audio-io.h 6.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228
  1. /******************************************************************************
  2. Copyright (C) 2023 by Lain Bailey <lain@obsproject.com>
  3. This program is free software: you can redistribute it and/or modify
  4. it under the terms of the GNU General Public License as published by
  5. the Free Software Foundation, either version 2 of the License, or
  6. (at your option) any later version.
  7. This program is distributed in the hope that it will be useful,
  8. but WITHOUT ANY WARRANTY; without even the implied warranty of
  9. MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  10. GNU General Public License for more details.
  11. You should have received a copy of the GNU General Public License
  12. along with this program. If not, see <http://www.gnu.org/licenses/>.
  13. ******************************************************************************/
  14. #pragma once
  15. #include "media-io-defs.h"
  16. #include "../util/c99defs.h"
  17. #include "../util/util_uint64.h"
  18. #ifdef __cplusplus
  19. extern "C" {
  20. #endif
  21. #define MAX_AUDIO_MIXES 6
  22. #define MAX_AUDIO_CHANNELS 8
  23. #define MAX_DEVICE_INPUT_CHANNELS 64
  24. #define AUDIO_OUTPUT_FRAMES 1024
  25. #define TOTAL_AUDIO_SIZE (MAX_AUDIO_MIXES * MAX_AUDIO_CHANNELS * AUDIO_OUTPUT_FRAMES * sizeof(float))
  26. /*
  27. * Base audio output component. Use this to create an audio output track
  28. * for the media.
  29. */
  30. struct audio_output;
  31. typedef struct audio_output audio_t;
  32. enum audio_format {
  33. AUDIO_FORMAT_UNKNOWN,
  34. AUDIO_FORMAT_U8BIT,
  35. AUDIO_FORMAT_16BIT,
  36. AUDIO_FORMAT_32BIT,
  37. AUDIO_FORMAT_FLOAT,
  38. AUDIO_FORMAT_U8BIT_PLANAR,
  39. AUDIO_FORMAT_16BIT_PLANAR,
  40. AUDIO_FORMAT_32BIT_PLANAR,
  41. AUDIO_FORMAT_FLOAT_PLANAR,
  42. };
  43. /**
  44. * The speaker layout describes where the speakers are located in the room.
  45. * For OBS it dictates:
  46. * * how many channels are available and
  47. * * which channels are used for which speakers.
  48. *
  49. * Standard channel layouts where retrieved from ffmpeg documentation at:
  50. * https://trac.ffmpeg.org/wiki/AudioChannelManipulation
  51. */
  52. enum speaker_layout {
  53. SPEAKERS_UNKNOWN, /**< Unknown setting, fallback is stereo. */
  54. SPEAKERS_MONO, /**< Channels: MONO */
  55. SPEAKERS_STEREO, /**< Channels: FL, FR */
  56. SPEAKERS_2POINT1, /**< Channels: FL, FR, LFE */
  57. SPEAKERS_4POINT0, /**< Channels: FL, FR, FC, RC */
  58. SPEAKERS_4POINT1, /**< Channels: FL, FR, FC, LFE, RC */
  59. SPEAKERS_5POINT1, /**< Channels: FL, FR, FC, LFE, RL, RR */
  60. SPEAKERS_7POINT1 = 8, /**< Channels: FL, FR, FC, LFE, RL, RR, SL, SR */
  61. };
  62. struct audio_data {
  63. uint8_t *data[MAX_AV_PLANES];
  64. uint32_t frames;
  65. uint64_t timestamp;
  66. };
  67. struct audio_output_data {
  68. float *data[MAX_AUDIO_CHANNELS];
  69. };
  70. typedef bool (*audio_input_callback_t)(void *param, uint64_t start_ts, uint64_t end_ts, uint64_t *new_ts,
  71. uint32_t active_mixers, struct audio_output_data *mixes);
  72. struct audio_output_info {
  73. const char *name;
  74. uint32_t samples_per_sec;
  75. enum audio_format format;
  76. enum speaker_layout speakers;
  77. audio_input_callback_t input_callback;
  78. void *input_param;
  79. };
  80. struct audio_convert_info {
  81. uint32_t samples_per_sec;
  82. enum audio_format format;
  83. enum speaker_layout speakers;
  84. bool allow_clipping;
  85. };
  86. static inline uint32_t get_audio_channels(enum speaker_layout speakers)
  87. {
  88. switch (speakers) {
  89. case SPEAKERS_MONO:
  90. return 1;
  91. case SPEAKERS_STEREO:
  92. return 2;
  93. case SPEAKERS_2POINT1:
  94. return 3;
  95. case SPEAKERS_4POINT0:
  96. return 4;
  97. case SPEAKERS_4POINT1:
  98. return 5;
  99. case SPEAKERS_5POINT1:
  100. return 6;
  101. case SPEAKERS_7POINT1:
  102. return 8;
  103. case SPEAKERS_UNKNOWN:
  104. return 0;
  105. }
  106. return 0;
  107. }
  108. static inline size_t get_audio_bytes_per_channel(enum audio_format format)
  109. {
  110. switch (format) {
  111. case AUDIO_FORMAT_U8BIT:
  112. case AUDIO_FORMAT_U8BIT_PLANAR:
  113. return 1;
  114. case AUDIO_FORMAT_16BIT:
  115. case AUDIO_FORMAT_16BIT_PLANAR:
  116. return 2;
  117. case AUDIO_FORMAT_FLOAT:
  118. case AUDIO_FORMAT_FLOAT_PLANAR:
  119. case AUDIO_FORMAT_32BIT:
  120. case AUDIO_FORMAT_32BIT_PLANAR:
  121. return 4;
  122. case AUDIO_FORMAT_UNKNOWN:
  123. return 0;
  124. }
  125. return 0;
  126. }
  127. static inline bool is_audio_planar(enum audio_format format)
  128. {
  129. switch (format) {
  130. case AUDIO_FORMAT_U8BIT:
  131. case AUDIO_FORMAT_16BIT:
  132. case AUDIO_FORMAT_32BIT:
  133. case AUDIO_FORMAT_FLOAT:
  134. return false;
  135. case AUDIO_FORMAT_U8BIT_PLANAR:
  136. case AUDIO_FORMAT_FLOAT_PLANAR:
  137. case AUDIO_FORMAT_16BIT_PLANAR:
  138. case AUDIO_FORMAT_32BIT_PLANAR:
  139. return true;
  140. case AUDIO_FORMAT_UNKNOWN:
  141. return false;
  142. }
  143. return false;
  144. }
  145. static inline size_t get_audio_planes(enum audio_format format, enum speaker_layout speakers)
  146. {
  147. return (is_audio_planar(format) ? get_audio_channels(speakers) : 1);
  148. }
  149. static inline size_t get_audio_size(enum audio_format format, enum speaker_layout speakers, uint32_t frames)
  150. {
  151. bool planar = is_audio_planar(format);
  152. return (planar ? 1 : get_audio_channels(speakers)) * get_audio_bytes_per_channel(format) * frames;
  153. }
  154. static inline size_t get_total_audio_size(enum audio_format format, enum speaker_layout speakers, uint32_t frames)
  155. {
  156. return get_audio_channels(speakers) * get_audio_bytes_per_channel(format) * frames;
  157. }
  158. static inline uint64_t audio_frames_to_ns(size_t sample_rate, uint64_t frames)
  159. {
  160. return util_mul_div64(frames, 1000000000ULL, sample_rate);
  161. }
  162. static inline uint64_t ns_to_audio_frames(size_t sample_rate, uint64_t frames)
  163. {
  164. return util_mul_div64(frames, sample_rate, 1000000000ULL);
  165. }
  166. #define AUDIO_OUTPUT_SUCCESS 0
  167. #define AUDIO_OUTPUT_INVALIDPARAM -1
  168. #define AUDIO_OUTPUT_FAIL -2
  169. EXPORT int audio_output_open(audio_t **audio, struct audio_output_info *info);
  170. EXPORT void audio_output_close(audio_t *audio);
  171. typedef void (*audio_output_callback_t)(void *param, size_t mix_idx, struct audio_data *data);
  172. EXPORT bool audio_output_connect(audio_t *video, size_t mix_idx, const struct audio_convert_info *conversion,
  173. audio_output_callback_t callback, void *param);
  174. EXPORT void audio_output_disconnect(audio_t *video, size_t mix_idx, audio_output_callback_t callback, void *param);
  175. EXPORT bool audio_output_active(const audio_t *audio);
  176. EXPORT size_t audio_output_get_block_size(const audio_t *audio);
  177. EXPORT size_t audio_output_get_planes(const audio_t *audio);
  178. EXPORT size_t audio_output_get_channels(const audio_t *audio);
  179. EXPORT uint32_t audio_output_get_sample_rate(const audio_t *audio);
  180. EXPORT const struct audio_output_info *audio_output_get_info(const audio_t *audio);
  181. #ifdef __cplusplus
  182. }
  183. #endif