] = _mm_extract_epi16(pcm8, 0);
+ out[i+1] = _mm_extract_epi16(pcm8, 1);
+ out[i+2] = _mm_extract_epi16(pcm8, 2);
+ out[i+3] = _mm_extract_epi16(pcm8, 3);
+ out[i+4] = _mm_extract_epi16(pcm8, 4);
+ out[i+5] = _mm_extract_epi16(pcm8, 5);
+ out[i+6] = _mm_extract_epi16(pcm8, 6);
+ out[i+7] = _mm_extract_epi16(pcm8, 7);
+#else /* HAVE_SSE */
+ int16x4_t pcma, pcmb;
+ a = VADD(a, VSET(0.5f));
+ b = VADD(b, VSET(0.5f));
+ pcma = vqmovn_s32(vqaddq_s32(vcvtq_s32_f32(a), vreinterpretq_s32_u32(vcltq_f32(a, VSET(0)))));
+ pcmb = vqmovn_s32(vqaddq_s32(vcvtq_s32_f32(b), vreinterpretq_s32_u32(vcltq_f32(b, VSET(0)))));
+ vst1_lane_s16(out+i , pcma, 0);
+ vst1_lane_s16(out+i+1, pcma, 1);
+ vst1_lane_s16(out+i+2, pcma, 2);
+ vst1_lane_s16(out+i+3, pcma, 3);
+ vst1_lane_s16(out+i+4, pcmb, 0);
+ vst1_lane_s16(out+i+5, pcmb, 1);
+ vst1_lane_s16(out+i+6, pcmb, 2);
+ vst1_lane_s16(out+i+7, pcmb, 3);
+#endif /* HAVE_SSE */
+ }
+#endif /* HAVE_SIMD */
+ for(; i < num_samples; i++)
+ {
+ float sample = in[i] * 32768.0f;
+ if (sample >= 32766.5)
+ out[i] = (int16_t) 32767;
+ else if (sample <= -32767.5)
+ out[i] = (int16_t)-32768;
+ else
+ {
+ int16_t s = (int16_t)(sample + .5f);
+ s -= (s < 0); /* away from zero, to be compliant */
+ out[i] = s;
+ }
+ }
+}
+#endif /* MINIMP3_FLOAT_OUTPUT */
+#endif /* MINIMP3_IMPLEMENTATION && !_MINIMP3_IMPLEMENTATION_GUARD */
diff --git a/src/SoundFormats/mp3.cpp b/src/SoundFormats/mp3.cpp
new file mode 100644
index 0000000..fcef95b
--- /dev/null
+++ b/src/SoundFormats/mp3.cpp
@@ -0,0 +1,80 @@
+#ifndef POMME_NO_MP3
+
+#include "PommeSound.h"
+
+#include
+#include
+
+#define MINIMP3_IMPLEMENTATION
+#include "SoundFormats/minimp3.h"
+
+#define MINIMP3_IO_SIZE (128*1024) // io buffer size for streaming functions, must be greater than MINIMP3_BUF_SIZE
+#define MINIMP3_BUF_SIZE (16*1024) // buffer which can hold minimum 10 consecutive mp3 frames (~16KB) worst case
+
+SndListHandle Pomme::Sound::LoadMP3AsResource(std::istream& stream)
+{
+ mp3dec_t context = {};
+ mp3dec_init(&context);
+
+ mp3dec_frame_info_t frameInfo;
+
+ std::vector fileBuf;
+ std::vector songPCM;
+ std::vector tempPCM(MINIMP3_MAX_SAMPLES_PER_FRAME);
+
+ int totalSamples = 0;
+
+ while (true)
+ {
+ // Refill buffer
+ if (fileBuf.size() < MINIMP3_BUF_SIZE)
+ {
+ if (stream.eof())
+ {
+ // bail
+ break;
+ }
+
+ auto oldSize = fileBuf.size();
+ auto toRead = MINIMP3_BUF_SIZE - oldSize;
+
+ fileBuf.resize(MINIMP3_BUF_SIZE);
+ stream.read((char*) (fileBuf.data() + oldSize), (int) toRead);
+
+ auto didRead = stream.gcount();
+ fileBuf.resize(oldSize + didRead);
+ }
+
+ int numDecodedSamples = mp3dec_decode_frame(&context, fileBuf.data(), (int) fileBuf.size(), tempPCM.data(), &frameInfo);
+
+ if (numDecodedSamples > 0)
+ {
+ size_t minCapacity = songPCM.size() + (numDecodedSamples * frameInfo.channels);
+ if (songPCM.capacity() < minCapacity)
+ {
+ songPCM.reserve(2 * minCapacity);
+ }
+
+ songPCM.insert(songPCM.end(), tempPCM.begin(), tempPCM.begin() + (numDecodedSamples * frameInfo.channels));
+ totalSamples += numDecodedSamples;
+ }
+
+ fileBuf.erase(fileBuf.begin(), fileBuf.begin() + frameInfo.frame_bytes);
+ }
+
+ Pomme::Sound::SampledSoundInfo info = {};
+ info.compressionType = 'swot';
+ info.bigEndian = false;
+ info.isCompressed = false;
+ info.baseNote = 60; // Middle C
+ info.codecBitDepth = 8 * sizeof(mp3d_sample_t);
+ info.sampleRate = frameInfo.hz;
+ info.nChannels = frameInfo.channels;
+ info.nPackets = totalSamples;
+ info.decompressedLength = totalSamples * info.nChannels * sizeof(mp3d_sample_t);
+ info.compressedLength = info.decompressedLength;
+ info.dataStart = (char*) songPCM.data();
+ return info.MakeStandaloneResource();
+}
+
+#endif // POMME_NO_MP3
diff --git a/src/SoundMixer/SoundManager.cpp b/src/SoundMixer/SoundManager.cpp
index f06b9ed..87d00f1 100644
--- a/src/SoundMixer/SoundManager.cpp
+++ b/src/SoundMixer/SoundManager.cpp
@@ -334,6 +334,12 @@ OSErr SndStartFilePlay(
}
SndListHandle sndListHandle = Pomme_SndLoadFileAsResource(fRefNum);
+
+ if (!sndListHandle)
+ {
+ return badFileFormat;
+ }
+
long offset = 0;
GetSoundHeaderOffset(sndListHandle, &offset);
InstallSoundInChannel(chan, ((Ptr) *sndListHandle) + offset, true);