libavcodec/libspeexenc.c

   1 /*
   2  * Copyright (C) 2009 Justin Ruggles
   3  * Copyright (c) 2009 Xuggle Incorporated
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav 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  * Libav 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 Libav; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 /**
  23  * @file
  24  * libspeex Speex audio encoder
  25  *
  26  * Usage Guide
  27  * This explains the values that need to be set prior to initialization in
  28  * order to control various encoding parameters.
  29  *
  30  * Channels
  31  *     Speex only supports mono or stereo, so avctx->channels must be set to
  32  *     1 or 2.
  33  *
  34  * Sample Rate / Encoding Mode
  35  *     Speex has 3 modes, each of which uses a specific sample rate.
  36  *         narrowband     :  8 kHz
  37  *         wideband       : 16 kHz
  38  *         ultra-wideband : 32 kHz
  39  *     avctx->sample_rate must be set to one of these 3 values.  This will be
  40  *     used to set the encoding mode.
  41  *
  42  * Rate Control
  43  *     VBR mode is turned on by setting AV_CODEC_FLAG_QSCALE in avctx->flags.
  44  *     avctx->global_quality is used to set the encoding quality.
  45  *     For CBR mode, avctx->bit_rate can be used to set the constant bitrate.
  46  *     Alternatively, the 'cbr_quality' option can be set from 0 to 10 to set
  47  *     a constant bitrate based on quality.
  48  *     For ABR mode, set avctx->bit_rate and set the 'abr' option to 1.
  49  *     Approx. Bitrate Range:
  50  *         narrowband     : 2400 - 25600 bps
  51  *         wideband       : 4000 - 43200 bps
  52  *         ultra-wideband : 4400 - 45200 bps
  53  *
  54  * Complexity
  55  *     Encoding complexity is controlled by setting avctx->compression_level.
  56  *     The valid range is 0 to 10.  A higher setting gives generally better
  57  *     quality at the expense of encoding speed.  This does not affect the
  58  *     bit rate.
  59  *
  60  * Frames-per-Packet
  61  *     The encoder defaults to using 1 frame-per-packet.  However, it is
  62  *     sometimes desirable to use multiple frames-per-packet to reduce the
  63  *     amount of container overhead.  This can be done by setting the
  64  *     'frames_per_packet' option to a value 1 to 8.
  65  *
  66  *
  67  * Optional features
  68  * Speex encoder supports several optional features, which can be useful
  69  * for some conditions.
  70  *
  71  * Voice Activity Detection
  72  *     When enabled, voice activity detection detects whether the audio
  73  *     being encoded is speech or silence/background noise. VAD is always
  74  *     implicitly activated when encoding in VBR, so the option is only useful
  75  *     in non-VBR operation. In this case, Speex detects non-speech periods and
  76  *     encodes them with just enough bits to reproduce the background noise.
  77  *
  78  * Discontinuous Transmission (DTX)
  79  *     DTX is an addition to VAD/VBR operation, that allows to stop transmitting
  80  *     completely when the background noise is stationary.
  81  *     In file-based operation only 5 bits are used for such frames.
  82  */
  83
  84 #include <speex/speex.h>
  85 #include <speex/speex_header.h>
  86 #include <speex/speex_stereo.h>
  87
  88 #include "libavutil/channel_layout.h"
  89 #include "libavutil/common.h"
  90 #include "libavutil/opt.h"
  91 #include "avcodec.h"
  92 #include "internal.h"
  93 #include "audio_frame_queue.h"
  94
  95 typedef struct LibSpeexEncContext {
  96     AVClass *class;             ///< AVClass for private options
  97     SpeexBits bits;             ///< libspeex bitwriter context
  98     SpeexHeader header;         ///< libspeex header struct
  99     void *enc_state;            ///< libspeex encoder state
 100     int frames_per_packet;      ///< number of frames to encode in each packet
 101     float vbr_quality;          ///< VBR quality 0.0 to 10.0
 102     int cbr_quality;            ///< CBR quality 0 to 10
 103     int abr;                    ///< flag to enable ABR
 104     int vad;                    ///< flag to enable VAD
 105     int dtx;                    ///< flag to enable DTX
 106     int pkt_frame_count;        ///< frame count for the current packet
 107     AudioFrameQueue afq;        ///< frame queue
 108 } LibSpeexEncContext;
 109
 110 static av_cold void print_enc_params(AVCodecContext *avctx,
 111                                      LibSpeexEncContext *s)
 112 {
 113     const char *mode_str = "unknown";
 114
 115     av_log(avctx, AV_LOG_DEBUG, "channels: %d\n", avctx->channels);
 116     switch (s->header.mode) {
 117     case SPEEX_MODEID_NB:  mode_str = "narrowband";     break;
 118     case SPEEX_MODEID_WB:  mode_str = "wideband";       break;
 119     case SPEEX_MODEID_UWB: mode_str = "ultra-wideband"; break;
 120     }
 121     av_log(avctx, AV_LOG_DEBUG, "mode: %s\n", mode_str);
 122     if (s->header.vbr) {
 123         av_log(avctx, AV_LOG_DEBUG, "rate control: VBR\n");
 124         av_log(avctx, AV_LOG_DEBUG, "  quality: %f\n", s->vbr_quality);
 125     } else if (s->abr) {
 126         av_log(avctx, AV_LOG_DEBUG, "rate control: ABR\n");
 127         av_log(avctx, AV_LOG_DEBUG, "  bitrate: %d bps\n", avctx->bit_rate);
 128     } else {
 129         av_log(avctx, AV_LOG_DEBUG, "rate control: CBR\n");
 130         av_log(avctx, AV_LOG_DEBUG, "  bitrate: %d bps\n", avctx->bit_rate);
 131     }
 132     av_log(avctx, AV_LOG_DEBUG, "complexity: %d\n",
 133            avctx->compression_level);
 134     av_log(avctx, AV_LOG_DEBUG, "frame size: %d samples\n",
 135            avctx->frame_size);
 136     av_log(avctx, AV_LOG_DEBUG, "frames per packet: %d\n",
 137            s->frames_per_packet);
 138     av_log(avctx, AV_LOG_DEBUG, "packet size: %d\n",
 139            avctx->frame_size * s->frames_per_packet);
 140     av_log(avctx, AV_LOG_DEBUG, "voice activity detection: %d\n", s->vad);
 141     av_log(avctx, AV_LOG_DEBUG, "discontinuous transmission: %d\n", s->dtx);
 142 }
 143
 144 static av_cold int encode_init(AVCodecContext *avctx)
 145 {
 146     LibSpeexEncContext *s = avctx->priv_data;
 147     const SpeexMode *mode;
 148     uint8_t *header_data;
 149     int header_size;
 150     int32_t complexity;
 151
 152     /* channels */
 153     if (avctx->channels < 1 || avctx->channels > 2) {
 154         av_log(avctx, AV_LOG_ERROR, "Invalid channels (%d). Only stereo and "
 155                "mono are supported\n", avctx->channels);
 156         return AVERROR(EINVAL);
 157     }
 158
 159     /* sample rate and encoding mode */
 160     switch (avctx->sample_rate) {
 161     case  8000: mode = speex_lib_get_mode(SPEEX_MODEID_NB);  break;
 162     case 16000: mode = speex_lib_get_mode(SPEEX_MODEID_WB);  break;
 163     case 32000: mode = speex_lib_get_mode(SPEEX_MODEID_UWB); break;
 164     default:
 165         av_log(avctx, AV_LOG_ERROR, "Sample rate of %d Hz is not supported. "
 166                "Resample to 8, 16, or 32 kHz.\n", avctx->sample_rate);
 167         return AVERROR(EINVAL);
 168     }
 169
 170     /* initialize libspeex */
 171     s->enc_state = speex_encoder_init(mode);
 172     if (!s->enc_state) {
 173         av_log(avctx, AV_LOG_ERROR, "Error initializing libspeex\n");
 174         return -1;
 175     }
 176     speex_init_header(&s->header, avctx->sample_rate, avctx->channels, mode);
 177
 178     /* rate control method and parameters */
 179     if (avctx->flags & AV_CODEC_FLAG_QSCALE) {
 180         /* VBR */
 181         s->header.vbr = 1;
 182         s->vad = 1; /* VAD is always implicitly activated for VBR */
 183         speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR, &s->header.vbr);
 184         s->vbr_quality = av_clipf(avctx->global_quality / (float)FF_QP2LAMBDA,
 185                                   0.0f, 10.0f);
 186         speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR_QUALITY, &s->vbr_quality);
 187     } else {
 188         s->header.bitrate = avctx->bit_rate;
 189         if (avctx->bit_rate > 0) {
 190             /* CBR or ABR by bitrate */
 191             if (s->abr) {
 192                 speex_encoder_ctl(s->enc_state, SPEEX_SET_ABR,
 193                                   &s->header.bitrate);
 194                 speex_encoder_ctl(s->enc_state, SPEEX_GET_ABR,
 195                                   &s->header.bitrate);
 196             } else {
 197                 speex_encoder_ctl(s->enc_state, SPEEX_SET_BITRATE,
 198                                   &s->header.bitrate);
 199                 speex_encoder_ctl(s->enc_state, SPEEX_GET_BITRATE,
 200                                   &s->header.bitrate);
 201             }
 202         } else {
 203             /* CBR by quality */
 204             speex_encoder_ctl(s->enc_state, SPEEX_SET_QUALITY,
 205                               &s->cbr_quality);
 206             speex_encoder_ctl(s->enc_state, SPEEX_GET_BITRATE,
 207                               &s->header.bitrate);
 208         }
 209         /* stereo side information adds about 800 bps to the base bitrate */
 210         /* TODO: this should be calculated exactly */
 211         avctx->bit_rate = s->header.bitrate + (avctx->channels == 2 ? 800 : 0);
 212     }
 213
 214     /* VAD is activated with VBR or can be turned on by itself */
 215     if (s->vad)
 216         speex_encoder_ctl(s->enc_state, SPEEX_SET_VAD, &s->vad);
 217
 218     /* Activating Discontinuous Transmission */
 219     if (s->dtx) {
 220         speex_encoder_ctl(s->enc_state, SPEEX_SET_DTX, &s->dtx);
 221         if (!(s->abr || s->vad || s->header.vbr))
 222             av_log(avctx, AV_LOG_WARNING, "DTX is not much of use without ABR, VAD or VBR\n");
 223     }
 224
 225     /* set encoding complexity */
 226     if (avctx->compression_level > FF_COMPRESSION_DEFAULT) {
 227         complexity = av_clip(avctx->compression_level, 0, 10);
 228         speex_encoder_ctl(s->enc_state, SPEEX_SET_COMPLEXITY, &complexity);
 229     }
 230     speex_encoder_ctl(s->enc_state, SPEEX_GET_COMPLEXITY, &complexity);
 231     avctx->compression_level = complexity;
 232
 233     /* set packet size */
 234     avctx->frame_size = s->header.frame_size;
 235     s->header.frames_per_packet = s->frames_per_packet;
 236
 237     /* set encoding delay */
 238     speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &avctx->initial_padding);
 239     ff_af_queue_init(avctx, &s->afq);
 240
 241     /* create header packet bytes from header struct */
 242     /* note: libspeex allocates the memory for header_data, which is freed
 243              below with speex_header_free() */
 244     header_data = speex_header_to_packet(&s->header, &header_size);
 245
 246     /* allocate extradata */
 247     avctx->extradata = av_malloc(header_size + AV_INPUT_BUFFER_PADDING_SIZE);
 248     if (!avctx->extradata) {
 249         speex_header_free(header_data);
 250         speex_encoder_destroy(s->enc_state);
 251         av_log(avctx, AV_LOG_ERROR, "memory allocation error\n");
 252         return AVERROR(ENOMEM);
 253     }
 254
 255     /* copy header packet to extradata */
 256     memcpy(avctx->extradata, header_data, header_size);
 257     avctx->extradata_size = header_size;
 258     speex_header_free(header_data);
 259
 260     /* init libspeex bitwriter */
 261     speex_bits_init(&s->bits);
 262
 263     print_enc_params(avctx, s);
 264     return 0;
 265 }
 266
 267 static int encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
 268                         const AVFrame *frame, int *got_packet_ptr)
 269 {
 270     LibSpeexEncContext *s = avctx->priv_data;
 271     int16_t *samples      = frame ? (int16_t *)frame->data[0] : NULL;
 272     int ret;
 273
 274     if (samples) {
 275         /* encode Speex frame */
 276         if (avctx->channels == 2)
 277             speex_encode_stereo_int(samples, s->header.frame_size, &s->bits);
 278         speex_encode_int(s->enc_state, samples, &s->bits);
 279         s->pkt_frame_count++;
 280         if ((ret = ff_af_queue_add(&s->afq, frame)) < 0)
 281             return ret;
 282     } else {
 283         /* handle end-of-stream */
 284         if (!s->pkt_frame_count)
 285             return 0;
 286         /* add extra terminator codes for unused frames in last packet */
 287         while (s->pkt_frame_count < s->frames_per_packet) {
 288             speex_bits_pack(&s->bits, 15, 5);
 289             s->pkt_frame_count++;
 290         }
 291     }
 292
 293     /* write output if all frames for the packet have been encoded */
 294     if (s->pkt_frame_count == s->frames_per_packet) {
 295         s->pkt_frame_count = 0;
 296         if ((ret = ff_alloc_packet(avpkt, speex_bits_nbytes(&s->bits)))) {
 297             av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n");
 298             return ret;
 299         }
 300         ret = speex_bits_write(&s->bits, avpkt->data, avpkt->size);
 301         speex_bits_reset(&s->bits);
 302
 303         /* Get the next frame pts/duration */
 304         ff_af_queue_remove(&s->afq, s->frames_per_packet * avctx->frame_size,
 305                            &avpkt->pts, &avpkt->duration);
 306
 307         avpkt->size = ret;
 308         *got_packet_ptr = 1;
 309         return 0;
 310     }
 311     return 0;
 312 }
 313
 314 static av_cold int encode_close(AVCodecContext *avctx)
 315 {
 316     LibSpeexEncContext *s = avctx->priv_data;
 317
 318     speex_bits_destroy(&s->bits);
 319     speex_encoder_destroy(s->enc_state);
 320
 321     ff_af_queue_close(&s->afq);
 322     av_freep(&avctx->extradata);
 323
 324     return 0;
 325 }
 326
 327 #define OFFSET(x) offsetof(LibSpeexEncContext, x)
 328 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
 329 static const AVOption options[] = {
 330     { "abr",               "Use average bit rate",                      OFFSET(abr),               AV_OPT_TYPE_INT, { .i64 = 0 }, 0,   1, AE },
 331     { "cbr_quality",       "Set quality value (0 to 10) for CBR",       OFFSET(cbr_quality),       AV_OPT_TYPE_INT, { .i64 = 8 }, 0,  10, AE },
 332     { "frames_per_packet", "Number of frames to encode in each packet", OFFSET(frames_per_packet), AV_OPT_TYPE_INT, { .i64 = 1 }, 1,   8, AE },
 333     { "vad",               "Voice Activity Detection",                  OFFSET(vad),               AV_OPT_TYPE_INT, { .i64 = 0 }, 0,   1, AE },
 334     { "dtx",               "Discontinuous Transmission",                OFFSET(dtx),               AV_OPT_TYPE_INT, { .i64 = 0 }, 0,   1, AE },
 335     { NULL },
 336 };
 337
 338 static const AVClass class = {
 339     .class_name = "libspeex",
 340     .item_name  = av_default_item_name,
 341     .option     = options,
 342     .version    = LIBAVUTIL_VERSION_INT,
 343 };
 344
 345 static const AVCodecDefault defaults[] = {
 346     { "b",                 "0" },
 347     { "compression_level", "3" },
 348     { NULL },
 349 };
 350
 351 AVCodec ff_libspeex_encoder = {
 352     .name           = "libspeex",
 353     .long_name      = NULL_IF_CONFIG_SMALL("libspeex Speex"),
 354     .type           = AVMEDIA_TYPE_AUDIO,
 355     .id             = AV_CODEC_ID_SPEEX,
 356     .priv_data_size = sizeof(LibSpeexEncContext),
 357     .init           = encode_init,
 358     .encode2        = encode_frame,
 359     .close          = encode_close,
 360     .capabilities   = AV_CODEC_CAP_DELAY,
 361     .sample_fmts    = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
 362                                                      AV_SAMPLE_FMT_NONE },
 363     .channel_layouts = (const uint64_t[]){ AV_CH_LAYOUT_MONO,
 364                                            AV_CH_LAYOUT_STEREO,
 365                                            0 },
 366     .supported_samplerates = (const int[]){ 8000, 16000, 32000, 0 },
 367     .priv_class     = &class,
 368     .defaults       = defaults,
 369     .wrapper_name   = "libspeex",
 370 };