You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

238 lines
6.8KB

  1. /*
  2. * RealAudio 2.0 (28.8K)
  3. * Copyright (c) 2003 the ffmpeg project
  4. *
  5. * This file is part of FFmpeg.
  6. *
  7. * FFmpeg is free software; you can redistribute it and/or
  8. * modify it under the terms of the GNU Lesser General Public
  9. * License as published by the Free Software Foundation; either
  10. * version 2.1 of the License, or (at your option) any later version.
  11. *
  12. * FFmpeg is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  15. * Lesser General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU Lesser General Public
  18. * License along with FFmpeg; if not, write to the Free Software
  19. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20. */
  21. #include "avcodec.h"
  22. #define ALT_BITSTREAM_READER_LE
  23. #include "bitstream.h"
  24. #include "ra288.h"
  25. #include "lpc.h"
  26. typedef struct {
  27. float sp_lpc[36]; ///< LPC coefficients for speech data (spec: A)
  28. float gain_lpc[10]; ///< LPC coefficients for gain (spec: GB)
  29. /** speech data history (spec: SB).
  30. * Its first 70 coefficients are updated only at backward filtering.
  31. */
  32. float sp_hist[111];
  33. /// speech part of the gain autocorrelation (spec: REXP)
  34. float sp_rec[37];
  35. /** log-gain history (spec: SBLG).
  36. * Its first 28 coefficients are updated only at backward filtering.
  37. */
  38. float gain_hist[38];
  39. /// recursive part of the gain autocorrelation (spec: REXPLG)
  40. float gain_rec[11];
  41. } RA288Context;
  42. static av_cold int ra288_decode_init(AVCodecContext *avctx)
  43. {
  44. avctx->sample_fmt = SAMPLE_FMT_S16;
  45. return 0;
  46. }
  47. static inline float scalar_product_float(const float * v1, const float * v2,
  48. int size)
  49. {
  50. float res = 0.;
  51. while (size--)
  52. res += *v1++ * *v2++;
  53. return res;
  54. }
  55. static void apply_window(float *tgt, const float *m1, const float *m2, int n)
  56. {
  57. while (n--)
  58. *tgt++ = *m1++ * *m2++;
  59. }
  60. static void decode(RA288Context *ractx, float gain, int cb_coef)
  61. {
  62. int i, j;
  63. double sumsum;
  64. float sum, buffer[5];
  65. float *block = ractx->sp_hist + 70 + 36; // current block
  66. float *gain_block = ractx->gain_hist + 28;
  67. memmove(ractx->sp_hist + 70, ractx->sp_hist + 75, 36*sizeof(*block));
  68. for (i=0; i < 5; i++) {
  69. block[i] = 0.;
  70. for (j=0; j < 36; j++)
  71. block[i] -= block[i-1-j]*ractx->sp_lpc[j];
  72. }
  73. /* block 46 of G.728 spec */
  74. sum = 32.;
  75. for (i=0; i < 10; i++)
  76. sum -= gain_block[9-i] * ractx->gain_lpc[i];
  77. /* block 47 of G.728 spec */
  78. sum = av_clipf(sum, 0, 60);
  79. /* block 48 of G.728 spec */
  80. sumsum = exp(sum * 0.1151292546497) * gain; /* pow(10.0,sum/20)*gain */
  81. for (i=0; i < 5; i++)
  82. buffer[i] = codetable[cb_coef][i] * sumsum * (1./2048.);
  83. sum = scalar_product_float(buffer, buffer, 5) / 5;
  84. sum = FFMAX(sum, 1);
  85. /* shift and store */
  86. memmove(gain_block, gain_block + 1, 9 * sizeof(*gain_block));
  87. gain_block[9] = 10 * log10(sum) - 32;
  88. for (i=1; i < 5; i++)
  89. for (j=i-1; j >= 0; j--)
  90. buffer[i] -= ractx->sp_lpc[i-j-1] * buffer[j];
  91. /* output */
  92. for (i=0; i < 5; i++)
  93. block[i] = av_clipf(block[i] + buffer[i], -4095, 4095);
  94. }
  95. static void convolve(float *tgt, const float *src, int len, int n)
  96. {
  97. for (; n >= 0; n--)
  98. tgt[n] = scalar_product_float(src, src - n, len);
  99. }
  100. /**
  101. * Hybrid window filtering, see blocks 36 and 49 of the G.728 specification.
  102. *
  103. * @param order filter order
  104. * @param n input length
  105. * @param non_rec number of non-recursive samples
  106. * @param out filter output
  107. * @param hist pointer to the input history of the filter
  108. * @param out pointer to the non-recursive part of the output
  109. * @param out2 pointer to the recursive part of the output
  110. * @param window pointer to the windowing function table
  111. */
  112. static void do_hybrid_window(int order, int n, int non_rec, float *out,
  113. float *hist, float *out2, const float *window)
  114. {
  115. int i;
  116. float buffer1[order + 1];
  117. float buffer2[order + 1];
  118. float work[order + n + non_rec];
  119. apply_window(work, window, hist, order + n + non_rec);
  120. convolve(buffer1, work + order , n , order);
  121. convolve(buffer2, work + order + n, non_rec, order);
  122. for (i=0; i <= order; i++) {
  123. out2[i] = out2[i] * 0.5625 + buffer1[i];
  124. out [i] = out2[i] + buffer2[i];
  125. }
  126. /* Multiply by the white noise correcting factor (WNCF). */
  127. *out *= 257./256.;
  128. }
  129. /**
  130. * Backward synthesis filter, find the LPC coefficients from past speech data.
  131. */
  132. static void backward_filter(RA288Context *ractx)
  133. {
  134. float temp1[37]; // RTMP in the spec
  135. float temp2[11]; // GPTPMP in the spec
  136. do_hybrid_window(36, 40, 35, temp1, ractx->sp_hist,
  137. ractx->sp_rec, syn_window);
  138. if (!compute_lpc_coefs(temp1, 36, ractx->sp_lpc, 0, 1, 1))
  139. apply_window(ractx->sp_lpc, ractx->sp_lpc, syn_bw_tab, 36);
  140. do_hybrid_window(10, 8, 20, temp2, ractx->gain_hist,
  141. ractx->gain_rec, gain_window);
  142. if (!compute_lpc_coefs(temp2, 10, ractx->gain_lpc, 0, 1, 1))
  143. apply_window(ractx->gain_lpc, ractx->gain_lpc, gain_bw_tab, 10);
  144. memmove(ractx->gain_hist, ractx->gain_hist + 8,
  145. 28*sizeof(*ractx->gain_hist));
  146. memmove(ractx->sp_hist , ractx->sp_hist + 40,
  147. 70*sizeof(*ractx->sp_hist ));
  148. }
  149. static int ra288_decode_frame(AVCodecContext * avctx, void *data,
  150. int *data_size, const uint8_t * buf,
  151. int buf_size)
  152. {
  153. int16_t *out = data;
  154. int i, j;
  155. RA288Context *ractx = avctx->priv_data;
  156. GetBitContext gb;
  157. if (buf_size < avctx->block_align) {
  158. av_log(avctx, AV_LOG_ERROR,
  159. "Error! Input buffer is too small [%d<%d]\n",
  160. buf_size, avctx->block_align);
  161. return 0;
  162. }
  163. if (*data_size < 32*5*2)
  164. return -1;
  165. init_get_bits(&gb, buf, avctx->block_align * 8);
  166. for (i=0; i < 32; i++) {
  167. float gain = amptable[get_bits(&gb, 3)];
  168. int cb_coef = get_bits(&gb, 6 + (i&1));
  169. decode(ractx, gain, cb_coef);
  170. for (j=0; j < 5; j++)
  171. *(out++) = 8 * ractx->sp_hist[70 + 36 + j];
  172. if ((i & 7) == 3)
  173. backward_filter(ractx);
  174. }
  175. *data_size = (char *)out - (char *)data;
  176. return avctx->block_align;
  177. }
  178. AVCodec ra_288_decoder =
  179. {
  180. "real_288",
  181. CODEC_TYPE_AUDIO,
  182. CODEC_ID_RA_288,
  183. sizeof(RA288Context),
  184. ra288_decode_init,
  185. NULL,
  186. NULL,
  187. ra288_decode_frame,
  188. .long_name = NULL_IF_CONFIG_SMALL("RealAudio 2.0 (28.8K)"),
  189. };