2 * Copyright (C) 2009 Justin Ruggles
3 * Copyright (c) 2009 Xuggle Incorporated
5 * This file is part of FFmpeg.
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.
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.
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
24 * libspeex Speex audio encoder
27 * This explains the values that need to be set prior to initialization in
28 * order to control various encoding parameters.
31 * Speex only supports mono or stereo, so avctx->channels must be set to
34 * Sample Rate / Encoding Mode
35 * Speex has 3 modes, each of which uses a specific sample rate.
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.
43 * VBR mode is turned on by setting 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
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
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.
67 #include <speex/speex.h>
68 #include <speex/speex_header.h>
69 #include <speex/speex_stereo.h>
70 #include "libavutil/opt.h"
75 AVClass *class; ///< AVClass for private options
76 SpeexBits bits; ///< libspeex bitwriter context
77 SpeexHeader header; ///< libspeex header struct
78 void *enc_state; ///< libspeex encoder state
79 int frames_per_packet; ///< number of frames to encode in each packet
80 float vbr_quality; ///< VBR quality 0.0 to 10.0
81 int cbr_quality; ///< CBR quality 0 to 10
82 int abr; ///< flag to enable ABR
83 int pkt_frame_count; ///< frame count for the current packet
84 int64_t next_pts; ///< next pts, in sample_rate time base
85 int pkt_sample_count; ///< sample count in the current packet
88 static av_cold void print_enc_params(AVCodecContext *avctx,
89 LibSpeexEncContext *s)
91 const char *mode_str = "unknown";
93 av_log(avctx, AV_LOG_DEBUG, "channels: %d\n", avctx->channels);
94 switch (s->header.mode) {
95 case SPEEX_MODEID_NB: mode_str = "narrowband"; break;
96 case SPEEX_MODEID_WB: mode_str = "wideband"; break;
97 case SPEEX_MODEID_UWB: mode_str = "ultra-wideband"; break;
99 av_log(avctx, AV_LOG_DEBUG, "mode: %s\n", mode_str);
101 av_log(avctx, AV_LOG_DEBUG, "rate control: VBR\n");
102 av_log(avctx, AV_LOG_DEBUG, " quality: %f\n", s->vbr_quality);
104 av_log(avctx, AV_LOG_DEBUG, "rate control: ABR\n");
105 av_log(avctx, AV_LOG_DEBUG, " bitrate: %d bps\n", avctx->bit_rate);
107 av_log(avctx, AV_LOG_DEBUG, "rate control: CBR\n");
108 av_log(avctx, AV_LOG_DEBUG, " bitrate: %d bps\n", avctx->bit_rate);
110 av_log(avctx, AV_LOG_DEBUG, "complexity: %d\n",
111 avctx->compression_level);
112 av_log(avctx, AV_LOG_DEBUG, "frame size: %d samples\n",
114 av_log(avctx, AV_LOG_DEBUG, "frames per packet: %d\n",
115 s->frames_per_packet);
116 av_log(avctx, AV_LOG_DEBUG, "packet size: %d\n",
117 avctx->frame_size * s->frames_per_packet);
120 static av_cold int encode_init(AVCodecContext *avctx)
122 LibSpeexEncContext *s = avctx->priv_data;
123 const SpeexMode *mode;
124 uint8_t *header_data;
129 if (avctx->channels < 1 || avctx->channels > 2) {
130 av_log(avctx, AV_LOG_ERROR, "Invalid channels (%d). Only stereo and "
131 "mono are supported\n", avctx->channels);
132 return AVERROR(EINVAL);
135 /* sample rate and encoding mode */
136 switch (avctx->sample_rate) {
137 case 8000: mode = &speex_nb_mode; break;
138 case 16000: mode = &speex_wb_mode; break;
139 case 32000: mode = &speex_uwb_mode; break;
141 av_log(avctx, AV_LOG_ERROR, "Sample rate of %d Hz is not supported. "
142 "Resample to 8, 16, or 32 kHz.\n", avctx->sample_rate);
143 return AVERROR(EINVAL);
146 /* initialize libspeex */
147 s->enc_state = speex_encoder_init(mode);
149 av_log(avctx, AV_LOG_ERROR, "Error initializing libspeex\n");
152 speex_init_header(&s->header, avctx->sample_rate, avctx->channels, mode);
154 /* rate control method and parameters */
155 if (avctx->flags & CODEC_FLAG_QSCALE) {
158 speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR, &s->header.vbr);
159 s->vbr_quality = av_clipf(avctx->global_quality / (float)FF_QP2LAMBDA,
161 speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR_QUALITY, &s->vbr_quality);
163 s->header.bitrate = avctx->bit_rate;
164 if (avctx->bit_rate > 0) {
165 /* CBR or ABR by bitrate */
167 speex_encoder_ctl(s->enc_state, SPEEX_SET_ABR,
169 speex_encoder_ctl(s->enc_state, SPEEX_GET_ABR,
172 speex_encoder_ctl(s->enc_state, SPEEX_SET_BITRATE,
174 speex_encoder_ctl(s->enc_state, SPEEX_GET_BITRATE,
179 speex_encoder_ctl(s->enc_state, SPEEX_SET_QUALITY,
181 speex_encoder_ctl(s->enc_state, SPEEX_GET_BITRATE,
184 /* stereo side information adds about 800 bps to the base bitrate */
185 /* TODO: this should be calculated exactly */
186 avctx->bit_rate = s->header.bitrate + (avctx->channels == 2 ? 800 : 0);
189 /* set encoding complexity */
190 if (avctx->compression_level > FF_COMPRESSION_DEFAULT) {
191 complexity = av_clip(avctx->compression_level, 0, 10);
192 speex_encoder_ctl(s->enc_state, SPEEX_SET_COMPLEXITY, &complexity);
194 speex_encoder_ctl(s->enc_state, SPEEX_GET_COMPLEXITY, &complexity);
195 avctx->compression_level = complexity;
197 /* set packet size */
198 avctx->frame_size = s->header.frame_size;
199 s->header.frames_per_packet = s->frames_per_packet;
201 /* set encoding delay */
202 speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &avctx->delay);
204 /* create header packet bytes from header struct */
205 /* note: libspeex allocates the memory for header_data, which is freed
206 below with speex_header_free() */
207 header_data = speex_header_to_packet(&s->header, &header_size);
209 /* allocate extradata and coded_frame */
210 avctx->extradata = av_malloc(header_size + FF_INPUT_BUFFER_PADDING_SIZE);
211 avctx->coded_frame = avcodec_alloc_frame();
212 if (!avctx->extradata || !avctx->coded_frame) {
213 speex_header_free(header_data);
214 speex_encoder_destroy(s->enc_state);
215 av_log(avctx, AV_LOG_ERROR, "memory allocation error\n");
216 return AVERROR(ENOMEM);
219 /* copy header packet to extradata */
220 memcpy(avctx->extradata, header_data, header_size);
221 avctx->extradata_size = header_size;
222 speex_header_free(header_data);
224 /* init libspeex bitwriter */
225 speex_bits_init(&s->bits);
227 print_enc_params(avctx, s);
231 static int encode_frame(AVCodecContext *avctx, uint8_t *frame, int buf_size,
234 LibSpeexEncContext *s = avctx->priv_data;
235 int16_t *samples = data;
238 /* encode Speex frame */
239 if (avctx->channels == 2)
240 speex_encode_stereo_int(samples, s->header.frame_size, &s->bits);
241 speex_encode_int(s->enc_state, samples, &s->bits);
242 s->pkt_frame_count++;
243 s->pkt_sample_count += avctx->frame_size;
245 /* handle end-of-stream */
246 if (!s->pkt_frame_count)
248 /* add extra terminator codes for unused frames in last packet */
249 while (s->pkt_frame_count < s->frames_per_packet) {
250 speex_bits_pack(&s->bits, 15, 5);
251 s->pkt_frame_count++;
255 /* write output if all frames for the packet have been encoded */
256 if (s->pkt_frame_count == s->frames_per_packet) {
257 s->pkt_frame_count = 0;
258 avctx->coded_frame->pts = ff_samples_to_time_base(avctx, s->next_pts -
260 s->next_pts += s->pkt_sample_count;
261 s->pkt_sample_count = 0;
262 if (buf_size > speex_bits_nbytes(&s->bits)) {
263 int ret = speex_bits_write(&s->bits, frame, buf_size);
264 speex_bits_reset(&s->bits);
267 av_log(avctx, AV_LOG_ERROR, "output buffer too small");
268 return AVERROR(EINVAL);
274 static av_cold int encode_close(AVCodecContext *avctx)
276 LibSpeexEncContext *s = avctx->priv_data;
278 speex_bits_destroy(&s->bits);
279 speex_encoder_destroy(s->enc_state);
281 av_freep(&avctx->coded_frame);
282 av_freep(&avctx->extradata);
287 #define OFFSET(x) offsetof(LibSpeexEncContext, x)
288 #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
289 static const AVOption options[] = {
290 { "abr", "Use average bit rate", OFFSET(abr), AV_OPT_TYPE_INT, { 0 }, 0, 1, AE },
291 { "cbr_quality", "Set quality value (0 to 10) for CBR", OFFSET(cbr_quality), AV_OPT_TYPE_INT, { 8 }, 0, 10, AE },
292 { "frames_per_packet", "Number of frames to encode in each packet", OFFSET(frames_per_packet), AV_OPT_TYPE_INT, { 1 }, 1, 8, AE },
296 static const AVClass class = {
297 .class_name = "libspeex",
298 .item_name = av_default_item_name,
300 .version = LIBAVUTIL_VERSION_INT,
303 static const AVCodecDefault defaults[] = {
305 { "compression_level", "3" },
309 AVCodec ff_libspeex_encoder = {
311 .type = AVMEDIA_TYPE_AUDIO,
312 .id = CODEC_ID_SPEEX,
313 .priv_data_size = sizeof(LibSpeexEncContext),
315 .encode = encode_frame,
316 .close = encode_close,
317 .capabilities = CODEC_CAP_DELAY,
318 .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE },
319 .long_name = NULL_IF_CONFIG_SMALL("libspeex Speex"),
320 .priv_class = &class,
321 .defaults = defaults,