} G729FormatDescription;
typedef struct {
+ int pitch_delay_int_prev; ///< integer part of previous subframe's pitch delay (4.1.3)
+
+ /// (2.13) LSP quantizer outputs
+ int16_t past_quantizer_output_buf[MA_NP + 1][10];
+ int16_t* past_quantizer_outputs[MA_NP + 1];
+
+ int16_t lsfq[10]; ///< (2.13) quantized LSF coefficients from previous frame
int16_t lsp_buf[2][10]; ///< (0.15) LSP coefficients (previous and current frames) (3.2.5)
int16_t *lsp[2]; ///< pointers to lsp_buf
} G729Context;
return (0x6996966996696996ULL >> (value >> 2)) & 1;
}
+static void lsf_decode(int16_t* lsfq, int16_t* past_quantizer_outputs[MA_NP + 1],
+ int16_t ma_predictor,
+ int16_t vq_1st, int16_t vq_2nd_low, int16_t vq_2nd_high)
+{
+ int i,j;
+ static const uint8_t min_distance[2]={10, 5}; //(2.13)
+ int16_t* quantizer_output = past_quantizer_outputs[MA_NP];
+
+ for (i = 0; i < 5; i++) {
+ quantizer_output[i] = cb_lsp_1st[vq_1st][i ] + cb_lsp_2nd[vq_2nd_low ][i ];
+ quantizer_output[i + 5] = cb_lsp_1st[vq_1st][i + 5] + cb_lsp_2nd[vq_2nd_high][i + 5];
+ }
+
+ for (j = 0; j < 2; j++) {
+ for (i = 1; i < 10; i++) {
+ int diff = (quantizer_output[i - 1] - quantizer_output[i] + min_distance[j]) >> 1;
+ if (diff > 0) {
+ quantizer_output[i - 1] -= diff;
+ quantizer_output[i ] += diff;
+ }
+ }
+ }
+
+ for (i = 0; i < 10; i++) {
+ int sum = quantizer_output[i] * cb_ma_predictor_sum[ma_predictor][i];
+ for (j = 0; j < MA_NP; j++)
+ sum += past_quantizer_outputs[j][i] * cb_ma_predictor[ma_predictor][j][i];
+
+ lsfq[i] = sum >> 15;
+ }
+
+ /* Rotate past_quantizer_outputs. */
+ memmove(past_quantizer_outputs + 1, past_quantizer_outputs, MA_NP * sizeof(int16_t*));
+ past_quantizer_outputs[0] = quantizer_output;
+
+ ff_acelp_reorder_lsf(lsfq, LSFQ_DIFF_MIN, LSFQ_MIN, LSFQ_MAX, 10);
+}
+
static av_cold int decoder_init(AVCodecContext * avctx)
{
+ G729Context* ctx = avctx->priv_data;
+ int i,k;
+
if (avctx->channels != 1) {
av_log(avctx, AV_LOG_ERROR, "Only mono sound is supported (requested channels: %d).\n", avctx->channels);
return AVERROR_NOFMT;
/* Both 8kbit/s and 6.4kbit/s modes uses two subframes per frame. */
avctx->frame_size = SUBFRAME_SIZE << 1;
+ for (k = 0; k < MA_NP + 1; k++) {
+ ctx->past_quantizer_outputs[k] = ctx->past_quantizer_output_buf[k];
+ for (i = 1; i < 11; i++)
+ ctx->past_quantizer_outputs[k][i - 1] = (18717 * i) >> 3;
+ }
+
ctx->lsp[0] = ctx->lsp_buf[0];
ctx->lsp[1] = ctx->lsp_buf[1];
memcpy(ctx->lsp[0], lsp_init, 10 * sizeof(int16_t));
uint8_t quantizer_2nd_lo; ///< second stage lower vector of quantizer (size in bits)
uint8_t quantizer_2nd_hi; ///< second stage higher vector of quantizer (size in bits)
+ int pitch_delay_int; // pitch delay, integer part
+ int pitch_delay_3x; // pitch delay, multiplied by 3
+
if (*data_size < SUBFRAME_SIZE << 2) {
av_log(avctx, AV_LOG_ERROR, "Error processing packet: output buffer too small\n");
return AVERROR(EIO);
quantizer_2nd_lo = get_bits(&gb, VQ_2ND_BITS);
quantizer_2nd_hi = get_bits(&gb, VQ_2ND_BITS);
+ lsf_decode(ctx->lsfq, ctx->past_quantizer_outputs,
+ ma_predictor,
+ quantizer_1st, quantizer_2nd_lo, quantizer_2nd_hi);
+
ff_acelp_lsf2lsp(ctx->lsp[1], ctx->lsfq, 10);
ff_acelp_lp_decode(&lp[0][0], &lp[1][0], ctx->lsp[1], ctx->lsp[0], 10);
gc_1st_index = get_bits(&gb, format.gc_1st_index_bits);
gc_2nd_index = get_bits(&gb, format.gc_2nd_index_bits);
- ff_acelp_weighted_vector_sum(fc + pitch_delay_int[i],
- fc + pitch_delay_int[i],
+ if(!i) {
+ if (bad_pitch)
+ pitch_delay_3x = 3 * ctx->pitch_delay_int_prev;
+ else
+ pitch_delay_3x = ff_acelp_decode_8bit_to_1st_delay3(ac_index);
+ } else {
+ int pitch_delay_min = av_clip(ctx->pitch_delay_int_prev - 5,
+ PITCH_DELAY_MIN, PITCH_DELAY_MAX - 9);
+
+ if(packet_type == FORMAT_G729D_6K4)
+ pitch_delay_3x = ff_acelp_decode_4bit_to_2nd_delay3(ac_index, pitch_delay_min);
+ else
+ pitch_delay_3x = ff_acelp_decode_5_6_bit_to_2nd_delay3(ac_index, pitch_delay_min);
+ }
+
+ /* Round pitch delay to nearest (used everywhere except ff_acelp_interpolate). */
+ pitch_delay_int = (pitch_delay_3x + 1) / 3;
+
+ ff_acelp_weighted_vector_sum(fc + pitch_delay_int,
+ fc + pitch_delay_int,
fc, 1 << 14,
av_clip(ctx->gain_pitch, SHARP_MIN, SHARP_MAX),
0, 14,
- SUBFRAME_SIZE - pitch_delay_int[i]);
+ SUBFRAME_SIZE - pitch_delay_int);
if (frame_erasure) {
ctx->gain_pitch = (29491 * ctx->gain_pitch) >> 15; // 0.90 (0.15)
(!voicing && frame_erasure) ? 0 : ctx->gain_pitch,
( voicing && frame_erasure) ? 0 : ctx->gain_code,
1 << 13, 14, SUBFRAME_SIZE);
+
+ ctx->pitch_delay_int_prev = pitch_delay_int;
}
*data_size = SUBFRAME_SIZE << 2;