X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Flibspeexenc.c;h=cbfa5be92b9ae82be1f99ec95e0d887871edfcf8;hb=e00eb03cd8bfb6993d33c30ccd560947b5f6bad5;hp=290718c5f124704ff5345479f3ffe0590c6e16a7;hpb=9785082c15360303510059a79b61c88c489548c5;p=ffmpeg diff --git a/libavcodec/libspeexenc.c b/libavcodec/libspeexenc.c index 290718c5f12..cbfa5be92b9 100644 --- a/libavcodec/libspeexenc.c +++ b/libavcodec/libspeexenc.c @@ -62,15 +62,35 @@ * sometimes desirable to use multiple frames-per-packet to reduce the * amount of container overhead. This can be done by setting the * 'frames_per_packet' option to a value 1 to 8. + * + * + * Optional features + * Speex encoder supports several optional features, which can be useful + * for some conditions. + * + * Voice Activity Detection + * When enabled, voice activity detection detects whether the audio + * being encoded is speech or silence/background noise. VAD is always + * implicitly activated when encoding in VBR, so the option is only useful + * in non-VBR operation. In this case, Speex detects non-speech periods and + * encodes them with just enough bits to reproduce the background noise. + * + * Discontinuous Transmission (DTX) + * DTX is an addition to VAD/VBR operation, that allows to stop transmitting + * completely when the background noise is stationary. + * In file-based operation only 5 bits are used for such frames. */ #include #include #include -#include "libavutil/mathematics.h" + +#include "libavutil/audioconvert.h" +#include "libavutil/common.h" #include "libavutil/opt.h" #include "avcodec.h" #include "internal.h" +#include "audio_frame_queue.h" typedef struct { AVClass *class; ///< AVClass for private options @@ -81,10 +101,10 @@ typedef struct { float vbr_quality; ///< VBR quality 0.0 to 10.0 int cbr_quality; ///< CBR quality 0 to 10 int abr; ///< flag to enable ABR + int vad; ///< flag to enable VAD + int dtx; ///< flag to enable DTX int pkt_frame_count; ///< frame count for the current packet - int lookahead; ///< encoder delay - int64_t next_pts; ///< next pts, in sample_rate time base - int pkt_sample_count; ///< sample count in the current packet + AudioFrameQueue afq; ///< frame queue } LibSpeexEncContext; static av_cold void print_enc_params(AVCodecContext *avctx, @@ -117,6 +137,8 @@ static av_cold void print_enc_params(AVCodecContext *avctx, s->frames_per_packet); av_log(avctx, AV_LOG_DEBUG, "packet size: %d\n", avctx->frame_size * s->frames_per_packet); + av_log(avctx, AV_LOG_DEBUG, "voice activity detection: %d\n", s->vad); + av_log(avctx, AV_LOG_DEBUG, "discontinuous transmission: %d\n", s->dtx); } static av_cold int encode_init(AVCodecContext *avctx) @@ -157,6 +179,7 @@ static av_cold int encode_init(AVCodecContext *avctx) if (avctx->flags & CODEC_FLAG_QSCALE) { /* VBR */ s->header.vbr = 1; + s->vad = 1; /* VAD is always implicitly activated for VBR */ speex_encoder_ctl(s->enc_state, SPEEX_SET_VBR, &s->header.vbr); s->vbr_quality = av_clipf(avctx->global_quality / (float)FF_QP2LAMBDA, 0.0f, 10.0f); @@ -188,6 +211,17 @@ static av_cold int encode_init(AVCodecContext *avctx) avctx->bit_rate = s->header.bitrate + (avctx->channels == 2 ? 800 : 0); } + /* VAD is activated with VBR or can be turned on by itself */ + if (s->vad) + speex_encoder_ctl(s->enc_state, SPEEX_SET_VAD, &s->vad); + + /* Activiting Discontinuous Transmission */ + if (s->dtx) { + speex_encoder_ctl(s->enc_state, SPEEX_SET_DTX, &s->dtx); + if (!(s->abr || s->vad || s->header.vbr)) + av_log(avctx, AV_LOG_WARNING, "DTX is not much of use without ABR, VAD or VBR\n"); + } + /* set encoding complexity */ if (avctx->compression_level > FF_COMPRESSION_DEFAULT) { complexity = av_clip(avctx->compression_level, 0, 10); @@ -201,8 +235,8 @@ static av_cold int encode_init(AVCodecContext *avctx) s->header.frames_per_packet = s->frames_per_packet; /* set encoding delay */ - speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &s->lookahead); - s->next_pts = -s->lookahead; + speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &avctx->delay); + ff_af_queue_init(avctx, &s->afq); /* create header packet bytes from header struct */ /* note: libspeex allocates the memory for header_data, which is freed @@ -211,13 +245,22 @@ static av_cold int encode_init(AVCodecContext *avctx) /* allocate extradata and coded_frame */ avctx->extradata = av_malloc(header_size + FF_INPUT_BUFFER_PADDING_SIZE); + if (!avctx->extradata) { + speex_header_free(header_data); + speex_encoder_destroy(s->enc_state); + av_log(avctx, AV_LOG_ERROR, "memory allocation error\n"); + return AVERROR(ENOMEM); + } +#if FF_API_OLD_ENCODE_AUDIO avctx->coded_frame = avcodec_alloc_frame(); - if (!avctx->extradata || !avctx->coded_frame) { + if (!avctx->coded_frame) { + av_freep(&avctx->extradata); speex_header_free(header_data); speex_encoder_destroy(s->enc_state); av_log(avctx, AV_LOG_ERROR, "memory allocation error\n"); return AVERROR(ENOMEM); } +#endif /* copy header packet to extradata */ memcpy(avctx->extradata, header_data, header_size); @@ -231,19 +274,21 @@ static av_cold int encode_init(AVCodecContext *avctx) return 0; } -static int encode_frame(AVCodecContext *avctx, uint8_t *frame, int buf_size, - void *data) +static int encode_frame(AVCodecContext *avctx, AVPacket *avpkt, + const AVFrame *frame, int *got_packet_ptr) { LibSpeexEncContext *s = avctx->priv_data; - int16_t *samples = data; + int16_t *samples = frame ? (int16_t *)frame->data[0] : NULL; + int ret; - if (data) { + if (samples) { /* encode Speex frame */ if (avctx->channels == 2) speex_encode_stereo_int(samples, s->header.frame_size, &s->bits); speex_encode_int(s->enc_state, samples, &s->bits); s->pkt_frame_count++; - s->pkt_sample_count += avctx->frame_size; + if ((ret = ff_af_queue_add(&s->afq, frame) < 0)) + return ret; } else { /* handle end-of-stream */ if (!s->pkt_frame_count) @@ -258,19 +303,20 @@ static int encode_frame(AVCodecContext *avctx, uint8_t *frame, int buf_size, /* write output if all frames for the packet have been encoded */ if (s->pkt_frame_count == s->frames_per_packet) { s->pkt_frame_count = 0; - avctx->coded_frame->pts = - av_rescale_q(s->next_pts, (AVRational){ 1, avctx->sample_rate }, - avctx->time_base); - s->next_pts += s->pkt_sample_count; - s->pkt_sample_count = 0; - if (buf_size > speex_bits_nbytes(&s->bits)) { - int ret = speex_bits_write(&s->bits, frame, buf_size); - speex_bits_reset(&s->bits); + if ((ret = ff_alloc_packet(avpkt, speex_bits_nbytes(&s->bits)))) { + av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n"); return ret; - } else { - av_log(avctx, AV_LOG_ERROR, "output buffer too small"); - return AVERROR(EINVAL); } + ret = speex_bits_write(&s->bits, avpkt->data, avpkt->size); + speex_bits_reset(&s->bits); + + /* Get the next frame pts/duration */ + ff_af_queue_remove(&s->afq, s->frames_per_packet * avctx->frame_size, + &avpkt->pts, &avpkt->duration); + + avpkt->size = ret; + *got_packet_ptr = 1; + return 0; } return 0; } @@ -282,7 +328,10 @@ static av_cold int encode_close(AVCodecContext *avctx) speex_bits_destroy(&s->bits); speex_encoder_destroy(s->enc_state); + ff_af_queue_close(&s->afq); +#if FF_API_OLD_ENCODE_AUDIO av_freep(&avctx->coded_frame); +#endif av_freep(&avctx->extradata); return 0; @@ -291,9 +340,11 @@ static av_cold int encode_close(AVCodecContext *avctx) #define OFFSET(x) offsetof(LibSpeexEncContext, x) #define AE AV_OPT_FLAG_AUDIO_PARAM | AV_OPT_FLAG_ENCODING_PARAM static const AVOption options[] = { - { "abr", "Use average bit rate", OFFSET(abr), AV_OPT_TYPE_INT, { 0 }, 0, 1, AE }, - { "cbr_quality", "Set quality value (0 to 10) for CBR", OFFSET(cbr_quality), AV_OPT_TYPE_INT, { 8 }, 0, 10, AE }, - { "frames_per_packet", "Number of frames to encode in each packet", OFFSET(frames_per_packet), AV_OPT_TYPE_INT, { 1 }, 1, 8, AE }, + { "abr", "Use average bit rate", OFFSET(abr), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE }, + { "cbr_quality", "Set quality value (0 to 10) for CBR", OFFSET(cbr_quality), AV_OPT_TYPE_INT, { .i64 = 8 }, 0, 10, AE }, + { "frames_per_packet", "Number of frames to encode in each packet", OFFSET(frames_per_packet), AV_OPT_TYPE_INT, { .i64 = 1 }, 1, 8, AE }, + { "vad", "Voice Activity Detection", OFFSET(vad), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE }, + { "dtx", "Discontinuous Transmission", OFFSET(dtx), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AE }, { NULL }, }; @@ -313,13 +364,18 @@ static const AVCodecDefault defaults[] = { AVCodec ff_libspeex_encoder = { .name = "libspeex", .type = AVMEDIA_TYPE_AUDIO, - .id = CODEC_ID_SPEEX, + .id = AV_CODEC_ID_SPEEX, .priv_data_size = sizeof(LibSpeexEncContext), .init = encode_init, - .encode = encode_frame, + .encode2 = encode_frame, .close = encode_close, .capabilities = CODEC_CAP_DELAY, - .sample_fmts = (const enum SampleFormat[]){ AV_SAMPLE_FMT_S16, AV_SAMPLE_FMT_NONE }, + .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16, + AV_SAMPLE_FMT_NONE }, + .channel_layouts = (const uint64_t[]){ AV_CH_LAYOUT_MONO, + AV_CH_LAYOUT_STEREO, + 0 }, + .supported_samplerates = (const int[]){ 8000, 16000, 32000, 0 }, .long_name = NULL_IF_CONFIG_SMALL("libspeex Speex"), .priv_class = &class, .defaults = defaults,