X-Git-Url: http://git.ieval.ro/?a=blobdiff_plain;f=src%2Fmain.c;h=49051307ebaf7631db909ebab4f3d504aa6403dc;hb=b5d0b6b1af5afcabe233e256fdda8783d2984cb7;hp=0d076907631df6f8b76c5589350594120db7988e;hpb=6709cf694c2bdab475488397196520d9cc274dd8;p=fdkaac.git diff --git a/src/main.c b/src/main.c index 0d07690..4905130 100644 --- a/src/main.c +++ b/src/main.c @@ -33,8 +33,7 @@ #include #endif #include "compat.h" -#include "wav_reader.h" -#include "caf_reader.h" +#include "pcm_reader.h" #include "aacenc.h" #include "m4af.h" #include "progress.h" @@ -54,7 +53,8 @@ static void handle_signals(void) { int i, sigs[] = { SIGINT, SIGHUP, SIGTERM }; for (i = 0; i < sizeof(sigs)/sizeof(sigs[0]); ++i) { - struct sigaction sa = { 0 }; + struct sigaction sa; + memset(&sa, 0, sizeof sa); sa.sa_handler = signal_handler; sa.sa_flags |= SA_RESTART; sigaction(sigs[i], &sa, 0); @@ -117,9 +117,6 @@ PROGNAME " %s\n" " 29: MPEG-4 HE-AAC v2 (SBR+PS)\n" " 23: MPEG-4 AAC LD\n" " 39: MPEG-4 AAC ELD\n" -" 129: MPEG-2 AAC LC\n" -" 132: MPEG-2 HE-AAC (SBR)\n" -" 156: MPEG-2 HE-AAC v2 (SBR+PS)\n" " -b, --bitrate Bitrate in bits per seconds (for CBR)\n" " -m, --bitrate-mode Bitrate configuration\n" " 0: CBR (default)\n" @@ -132,7 +129,14 @@ PROGNAME " %s\n" " -a, --afterburner Afterburner\n" " 0: Off\n" " 1: On(default)\n" -" -L, --lowdelay-sbr Enable ELD-SBR (AAC ELD only)\n" +" -L, --lowdelay-sbr <-1|0|1> Configure SBR activity on AAC ELD\n" +" -1: Use ELD SBR auto configurator\n" +" 0: Disable SBR on ELD (default)\n" +" 1: Enable SBR on ELD\n" +" -s, --sbr-ratio <0|1|2> Controls activation of downsampled SBR\n" +" 0: Use lib default (default)\n" +" 1: downsampled SBR (default for ELD+SBR)\n" +" 2: dual-rate SBR (default for HE-AAC)\n" " -f, --transport-format Transport format\n" " 0: RAW (default, muxed into M4A)\n" " 1: ADIF\n" @@ -228,7 +232,7 @@ static int parse_options(int argc, char **argv, aacenc_param_ex_t *params) { int ch; - unsigned n; + int n; #define OPT_INCLUDE_SBR_DELAY M4AF_FOURCC('s','d','l','y') #define OPT_MOOV_BEFORE_MDAT M4AF_FOURCC('m','o','o','v') @@ -247,7 +251,8 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params) { "bitrate-mode", required_argument, 0, 'm' }, { "bandwidth", required_argument, 0, 'w' }, { "afterburner", required_argument, 0, 'a' }, - { "lowdelay-sbr", no_argument, 0, 'L' }, + { "lowdelay-sbr", required_argument, 0, 'L' }, + { "sbr-ratio", required_argument, 0, 's' }, { "transport-format", required_argument, 0, 'f' }, { "adts-crc-check", no_argument, 0, 'C' }, { "header-period", required_argument, 0, 'P' }, @@ -284,7 +289,7 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params) params->afterburner = 1; aacenc_getmainargs(&argc, &argv); - while ((ch = getopt_long(argc, argv, "hp:b:m:w:a:Ls:f:CP:G:Io:SR", + while ((ch = getopt_long(argc, argv, "hp:b:m:w:a:L:s:f:CP:G:Io:SR", long_options, 0)) != EOF) { switch (ch) { case 'h': @@ -325,7 +330,18 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params) params->afterburner = n; break; case 'L': - params->lowdelay_sbr = 1; + if (sscanf(optarg, "%d", &n) != 1 || n < -1 || n > 1) { + fprintf(stderr, "invalid arg for lowdelay-sbr\n"); + return -1; + } + params->lowdelay_sbr = n; + break; + case 's': + if (sscanf(optarg, "%u", &n) != 1 || n > 2) { + fprintf(stderr, "invalid arg for sbr-ratio\n"); + return -1; + } + params->sbr_ratio = n; break; case 'f': if (sscanf(optarg, "%u", &n) != 1) { @@ -472,68 +488,104 @@ int parse_options(int argc, char **argv, aacenc_param_ex_t *params) }; static -int write_sample(FILE *ofp, m4af_ctx_t *m4af, - const void *data, uint32_t size, uint32_t duration) +int write_sample(FILE *ofp, m4af_ctx_t *m4af, aacenc_frame_t *frame) { if (!m4af) { - fwrite(data, 1, size, ofp); + fwrite(frame->data, 1, frame->size, ofp); if (ferror(ofp)) { fprintf(stderr, "ERROR: fwrite(): %s\n", strerror(errno)); return -1; } - } else if (m4af_write_sample(m4af, 0, data, size, duration) < 0) { + } else if (m4af_write_sample(m4af, 0, frame->data, frame->size, 0) < 0) { fprintf(stderr, "ERROR: failed to write m4a sample\n"); return -1; } return 0; } +static int do_smart_padding(int profile) +{ + return profile == 2 || profile == 5 || profile == 29; +} + static -int encode(pcm_reader_t *reader, HANDLE_AACENCODER encoder, - uint32_t frame_length, FILE *ofp, m4af_ctx_t *m4af, - int show_progress) +int encode(aacenc_param_ex_t *params, pcm_reader_t *reader, + HANDLE_AACENCODER encoder, uint32_t frame_length, + m4af_ctx_t *m4af) { - int16_t *ibuf = 0; - uint8_t *obuf = 0; - uint32_t olen; - uint32_t osize = 0; + int16_t *ibuf = 0, *ip; + aacenc_frame_t obuf[2] = {{ 0 }}, *obp; + unsigned flip = 0; int nread = 1; - int consumed; int rc = -1; - int frames_written = 0; + int remaining, consumed; + int frames_written = 0, encoded = 0; aacenc_progress_t progress = { 0 }; const pcm_sample_description_t *fmt = pcm_get_format(reader); + const int is_padding = do_smart_padding(params->profile); ibuf = malloc(frame_length * fmt->bytes_per_frame); aacenc_progress_init(&progress, pcm_get_length(reader), fmt->sample_rate); - do { + + for (;;) { if (g_interrupted) nread = 0; - else if (nread) { + if (nread > 0) { if ((nread = pcm_read_frames(reader, ibuf, frame_length)) < 0) { fprintf(stderr, "ERROR: read failed\n"); goto END; } - if (show_progress) + if (!params->silent) aacenc_progress_update(&progress, pcm_get_position(reader), fmt->sample_rate * 2); } - if ((consumed = aac_encode_frame(encoder, fmt, ibuf, nread, - &obuf, &olen, &osize)) < 0) - goto END; - if (olen > 0) { - if (write_sample(ofp, m4af, obuf, olen, frame_length) < 0) + ip = ibuf; + remaining = nread; + do { + obp = &obuf[flip]; + consumed = aac_encode_frame(encoder, fmt, ip, remaining, obp); + if (consumed < 0) goto END; + if (consumed == 0 && obp->size == 0) goto DONE; + if (obp->size == 0) break; + + remaining -= consumed; + ip += consumed * fmt->channels_per_frame; + if (is_padding) { + /* + * As we pad 1 frame at beginning and ending by our extrapolator, + * we want to drop them. + * We delay output by 1 frame by double buffering, and discard + * second frame and final frame from the encoder. + * Since sbr_header is included in the first frame (in case of + * SBR), we cannot discard first frame. So we pick second instead. + */ + flip ^= 1; + ++encoded; + if (encoded == 1 || encoded == 3) + continue; + } + if (write_sample(params->output_fp, m4af, &obuf[flip]) < 0) goto END; ++frames_written; - } - } while (nread > 0 || olen > 0); - - if (show_progress) + } while (remaining > 0); + } +DONE: + /* + * When interrupted, we haven't pulled out last extrapolated frames + * from the reader. Therefore, we have to write the final outcome. + */ + if (g_interrupted) { + if (write_sample(params->output_fp, m4af, &obp[flip^1]) < 0) + goto END; + ++frames_written; + } + if (!params->silent) aacenc_progress_finish(&progress, pcm_get_position(reader)); rc = frames_written; END: if (ibuf) free(ibuf); - if (obuf) free(obuf); + if (obuf[0].data) free(obuf[0].data); + if (obuf[1].data) free(obuf[1].data); return rc; } @@ -543,19 +595,11 @@ void put_tool_tag(m4af_ctx_t *m4af, const aacenc_param_ex_t *params, { char tool_info[256]; char *p = tool_info; - LIB_INFO *lib_info = 0; + LIB_INFO lib_info; p += sprintf(p, PROGNAME " %s, ", fdkaac_version); - - lib_info = calloc(FDK_MODULE_LAST, sizeof(LIB_INFO)); - if (aacEncGetLibInfo(lib_info) == AACENC_OK) { - int i; - for (i = 0; i < FDK_MODULE_LAST; ++i) - if (lib_info[i].module_id == FDK_AACENC) - break; - p += sprintf(p, "libfdk-aac %s, ", lib_info[i].versionStr); - } - free(lib_info); + aacenc_get_lib_info(&lib_info); + p += sprintf(p, "libfdk-aac %s, ", lib_info.versionStr); if (params->bitrate_mode) sprintf(p, "VBR mode %d", params->bitrate_mode); else @@ -651,23 +695,21 @@ int parse_raw_spec(const char *spec, pcm_sample_description_t *desc) static pcm_io_vtbl_t pcm_io_vtbl = { read_callback, seek_callback, tell_callback }; -static pcm_io_vtbl_t pcm_io_vtbl_noseek = { read_callback, 0, 0 }; +static pcm_io_vtbl_t pcm_io_vtbl_noseek = { read_callback, 0, tell_callback }; static pcm_reader_t *open_input(aacenc_param_ex_t *params) { pcm_io_context_t io = { 0 }; pcm_reader_t *reader = 0; - struct stat stb = { 0 }; if ((params->input_fp = aacenc_fopen(params->input_filename, "rb")) == 0) { aacenc_fprintf(stderr, "ERROR: %s: %s\n", params->input_filename, strerror(errno)); - goto END; + goto FAIL; } io.cookie = params->input_fp; - if (fstat(fileno(params->input_fp), &stb) == 0 - && (stb.st_mode & S_IFMT) == S_IFREG) + if (aacenc_seekable(params->input_fp)) io.vtbl = &pcm_io_vtbl; else io.vtbl = &pcm_io_vtbl_noseek; @@ -677,7 +719,7 @@ pcm_reader_t *open_input(aacenc_param_ex_t *params) pcm_sample_description_t desc = { 0 }; if (parse_raw_spec(params->raw_format, &desc) < 0) { fprintf(stderr, "ERROR: invalid raw-format spec\n"); - goto END; + goto FAIL; } desc.sample_rate = params->raw_rate; desc.channels_per_frame = params->raw_channels; @@ -685,7 +727,7 @@ pcm_reader_t *open_input(aacenc_param_ex_t *params) desc.bytes_per_frame = params->raw_channels * bytes_per_channel; if ((reader = raw_open(&io, &desc)) == 0) { fprintf(stderr, "ERROR: failed to open raw input\n"); - goto END; + goto FAIL; } } else { int c; @@ -695,7 +737,7 @@ pcm_reader_t *open_input(aacenc_param_ex_t *params) case 'R': if ((reader = wav_open(&io, params->ignore_length)) == 0) { fprintf(stderr, "ERROR: broken / unsupported input file\n"); - goto END; + goto FAIL; } break; case 'c': @@ -705,15 +747,23 @@ pcm_reader_t *open_input(aacenc_param_ex_t *params) aacenc_translate_generic_text_tag, ¶ms->source_tag_ctx)) == 0) { fprintf(stderr, "ERROR: broken / unsupported input file\n"); - goto END; + goto FAIL; } break; default: - goto END; + fprintf(stderr, "ERROR: unsupported input file\n"); + goto FAIL; } } - return pcm_open_sint16_converter(reader); -END: + reader = pcm_open_native_converter(reader); + if (reader && PCM_IS_FLOAT(pcm_get_format(reader))) + reader = limiter_open(reader); + if (reader && (reader = pcm_open_sint16_converter(reader)) != 0) { + if (do_smart_padding(params->profile)) + reader = extrapolater_open(reader); + } + return reader; +FAIL: return 0; } @@ -731,9 +781,9 @@ int main(int argc, char **argv) AACENC_InfoStruct aacinfo = { 0 }; m4af_ctx_t *m4af = 0; const pcm_sample_description_t *sample_format; - int downsampled_timescale = 0; int frame_count = 0; int sbr_mode = 0; + unsigned scale_shift = 0; setlocale(LC_CTYPE, ""); setbuf(stderr, 0); @@ -746,17 +796,21 @@ int main(int argc, char **argv) sample_format = pcm_get_format(reader); - /* - * We use explicit/hierarchical signaling for LOAS. - * Other than that, we request implicit signaling to FDK library, then - * append explicit/backward-compatible signaling to ASC in case of MP4FF. - * - * Explicit/backward-compatible signaling of SBR is the most recommended - * way in MPEG4 part3 spec, and seems the only way supported by iTunes. - * Since FDK library does not support it, we have to do it on our side. - */ - params.sbr_signaling = (params.transport_format == TT_MP4_LOAS) ? 2 : 0; - + sbr_mode = aacenc_is_sbr_active((aacenc_param_t*)¶ms); + if (sbr_mode && !aacenc_is_sbr_ratio_available()) { + fprintf(stderr, "WARNING: Only dual-rate SBR is available " + "for this version\n"); + params.sbr_ratio = 2; + } + scale_shift = aacenc_is_dual_rate_sbr((aacenc_param_t*)¶ms); + params.sbr_signaling = 0; + if (sbr_mode) { + if (params.transport_format == TT_MP4_LOAS || !scale_shift) + params.sbr_signaling = 2; + if (params.transport_format == TT_MP4_RAW && + aacenc_is_explicit_bw_compatible_sbr_signaling_available()) + params.sbr_signaling = 1; + } if (aacenc_init(&encoder, (aacenc_param_t*)¶ms, sample_format, &aacinfo) < 0) goto END; @@ -773,44 +827,48 @@ int main(int argc, char **argv) goto END; } handle_signals(); - sbr_mode = aacenc_is_sbr_active((aacenc_param_t*)¶ms); + if (!params.transport_format) { uint32_t scale; - uint8_t mp4asc[32]; - uint32_t ascsize = sizeof(mp4asc); unsigned framelen = aacinfo.frameLength; - if (sbr_mode) - downsampled_timescale = 1; - scale = sample_format->sample_rate >> downsampled_timescale; + scale = sample_format->sample_rate >> scale_shift; if ((m4af = m4af_create(M4AF_CODEC_MP4A, scale, &m4af_io, params.output_fp)) < 0) goto END; - aacenc_mp4asc((aacenc_param_t*)¶ms, aacinfo.confBuf, - aacinfo.confSize, mp4asc, &ascsize); - m4af_set_decoder_specific_info(m4af, 0, mp4asc, ascsize); - m4af_set_fixed_frame_duration(m4af, 0, - framelen >> downsampled_timescale); + m4af_set_num_channels(m4af, 0, sample_format->channels_per_frame); + m4af_set_fixed_frame_duration(m4af, 0, framelen >> scale_shift); + if (aacenc_is_explicit_bw_compatible_sbr_signaling_available()) + m4af_set_decoder_specific_info(m4af, 0, + aacinfo.confBuf, aacinfo.confSize); + else { + uint8_t mp4asc[32]; + uint32_t ascsize = sizeof(mp4asc); + aacenc_mp4asc((aacenc_param_t*)¶ms, aacinfo.confBuf, + aacinfo.confSize, mp4asc, &ascsize); + m4af_set_decoder_specific_info(m4af, 0, mp4asc, ascsize); + } m4af_set_vbr_mode(m4af, 0, params.bitrate_mode); m4af_set_priming_mode(m4af, params.gapless_mode + 1); m4af_begin_write(m4af); } - frame_count = encode(reader, encoder, aacinfo.frameLength, - params.output_fp, m4af, !params.silent); + frame_count = encode(¶ms, reader, encoder, aacinfo.frameLength, m4af); if (frame_count < 0) goto END; if (m4af) { - uint32_t delay = aacinfo.encoderDelay; uint32_t padding; +#if AACENCODER_LIB_VL0 < 4 + uint32_t delay = aacinfo.encoderDelay; + if (sbr_mode && params.profile != AOT_ER_AAC_ELD + && !params.include_sbr_delay) + delay -= 481 << scale_shift; +#else + uint32_t delay = params.include_sbr_delay ? aacinfo.nDelay + : aacinfo.nDelayCore; +#endif int64_t frames_read = pcm_get_position(reader); - if (sbr_mode && params.profile != AOT_ER_AAC_ELD && - !params.include_sbr_delay) - delay -= 481 << 1; - if (sbr_mode && (delay & 1)) - ++delay; padding = frame_count * aacinfo.frameLength - frames_read - delay; - m4af_set_priming(m4af, 0, delay >> downsampled_timescale, - padding >> downsampled_timescale); + m4af_set_priming(m4af, 0, delay >> scale_shift, padding >> scale_shift); if (finalize_m4a(m4af, ¶ms, encoder) < 0) goto END; }