FFmpeg: libavcodec/libspeexenc.c Source File

00001 /*
00002  * Copyright (C) 2009 Justin Ruggles
00003  * Copyright (c) 2009 Xuggle Incorporated
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * FFmpeg is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00067 #include <speex/speex.h>
00068 #include <speex/speex_header.h>
00069 #include <speex/speex_stereo.h>
00070 #include "libavutil/opt.h"
00071 #include "avcodec.h"
00072 #include "internal.h"
00073 #include "audio_frame_queue.h"
00074 
00075 typedef struct {
00076     AVClass *class;             
00077     SpeexBits bits;             
00078     SpeexHeader header;         
00079     void *enc_state;            
00080     int frames_per_packet;      
00081     float vbr_quality;          
00082     int cbr_quality;            
00083     int abr;                    
00084     int pkt_frame_count;        
00085     AudioFrameQueue afq;        
00086 } LibSpeexEncContext;
00087 
00088 static av_cold void print_enc_params(AVCodecContext *avctx,
00089                                      LibSpeexEncContext *s)
00090 {
00091     const char *mode_str = "unknown";
00092 
00093     av_log(avctx, AV_LOG_DEBUG, "channels: %d\n", avctx->channels);
00094     switch (s->header.mode) {
00095     case SPEEX_MODEID_NB:  mode_str = "narrowband";     break;
00096     case SPEEX_MODEID_WB:  mode_str = "wideband";       break;
00097     case SPEEX_MODEID_UWB: mode_str = "ultra-wideband"; break;
00098     }
00099     av_log(avctx, AV_LOG_DEBUG, "mode: %s\n", mode_str);
00100     if (s->header.vbr) {
00101         av_log(avctx, AV_LOG_DEBUG, "rate control: VBR\n");
00102         av_log(avctx, AV_LOG_DEBUG, "  quality: %f\n", s->vbr_quality);
00103     } else if (s->abr) {
00104         av_log(avctx, AV_LOG_DEBUG, "rate control: ABR\n");
00105         av_log(avctx, AV_LOG_DEBUG, "  bitrate: %d bps\n", avctx->bit_rate);
00106     } else {
00107         av_log(avctx, AV_LOG_DEBUG, "rate control: CBR\n");
00108         av_log(avctx, AV_LOG_DEBUG, "  bitrate: %d bps\n", avctx->bit_rate);
00109     }
00110     av_log(avctx, AV_LOG_DEBUG, "complexity: %d\n",
00111            avctx->compression_level);
00112     av_log(avctx, AV_LOG_DEBUG, "frame size: %d samples\n",
00113            avctx->frame_size);
00114     av_log(avctx, AV_LOG_DEBUG, "frames per packet: %d\n",
00115            s->frames_per_packet);
00116     av_log(avctx, AV_LOG_DEBUG, "packet size: %d\n",
00117            avctx->frame_size * s->frames_per_packet);
00118 }
00119 
00120 static av_cold int encode_init(AVCodecContext *avctx)
00121 {
00122     LibSpeexEncContext *s = avctx->priv_data;
00123     const SpeexMode *mode;
00124     uint8_t *header_data;
00125     int header_size;
00126     int32_t complexity;
00127 
00128     /* channels */
00129     if (avctx->channels < 1 || avctx->channels > 2) {
00130         av_log(avctx, AV_LOG_ERROR, "Invalid channels (%d). Only stereo and "
00131                "mono are supported\n", avctx->channels);
00132         return AVERROR(EINVAL);
00133     }
00134 
00135     /* sample rate and encoding mode */
00136     switch (avctx->sample_rate) {
00137     case  8000: mode = &speex_nb_mode;  break;
00138     case 16000: mode = &speex_wb_mode;  break;
00139     case 32000: mode = &speex_uwb_mode; break;
00140     default:
00141         av_log(avctx, AV_LOG_ERROR, "Sample rate of %d Hz is not supported. "
00142                "Resample to 8, 16, or 32 kHz.\n", avctx->sample_rate);
00143         return AVERROR(EINVAL);
00144     }
00145 
00146     /* initialize libspeex */
00147     s->enc_state = speex_encoder_init(mode);
00148     if (!s->enc_state) {
00149         av_log(avctx, AV_LOG_ERROR, "Error initializing libspeex\n");
00150         return -1;
00151     }
00152     speex_init_header(&s->header, avctx->sample_rate, avctx->channels, mode);
00153 
00154     /* rate control method and parameters */
00155     if (avctx->flags & CODEC_FLAG_QSCALE) {
00156         /* VBR */
00157         s->header.vbr = 1;
00158         speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR, &s->header.vbr);
00159         s->vbr_quality = av_clipf(avctx->global_quality / (float)FF_QP2LAMBDA,
00160                                   0.0f, 10.0f);
00161         speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR_QUALITY, &s->vbr_quality);
00162     } else {
00163         s->header.bitrate = avctx->bit_rate;
00164         if (avctx->bit_rate > 0) {
00165             /* CBR or ABR by bitrate */
00166             if (s->abr) {
00167                 speex_encoder_ctl(s->enc_state, SPEEX_SET_ABR,
00168                                   &s->header.bitrate);
00169                 speex_encoder_ctl(s->enc_state, SPEEX_GET_ABR,
00170                                   &s->header.bitrate);
00171             } else {
00172                 speex_encoder_ctl(s->enc_state, SPEEX_SET_BITRATE,
00173                                   &s->header.bitrate);
00174                 speex_encoder_ctl(s->enc_state, SPEEX_GET_BITRATE,
00175                                   &s->header.bitrate);
00176             }
00177         } else {
00178             /* CBR by quality */
00179             speex_encoder_ctl(s->enc_state, SPEEX_SET_QUALITY,
00180                               &s->cbr_quality);
00181             speex_encoder_ctl(s->enc_state, SPEEX_GET_BITRATE,
00182                               &s->header.bitrate);
00183         }
00184         /* stereo side information adds about 800 bps to the base bitrate */
00185         /* TODO: this should be calculated exactly */
00186         avctx->bit_rate = s->header.bitrate + (avctx->channels == 2 ? 800 : 0);
00187     }
00188 
00189     /* set encoding complexity */
00190     if (avctx->compression_level > FF_COMPRESSION_DEFAULT) {
00191         complexity = av_clip(avctx->compression_level, 0, 10);
00192         speex_encoder_ctl(s->enc_state, SPEEX_SET_COMPLEXITY, &complexity);
00193     }
00194     speex_encoder_ctl(s->enc_state, SPEEX_GET_COMPLEXITY, &complexity);
00195     avctx->compression_level = complexity;
00196 
00197     /* set packet size */
00198     avctx->frame_size = s->header.frame_size;
00199     s->header.frames_per_packet = s->frames_per_packet;
00200 
00201     /* set encoding delay */
00202     speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &avctx->delay);
00203     ff_af_queue_init(avctx, &s->afq);
00204 
00205     /* create header packet bytes from header struct */
00206     /* note: libspeex allocates the memory for header_data, which is freed
00207              below with speex_header_free() */
00208     header_data = speex_header_to_packet(&s->header, &header_size);
00209 
00210     /* allocate extradata and coded_frame */
00211     avctx->extradata   = av_malloc(header_size + FF_INPUT_BUFFER_PADDING_SIZE);
00212     if (!avctx->extradata) {
00213         speex_header_free(header_data);
00214         speex_encoder_destroy(s->enc_state);
00215         av_log(avctx, AV_LOG_ERROR, "memory allocation error\n");
00216         return AVERROR(ENOMEM);
00217     }
00218 #if FF_API_OLD_ENCODE_AUDIO
00219     avctx->coded_frame = avcodec_alloc_frame();
00220     if (!avctx->coded_frame) {
00221         av_freep(&avctx->extradata);
00222         speex_header_free(header_data);
00223         speex_encoder_destroy(s->enc_state);
00224         av_log(avctx, AV_LOG_ERROR, "memory allocation error\n");
00225         return AVERROR(ENOMEM);
00226     }
00227 #endif
00228 
00229     /* copy header packet to extradata */
00230     memcpy(avctx->extradata, header_data, header_size);
00231     avctx->extradata_size = header_size;
00232     speex_header_free(header_data);
00233 
00234     /* init libspeex bitwriter */
00235     speex_bits_init(&s->bits);
00236 
00237     print_enc_params(avctx, s);
00238     return 0;
00239 }
00240 
00241 static int encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
00242                         const AVFrame *frame, int *got_packet_ptr)
00243 {
00244     LibSpeexEncContext *s = avctx->priv_data;
00245     int16_t *samples      = frame ? (int16_t *)frame->data[0] : NULL;
00246     int ret;
00247 
00248     if (samples) {
00249         /* encode Speex frame */
00250         if (avctx->channels == 2)
00251             speex_encode_stereo_int(samples, s->header.frame_size, &s->bits);
00252         speex_encode_int(s->enc_state, samples, &s->bits);
00253         s->pkt_frame_count++;
00254         if ((ret = ff_af_queue_add(&s->afq, frame) < 0))
00255             return ret;
00256     } else {
00257         /* handle end-of-stream */
00258         if (!s->pkt_frame_count)
00259             return 0;
00260         /* add extra terminator codes for unused frames in last packet */
00261         while (s->pkt_frame_count < s->frames_per_packet) {
00262             speex_bits_pack(&s->bits, 15, 5);
00263             s->pkt_frame_count++;
00264         }
00265     }
00266 
00267     /* write output if all frames for the packet have been encoded */
00268     if (s->pkt_frame_count == s->frames_per_packet) {
00269         s->pkt_frame_count = 0;
00270         if ((ret = ff_alloc_packet2(avctx, avpkt, speex_bits_nbytes(&s->bits))))
00271             return ret;
00272         ret = speex_bits_write(&s->bits, avpkt->data, avpkt->size);
00273         speex_bits_reset(&s->bits);
00274 
00275         /* Get the next frame pts/duration */
00276         ff_af_queue_remove(&s->afq, s->frames_per_packet * avctx->frame_size,
00277                            &avpkt->pts, &avpkt->duration);
00278 
00279         avpkt->size = ret;
00280         *got_packet_ptr = 1;
00281         return 0;
00282     }
00283     return 0;
00284 }
00285 
00286 static av_cold int encode_close(AVCodecContext *avctx)
00287 {
00288     LibSpeexEncContext *s = avctx->priv_data;
00289 
00290     speex_bits_destroy(&s->bits);
00291     speex_encoder_destroy(s->enc_state);
00292 
00293     ff_af_queue_close(&s->afq);
00294 #if FF_API_OLD_ENCODE_AUDIO
00295     av_freep(&avctx->coded_frame);
00296 #endif
00297     av_freep(&avctx->extradata);
00298 
00299     return 0;
00300 }
00301 
00302 #define OFFSET(x) offsetof(LibSpeexEncContext, x)
00303 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
00304 static const AVOption options[] = {
00305     { "abr",               "Use average bit rate",                      OFFSET(abr),               AV_OPT_TYPE_INT, { 0 }, 0,   1, AE },
00306     { "cbr_quality",       "Set quality value (0 to 10) for CBR",       OFFSET(cbr_quality),       AV_OPT_TYPE_INT, { 8 }, 0,  10, AE },
00307     { "frames_per_packet", "Number of frames to encode in each packet", OFFSET(frames_per_packet), AV_OPT_TYPE_INT, { 1 }, 1,   8, AE },
00308     { NULL },
00309 };
00310 
00311 static const AVClass class = {
00312     .class_name = "libspeex",
00313     .item_name  = av_default_item_name,
00314     .option     = options,
00315     .version    = LIBAVUTIL_VERSION_INT,
00316 };
00317 
00318 static const AVCodecDefault defaults[] = {
00319     { "b",                 "0" },
00320     { "compression_level", "3" },
00321     { NULL },
00322 };
00323 
00324 AVCodec ff_libspeex_encoder = {
00325     .name           = "libspeex",
00326     .type           = AVMEDIA_TYPE_AUDIO,
00327     .id             = CODEC_ID_SPEEX,
00328     .priv_data_size = sizeof(LibSpeexEncContext),
00329     .init           = encode_init,
00330     .encode2        = encode_frame,
00331     .close          = encode_close,
00332     .capabilities   = CODEC_CAP_DELAY,
00333     .sample_fmts    = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
00334                                                      AV_SAMPLE_FMT_NONE },
00335     .long_name      = NULL_IF_CONFIG_SMALL("libspeex Speex"),
00336     .priv_class     = &class,
00337     .defaults       = defaults,
00338 };