curioustorvald/tsvm
1
0
Fork 0
mirror of https://github.com/curioustorvald/tsvm.git synced 2026-03-09 20:51:51 +09:00
Code Issues Packages Projects Releases Wiki Activity

TAD: better bit allocation using statistics

Browse Source
This commit is contained in:
minjaesong
2025-10-26 18:16:28 +09:00
parent 9fcb7fc95c
commit 370d511f44
5 changed files with 167 additions and 177 deletions
Download Patch File Download Diff File Expand all files Collapse all files

98
video_encoder/decoder_tad.c
View File

@@ -256,8 +256,8 @@ static void ms_correlate(const float *mid, const float *side, float *left, float
// Decode M/S → L/R
float m = mid[i];
float s = side[i];
left[i] = FCLAMP((m + s) * 1.7321f, -1.0f, 1.0f);
right[i] = FCLAMP((m - s) * 1.7321f, -1.0f, 1.0f);
left[i] = FCLAMP((m + s), -1.0f, 1.0f);
right[i] = FCLAMP((m - s), -1.0f, 1.0f);
}
}
@@ -320,38 +320,44 @@ static void pcm32f_to_pcm8(const float *fleft, const float *fright, uint8_t *lef
// Dequantization (inverse of quantization)
//=============================================================================
static void get_quantization_weights(int quality, int dwt_levels, float *weights) {
const float base_weights[16][16] = {
/* 0*/{1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f},
/* 1*/{1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f},
/* 2*/{1.0f, 1.0f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 3*/{0.2f, 1.0f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 4*/{0.2f, 0.8f, 1.0f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 5*/{0.2f, 0.8f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 6*/{0.2f, 0.2f, 0.8f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 7*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 8*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/* 9*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/*10*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/*11*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/*12*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f, 1.5f},
/*13*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f, 1.5f},
/*14*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f, 1.5f},
/*15*/{0.2f, 0.2f, 0.8f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.25f, 1.5f, 1.5f}
};
float quality_scale = 1.0f * (1.0f + FCLAMP((5 - quality) * 0.5f, 0.0f, 1000.0f));
#define LAMBDA_FIXED 5.8f
for (int i = 0; i < dwt_levels; i++) {
weights[i] = 1.0f;//base_weights[dwt_levels][i] * quality_scale;
// Lambda-based decompanding decoder (inverse of Laplacian CDF-based encoder)
// Converts quantized index back to normalized float in [-1, 1]
static float lambda_decompanding(int16_t quant_val, int quant_bits) {
// Handle zero
if (quant_val == 0) {
return 0.0f;
}
int sign = (quant_val < 0) ? -1 : 1;
int abs_index = abs(quant_val);
// Maximum index for the given quant_bits
int max_index = (1 << (quant_bits - 1)) - 1;
// Clamp to valid range
if (abs_index > max_index) abs_index = max_index;
// Map index back to normalized CDF [0, 1]
float normalized_cdf = (float)abs_index / max_index;
// Map from [0, 1] back to [0.5, 1.0] (CDF range for positive half)
float cdf = 0.5f + normalized_cdf * 0.5f;
// Inverse Laplacian CDF for x >= 0: x = -(1/λ) * ln(2*(1-F))
// For F in [0.5, 1.0]: x = -(1/λ) * ln(2*(1-F))
float abs_val = -(1.0f / LAMBDA_FIXED) * logf(2.0f * (1.0f - cdf));
// Clamp to [0, 1]
if (abs_val > 1.0f) abs_val = 1.0f;
if (abs_val < 0.0f) abs_val = 0.0f;
return sign * abs_val;
}
#define QUANT_STEPS 8.0f // 64 -> [-64..64] -> 7 bits for LL
static void dequantize_dwt_coefficients(const int16_t *quantized, float *coeffs, size_t count, int quality, int chunk_size, int dwt_levels) {
float weights[16];
get_quantization_weights(quality, dwt_levels, weights);
static void dequantize_dwt_coefficients(const int16_t *quantized, float *coeffs, size_t count, int chunk_size, int dwt_levels, int quant_bits) {
// Calculate sideband boundaries dynamically
int first_band_size = chunk_size >> dwt_levels;
@@ -372,12 +378,11 @@ static void dequantize_dwt_coefficients(const int16_t *quantized, float *coeffs,
}
}
// Map (dwt_levels+1) sidebands to dwt_levels weights
int weight_idx = (sideband == 0) ? 0 : sideband - 1;
if (weight_idx >= dwt_levels) weight_idx = dwt_levels - 1;
// Decode using lambda companding
float normalized_val = lambda_decompanding(quantized[i], quant_bits);
float weight = weights[weight_idx];
coeffs[i] = ((float)quantized[i] * TAD32_COEFF_SCALARS[sideband]) / (QUANT_STEPS * weight);
// Denormalize using the subband scalar
coeffs[i] = normalized_val * TAD32_COEFF_SCALARS[sideband];
}
free(sideband_starts);
@@ -423,12 +428,16 @@ static size_t decode_sigmap_2bit(const uint8_t *input, int16_t *values, size_t c
//=============================================================================
static int decode_chunk(const uint8_t *input, size_t input_size, uint8_t *pcmu8_stereo,
int quality, size_t *bytes_consumed, size_t *samples_decoded) {
size_t *bytes_consumed, size_t *samples_decoded) {
const uint8_t *read_ptr = input;
// Read chunk header
uint16_t sample_count = *((const uint16_t*)read_ptr);
read_ptr += sizeof(uint16_t);
uint8_t quant_bits = *read_ptr;
read_ptr += sizeof(uint8_t);
uint32_t payload_size = *((const uint32_t*)read_ptr);
read_ptr += sizeof(uint32_t);
@@ -479,8 +488,8 @@ static int decode_chunk(const uint8_t *input, size_t input_size, uint8_t *pcmu8_
side_bytes = decode_sigmap_2bit(payload_ptr + mid_bytes, quant_side, sample_count);
// Dequantize
dequantize_dwt_coefficients(quant_mid, dwt_mid, sample_count, quality, sample_count, dwt_levels);
dequantize_dwt_coefficients(quant_side, dwt_side, sample_count, quality, sample_count, dwt_levels);
dequantize_dwt_coefficients(quant_mid, dwt_mid, sample_count, sample_count, dwt_levels, quant_bits);
dequantize_dwt_coefficients(quant_side, dwt_side, sample_count, sample_count, dwt_levels, quant_bits);
// Inverse DWT
dwt_haar_inverse_multilevel(dwt_mid, sample_count, dwt_levels);
@@ -520,7 +529,6 @@ static void print_usage(const char *prog_name) {
printf("Options:\n");
printf(" -i <file> Input TAD file\n");
printf(" -o <file> Output PCMu8 file (raw 8-bit unsigned stereo @ 32kHz)\n");
printf(" -q <0-5> Quality level used during encoding (default: 3)\n");
printf(" -v Verbose output\n");
printf(" -h, --help Show this help\n");
printf("\nVersion: %s\n", DECODER_VENDOR_STRING);
@@ -531,11 +539,10 @@ static void print_usage(const char *prog_name) {
int main(int argc, char *argv[]) {
char *input_file = NULL;
char *output_file = NULL;
int quality = 3; // Must match encoder quality
int verbose = 0;
int opt;
while ((opt = getopt(argc, argv, "i:o:q:vh")) != -1) {
while ((opt = getopt(argc, argv, "i:o:vh")) != -1) {
switch (opt) {
case 'i':
input_file = optarg;
@@ -543,14 +550,6 @@ int main(int argc, char *argv[]) {
case 'o':
output_file = optarg;
break;
case 'q':
quality = atoi(optarg);
if (quality < TAD_QUALITY_MIN || quality > TAD_QUALITY_MAX) {
fprintf(stderr, "Error: Quality must be between %d and %d\n",
TAD_QUALITY_MIN, TAD_QUALITY_MAX);
return 1;
}
break;
case 'v':
verbose = 1;
break;
@@ -573,7 +572,6 @@ int main(int argc, char *argv[]) {
printf("%s\n", DECODER_VENDOR_STRING);
printf("Input: %s\n", input_file);
printf("Output: %s\n", output_file);
printf("Quality: %d\n", quality);
}
// Open input file
@@ -611,7 +609,7 @@ int main(int argc, char *argv[]) {
while (offset < input_size) {
size_t bytes_consumed, samples_decoded;
int result = decode_chunk(input_data + offset, input_size - offset,
chunk_output, quality, &bytes_consumed, &samples_decoded);
chunk_output, &bytes_consumed, &samples_decoded);
if (result != 0) {
fprintf(stderr, "Error: Chunk decoding failed at offset %zu\n", offset);