X-Git-Url: http://git.ieval.ro/?a=blobdiff_plain;f=src%2Fmain.c;h=49051307ebaf7631db909ebab4f3d504aa6403dc;hb=b5d0b6b1af5afcabe233e256fdda8783d2984cb7;hp=8521a33ebd9791b174356c8cf24386f43c6205ea;hpb=e1adc1783585421fdbd32d9a1d3edf5337e9af4f;p=fdkaac.git diff --git a/src/main.c b/src/main.c index 8521a33..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" @@ -292,7 +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': @@ -506,6 +503,11 @@ int write_sample(FILE *ofp, m4af_ctx_t *m4af, aacenc_frame_t *frame) return 0; } +static int do_smart_padding(int profile) +{ + return profile == 2 || profile == 5 || profile == 29; +} + static int encode(aacenc_param_ex_t *params, pcm_reader_t *reader, HANDLE_AACENCODER encoder, uint32_t frame_length, @@ -520,15 +522,12 @@ int encode(aacenc_param_ex_t *params, pcm_reader_t *reader, 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); for (;;) { - /* - * Since we delay the write, we cannot just exit loop when interrupted. - * Instead, we regard it as EOF. - */ if (g_interrupted) nread = 0; if (nread > 0) { @@ -551,7 +550,7 @@ int encode(aacenc_param_ex_t *params, pcm_reader_t *reader, remaining -= consumed; ip += consumed * fmt->channels_per_frame; - flip ^= 1; + if (is_padding) { /* * As we pad 1 frame at beginning and ending by our extrapolator, * we want to drop them. @@ -560,16 +559,26 @@ int encode(aacenc_param_ex_t *params, pcm_reader_t *reader, * Since sbr_header is included in the first frame (in case of * SBR), we cannot discard first frame. So we pick second instead. */ - ++encoded; - if (encoded == 1 || encoded == 3) - continue; - obp = &obuf[flip]; - if (write_sample(params->output_fp, m4af, obp) < 0) + flip ^= 1; + ++encoded; + if (encoded == 1 || encoded == 3) + continue; + } + if (write_sample(params->output_fp, m4af, &obuf[flip]) < 0) goto END; ++frames_written; } 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; @@ -693,16 +702,14 @@ 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; @@ -712,7 +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; @@ -720,7 +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; @@ -730,7 +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': @@ -740,18 +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: fprintf(stderr, "ERROR: unsupported input file\n"); - goto END; + goto FAIL; } } - if ((reader = pcm_open_sint16_converter(reader)) != 0) - reader = extrapolater_open(reader); + 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; -END: +FAIL: return 0; } @@ -767,10 +779,8 @@ int main(int argc, char **argv) pcm_reader_t *reader = 0; HANDLE_AACENCODER encoder = 0; AACENC_InfoStruct aacinfo = { 0 }; - LIB_INFO lib_info = { 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; @@ -786,15 +796,6 @@ 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. - */ 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 " @@ -802,10 +803,14 @@ int main(int argc, char **argv) params.sbr_ratio = 2; } scale_shift = aacenc_is_dual_rate_sbr((aacenc_param_t*)¶ms); - params.sbr_signaling = (params.transport_format == TT_MP4_LOAS) ? 2 : 0; - if (sbr_mode && !scale_shift) - params.sbr_signaling = 2; - + 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; @@ -825,45 +830,43 @@ int main(int argc, char **argv) if (!params.transport_format) { uint32_t scale; - uint8_t mp4asc[32]; - uint32_t ascsize = sizeof(mp4asc); unsigned framelen = aacinfo.frameLength; 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 >> scale_shift); + 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); } - if (scale_shift && (aacinfo.encoderDelay & 1)) { - /* - * Since odd delay cannot be exactly expressed in downsampled scale, - * we push one zero frame to the encoder here, to make delay even - */ - int16_t zero[8] = { 0 }; - aacenc_frame_t frame = { 0 }; - aac_encode_frame(encoder, sample_format, zero, 1, &frame); - free(frame.data); - } 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 << scale_shift; - if (scale_shift && (delay & 1)) - ++delay; padding = frame_count * aacinfo.frameLength - frames_read - delay; m4af_set_priming(m4af, 0, delay >> scale_shift, padding >> scale_shift); if (finalize_m4a(m4af, ¶ms, encoder) < 0)