More audio

.github/workflows/publish-pypi.yml

@@ -42,7 +42,7 @@ env:
 
     # nv-codec-headers (Linux/Windows only)
     if [[ "$EXTRA_FFMPEG_FLAGS" == *"--enable-nvdec"* ]]; then
-      git clone -b "$nv_version" --depth 1 https://git.videolan.org/git/ffmpeg/nv-codec-headers
+      git clone -b "$nv_version" --depth 1 https://github.com/FFmpeg/nv-codec-headers
       cd nv-codec-headers
       $SUDO make install PREFIX="$PREFIX"
       cd .. && rm -rf nv-codec-headers

AviSynth/README.md

@@ -158,7 +158,7 @@
 ###### LSMASHAudioSource
 
 * `LSMASHAudioSource(string source, int track = 0, bool skip_priming = true, string layout = "", int rate = 0,
-                    string decoder = "", int ff_loglevel = 0, float drc_scale = 1.0, string ff_options = "")`
+                    string decoder = "", int ff_loglevel = 0, float drc_scale = 1.0, string ff_options = "", bool skip_tail = false)`
 
         * This function uses libavcodec as audio decoder and L-SMASH as demuxer.
         [Arguments]
@@ -169,7 +169,7 @@
                 The value 0 means trying to get the first detected audio stream.
             + skip_priming (default : true)
                 Whether skip priming samples or not.
-                Priming samples is detected from iTunSMPB or the first non-empty edit.
+                Priming samples are detected from iTunSMPB or the first non-empty edit.
                 If any priming samples, do pre-roll whenever any seek of audio stream occurs.
             + layout (default : "")
                 Output audio channel layout.
@@ -257,8 +257,11 @@
                 0.0 < drc_scale <= 1.0 : DRC enabled. Applies a fraction of the stream DRC value. Audio reproduction is between full range and full compression.
                 > 1.0 : DRC enabled. Applies drc_scale asymmetrically. Loud sounds are fully compressed. Soft sounds are enhanced.
                 If `ff_options="drc_scale=x"` is used, `drc_scale` is ignored.
-            + ff_options (default: "")
+            + ff_options (default : "")
                 Same as 'ff_options' of LSMASHVideoSource().
+            * skip_tail (default : false)
+                Whether skip trailing samples or not.
+                Trailing samples are detected from iTunSMPB.
 
 ###### LWLibavVideoSource
 

AviSynth/libavsmash_source.cpp

@@ -252,32 +252,6 @@ uint32_t LSMASHAudioSource::open_file(const char* source, IScriptEnvironment* en
     return movie_param.number_of_tracks;
 }
 
-static int64_t get_start_time(lsmash_root_t* root, uint32_t track_ID)
-{
-    /* Consider start time of this media if any non-empty edit is present. */
-    uint32_t edit_count = lsmash_count_explicit_timeline_map(root, track_ID);
-    for (uint32_t edit_number = 1; edit_number <= edit_count; edit_number++) {
-        lsmash_edit_t edit;
-        if (lsmash_get_explicit_timeline_map(root, track_ID, edit_number, &edit))
-            return 0;
-        if (edit.duration == 0)
-            return 0; /* no edits */
-        if (edit.start_time >= 0)
-            return edit.start_time;
-    }
-    return 0;
-}
-
-static char* duplicate_as_string(void* src, size_t length)
-{
-    char* dst = new char[length + 1];
-    if (!dst)
-        return nullptr;
-    memcpy(dst, src, length);
-    dst[length] = '\0';
-    return dst;
-}
-
 void LSMASHAudioSource::get_audio_track(const char* source, uint32_t track_number, IScriptEnvironment* env)
 {
     libavsmash_audio_decode_handler_t* adhp = this->adhp.get();
@@ -289,70 +263,24 @@ void LSMASHAudioSource::get_audio_track(const char* source, uint32_t track_numbe
 }
 
 static void count_output_audio_samples(libavsmash_audio_decode_handler_t* adhp, libavsmash_audio_output_handler_t* aohp, bool skip_priming,
-    VideoInfo& vi, IScriptEnvironment* env)
+    bool skip_tail, VideoInfo& vi, IScriptEnvironment* env)
 {
-    lsmash_root_t* root = libavsmash_audio_get_root(adhp);
-    uint32_t track_id = libavsmash_audio_get_track_id(adhp);
-    uint64_t start_time = 0;
-    if (skip_priming) {
-        uint32_t media_timescale = libavsmash_audio_get_media_timescale(adhp);
-        uint32_t itunes_metadata_count = lsmash_count_itunes_metadata(root);
-        for (uint32_t i = 1; i <= itunes_metadata_count; i++) {
-            lsmash_itunes_metadata_t metadata;
-            if (lsmash_get_itunes_metadata(root, i, &metadata) < 0)
-                continue;
-            if (metadata.item != ITUNES_METADATA_ITEM_CUSTOM
-                || (metadata.type != ITUNES_METADATA_TYPE_STRING && metadata.type != ITUNES_METADATA_TYPE_BINARY) || !metadata.meaning
-                || !metadata.name || memcmp("com.apple.iTunes", metadata.meaning, strlen(metadata.meaning))
-                || memcmp("iTunSMPB", metadata.name, strlen(metadata.name))) {
-                lsmash_cleanup_itunes_metadata(&metadata);
-                continue;
-            }
-            char* value = nullptr;
-            if (metadata.type == ITUNES_METADATA_TYPE_STRING) {
-                size_t length = strlen(metadata.value.string);
-                if (length >= 116)
-                    value = duplicate_as_string(metadata.value.string, length);
-            } else /* metadata.type == ITUNES_METADATA_TYPE_BINARY */
-            {
-                if (metadata.value.binary.size >= 116)
-                    value = duplicate_as_string(metadata.value.binary.data, metadata.value.binary.size);
-            }
-            lsmash_cleanup_itunes_metadata(&metadata);
-            if (!value)
-                continue;
-            uint32_t dummy[9];
-            uint32_t priming_samples;
-            uint32_t padding;
-            uint64_t duration;
-            if (12
-                != sscanf(value, " %x %x %x %" SCNx64 " %x %x %x %x %x %x %x %x", &dummy[0], &priming_samples, &padding, &duration,
-                    &dummy[1], &dummy[2], &dummy[3], &dummy[4], &dummy[5], &dummy[6], &dummy[7], &dummy[8])) {
-                delete[] value;
-                continue;
-            }
-            delete[] value;
-            libavsmash_audio_set_implicit_preroll(adhp);
-            start_time = av_rescale(priming_samples, media_timescale, aohp->output_sample_rate);
-            aohp->skip_decoded_samples = priming_samples;
-            // vi.num_audio_samples = duration + priming_samples;
-            break;
-        }
-        if (aohp->skip_decoded_samples == 0) {
-            uint32_t ctd_shift;
-            if (lsmash_get_composition_to_decode_shift_from_media_timeline(root, track_id, &ctd_shift))
-                env->ThrowError("LSMASHAudioSource: failed to get the timeline shift.");
-            start_time = ctd_shift + get_start_time(root, track_id);
-            aohp->skip_decoded_samples = av_rescale(start_time, aohp->output_sample_rate, media_timescale);
-        }
+    uint64_t final_num_samples = 0;
+    char error_msg[256] = { 0 };
+    if (libavsmash_audio_setup_sample_count(
+            adhp, aohp, skip_priming ? 1 : 0, skip_tail ? 1 : 0, &final_num_samples, error_msg, sizeof(error_msg))
+        < 0) {
+        env->ThrowError("LSMASHAudioSource: %s", error_msg);
+    }
+    if (final_num_samples > (uint64_t)INT64_MAX) {
+        env->ThrowError("LSMASHAudioSource: audio sample count exceeds INT64_MAX.");
     }
-    vi.num_audio_samples = libavsmash_audio_count_overall_pcm_samples(adhp, aohp->output_sample_rate, start_time);
-    if (vi.num_audio_samples == 0)
-        env->ThrowError("LSMASHAudioSource: no valid audio frame.");
+    vi.num_audio_samples = (int64_t)final_num_samples;
 }
 
 static void prepare_audio_decoding(libavsmash_audio_decode_handler_t* adhp, libavsmash_audio_output_handler_t* aohp,
-    AVFormatContext* format_ctx, const char* channel_layout, int sample_rate, bool skip_priming, VideoInfo& vi, IScriptEnvironment* env)
+    AVFormatContext* format_ctx, const char* channel_layout, int sample_rate, bool skip_priming, bool skip_tail, VideoInfo& vi,
+    IScriptEnvironment* env)
 {
     /* Initialize the audio decoder configuration. */
     if (libavsmash_audio_initialize_decoder_configuration(adhp, format_ctx, 0) < 0)
@@ -363,13 +291,13 @@ static void prepare_audio_decoding(libavsmash_audio_decode_handler_t* adhp, liba
     aohp->output_bits_per_sample = libavsmash_audio_get_best_used_bits_per_sample(adhp);
     AVCodecContext* ctx = libavsmash_audio_get_codec_context(adhp);
     as_setup_audio_rendering(aohp, ctx, &vi, env, "LSMASHAudioSource", channel_layout, sample_rate);
-    count_output_audio_samples(adhp, aohp, skip_priming, vi, env);
+    count_output_audio_samples(adhp, aohp, skip_priming, skip_tail, vi, env);
     /* Force seeking at the first reading. */
     libavsmash_audio_force_seek(adhp);
 }
 
 LSMASHAudioSource::LSMASHAudioSource(const char* source, uint32_t track_number, bool skip_priming, const char* channel_layout,
-    int sample_rate, const char* preferred_decoder_names, const double drc, const char* ff_options, IScriptEnvironment* env)
+    int sample_rate, const char* preferred_decoder_names, const double drc, const char* ff_options, bool skip_tail, IScriptEnvironment* env)
     : LSMASHAudioSource {}
 {
     memset(&vi, 0, sizeof(VideoInfo));
@@ -380,7 +308,7 @@ LSMASHAudioSource::LSMASHAudioSource(const char* source, uint32_t track_number,
     libavsmash_audio_set_drc(adhp, drc);
     libavsmash_audio_set_decoder_options(adhp, ff_options);
     get_audio_track(source, track_number, env);
-    prepare_audio_decoding(adhp, aohp, format_ctx.get(), channel_layout, sample_rate, skip_priming, vi, env);
+    prepare_audio_decoding(adhp, aohp, format_ctx.get(), channel_layout, sample_rate, skip_priming, skip_tail, vi, env);
     lsmash_discard_boxes(libavsmash_audio_get_root(adhp));
 }
 
@@ -468,7 +396,8 @@ AVSValue __cdecl CreateLSMASHAudioSource(AVSValue args, void* user_data, IScript
     int ff_loglevel = args[6].AsInt(0);
     const double drc = args[7].AsFloat(-1.0);
     const char* ff_options = args[8].AsString(nullptr);
+    bool skip_tail = args[9].AsBool(false);
     set_av_log_level(ff_loglevel);
     return new LSMASHAudioSource(
-        source, track_number, skip_priming, layout_string, sample_rate, preferred_decoder_names, drc, ff_options, env);
+        source, track_number, skip_priming, layout_string, sample_rate, preferred_decoder_names, drc, ff_options, skip_tail, env);
 }

AviSynth/libavsmash_source.h

@@ -90,7 +90,7 @@ class LSMASHAudioSource : public LibavSMASHSource {
 
 public:
     LSMASHAudioSource(const char* source, uint32_t track_number, bool skip_priming, const char* channel_layout, int sample_rate,
-        const char* preferred_decoder_names, const double drc, const char* ff_options, IScriptEnvironment* env);
+        const char* preferred_decoder_names, const double drc, const char* ff_options, bool skip_tail, IScriptEnvironment* env);
     ~LSMASHAudioSource();
     PVideoFrame __stdcall GetFrame(int n, IScriptEnvironment* env)
     {

AviSynth/lsmashsource.cpp

@@ -57,7 +57,7 @@ extern "C" AVS_EXPORT const char* __stdcall AvisynthPluginInit3(IScriptEnvironme
         CreateLSMASHVideoSource, 0);
     /* LSMASHAudioSource */
     env->AddFunction("LSMASHAudioSource",
-        "[source]s[track]i[skip_priming]b[layout]s[rate]i[decoder]s[ff_loglevel]i[drc_scale]f[ff_options]s", CreateLSMASHAudioSource, 0);
+        "[source]s[track]i[skip_priming]b[layout]s[rate]i[decoder]s[ff_loglevel]i[drc_scale]f[ff_options]s[skip_tail]b", CreateLSMASHAudioSource, 0);
     /* LWLibavVideoSource */
     env->AddFunction("LWLibavVideoSource",
         "[source]s[stream_index]i[threads]i[cache]b[cachefile]s[seek_mode]i[seek_threshold]i[dr]b[fpsnum]i[fpsden]i[repeat]b[dominance]i["

AviSynth/lwlibav_source.cpp

@@ -324,7 +324,7 @@ LWLibavAudioSource::LWLibavAudioSource(lwlibav_option_t* opt, const char* channe
             sequence_pcm_count += info[i].length;
             const uint64_t frame_end = prior_sequences_resampled_count
                 + count_sequence_output_pcm_samples(sequence_pcm_count, current_sample_rate, aohp->output_sample_rate);
-            if (info[i].file_offset == -1)
+            if (lw_audio_is_gap_offset(info[i].file_offset))
                 gap_info_list.emplace_back(static_cast<int64_t>(frame_start) + lwh.av_gap, static_cast<int>(frame_end - frame_start));
         }
         const int gap_count = gap_info_list.size();

VapourSynth/libavsmash_audio_source.c

@@ -9,8 +9,8 @@
 
 #include <libavformat/avformat.h>
 #include <libavutil/log.h>
-#include <libavutil/mem.h>
 #include <libavutil/mathematics.h>
+#include <libavutil/mem.h>
 
 #include "audio_output.h"
 #include "lsmashsource.h"
@@ -63,86 +63,23 @@ static lsmas_audio_handler_t* alloc_handler(void)
     return hp;
 }
 
-static int64_t get_start_time(lsmash_root_t* root, uint32_t track_id)
-{
-    uint32_t edit_count = lsmash_count_explicit_timeline_map(root, track_id);
-    for (uint32_t edit_number = 1; edit_number <= edit_count; edit_number++) {
-        lsmash_edit_t edit;
-        if (lsmash_get_explicit_timeline_map(root, track_id, edit_number, &edit) || edit.duration == 0)
-            return 0;
-        if (edit.start_time >= 0)
-            return edit.start_time;
-    }
-    return 0;
-}
-
-static char* duplicate_string(const void* src, size_t length)
-{
-    char* dst = (char*)malloc(length + 1);
-    if (!dst)
-        return NULL;
-    memcpy(dst, src, length);
-    dst[length] = '\0';
-    return dst;
-}
-
-static int count_output_audio_samples(lsmas_audio_handler_t* hp, int skip_priming, VSMap* out, const VSAPI* vsapi)
+static int count_output_audio_samples(lsmas_audio_handler_t* hp, int skip_priming, int skip_tail, VSMap* out, const VSAPI* vsapi)
 {
-    libavsmash_audio_decode_handler_t* adhp = hp->adhp;
-    libavsmash_audio_output_handler_t* aohp = hp->aohp;
-    lsmash_root_t* root = libavsmash_audio_get_root(adhp);
-    uint32_t track_id = libavsmash_audio_get_track_id(adhp);
-    uint64_t start_time = 0;
-    if (skip_priming) {
-        uint32_t media_timescale = libavsmash_audio_get_media_timescale(adhp);
-        uint32_t metadata_count = lsmash_count_itunes_metadata(root);
-        for (uint32_t i = 1; i <= metadata_count; i++) {
-            lsmash_itunes_metadata_t metadata;
-            if (lsmash_get_itunes_metadata(root, i, &metadata) < 0)
-                continue;
-            int matches = metadata.item == ITUNES_METADATA_ITEM_CUSTOM
-                && (metadata.type == ITUNES_METADATA_TYPE_STRING || metadata.type == ITUNES_METADATA_TYPE_BINARY) && metadata.meaning
-                && metadata.name && !strcmp(metadata.meaning, "com.apple.iTunes") && !strcmp(metadata.name, "iTunSMPB");
-            char* value = NULL;
-            if (matches && metadata.type == ITUNES_METADATA_TYPE_STRING) {
-                size_t length = strlen(metadata.value.string);
-                if (length >= 116)
-                    value = duplicate_string(metadata.value.string, length);
-            } else if (matches && metadata.value.binary.size >= 116)
-                value = duplicate_string(metadata.value.binary.data, metadata.value.binary.size);
-            lsmash_cleanup_itunes_metadata(&metadata);
-            if (!value)
-                continue;
-            uint32_t dummy[9];
-            uint32_t priming_samples;
-            uint32_t padding;
-            uint64_t duration;
-            int fields = sscanf(value, " %x %x %x %" SCNx64 " %x %x %x %x %x %x %x %x", &dummy[0], &priming_samples, &padding,
-                &duration, &dummy[1], &dummy[2], &dummy[3], &dummy[4], &dummy[5], &dummy[6], &dummy[7], &dummy[8]);
-            lw_free(value);
-            if (fields != 12)
-                continue;
-            libavsmash_audio_set_implicit_preroll(adhp);
-            start_time = av_rescale(priming_samples, media_timescale, aohp->output_sample_rate);
-            aohp->skip_decoded_samples = priming_samples;
-            break;
-        }
-        if (aohp->skip_decoded_samples == 0) {
-            uint32_t ctd_shift;
-            if (lsmash_get_composition_to_decode_shift_from_media_timeline(root, track_id, &ctd_shift)) {
-                set_error_on_init(out, vsapi, "lsmas: failed to get the audio timeline shift.");
-                return -1;
-            }
-            start_time = ctd_shift + get_start_time(root, track_id);
-            aohp->skip_decoded_samples = av_rescale(start_time, aohp->output_sample_rate, media_timescale);
-        }
+    uint64_t final_num_samples = 0;
+    char error_msg[256] = { 0 };
+    if (libavsmash_audio_setup_sample_count(hp->adhp, hp->aohp, skip_priming, skip_tail, &final_num_samples, error_msg, sizeof(error_msg))
+        < 0) {
+        char vs_msg[300];
+        snprintf(vs_msg, sizeof(vs_msg), "lsmas: %s", error_msg);
+        set_error_on_init(out, vsapi, vs_msg);
+        return -1;
     }
-    hp->ai.numSamples = libavsmash_audio_count_overall_pcm_samples(adhp, aohp->output_sample_rate, start_time);
-    if (hp->ai.numSamples <= 0) {
-        set_error_on_init(out, vsapi, "lsmas: no valid audio frame.");
+    if (final_num_samples > (uint64_t)INT64_MAX) {
+        set_error_on_init(out, vsapi, "lsmas: audio sample count exceeds INT64_MAX.");
         return -1;
     }
-    hp->ai.numFrames = (int)((hp->ai.numSamples + VS_AUDIO_FRAME_SAMPLES - 1) / VS_AUDIO_FRAME_SAMPLES);
+    hp->ai.numSamples = (int64_t)final_num_samples;
+    hp->ai.numFrames = (int)((final_num_samples + VS_AUDIO_FRAME_SAMPLES - 1) / VS_AUDIO_FRAME_SAMPLES);
     return 0;
 }
 
@@ -208,6 +145,7 @@ void VS_CC vs_libavsmashaudiosource_create(const VSMap* in, VSMap* out, void* us
     const char* layout;
     const char* decoder;
     const char* ff_options;
+    int64_t skip_tail;
     int error;
     double drc = vsapi->mapGetFloat(in, "drc_scale", 0, &error);
     if (error)
@@ -219,6 +157,7 @@ void VS_CC vs_libavsmashaudiosource_create(const VSMap* in, VSMap* out, void* us
     set_option_string(&layout, NULL, "layout", in, vsapi);
     set_option_string(&decoder, NULL, "decoder", in, vsapi);
     set_option_string(&ff_options, NULL, "ff_options", in, vsapi);
+    set_option_int64(&skip_tail, 0, "skip_tail", in, vsapi);
     set_av_log_level(ff_loglevel);
 
     lsmas_audio_handler_t* hp = alloc_handler();
@@ -261,7 +200,7 @@ void VS_CC vs_libavsmashaudiosource_create(const VSMap* in, VSMap* out, void* us
     hp->aohp->output_bits_per_sample = libavsmash_audio_get_best_used_bits_per_sample(hp->adhp);
     AVCodecContext* ctx = libavsmash_audio_get_codec_context(hp->adhp);
     if (vs_setup_audio_rendering(hp->aohp, ctx, layout, (int)MAX(rate, 0), &hp->ai, core, vsapi) < 0
-        || count_output_audio_samples(hp, skip_priming != 0, out, vsapi) < 0) {
+        || count_output_audio_samples(hp, skip_priming != 0, skip_tail != 0, out, vsapi) < 0) {
         free_handler(&hp);
         if (!vsapi->mapGetError(out))
             set_error_on_init(out, vsapi, "lsmas: failed to configure audio output.");
@@ -270,8 +209,7 @@ void VS_CC vs_libavsmashaudiosource_create(const VSMap* in, VSMap* out, void* us
     libavsmash_audio_force_seek(hp->adhp);
     lsmash_discard_boxes(root);
 
-    VSNode* node = vsapi->createAudioFilter2(
-        "LibavSMASHAudioSource", &hp->ai, audio_get_frame, audio_free, fmUnordered, NULL, 0, hp, core);
+    VSNode* node = vsapi->createAudioFilter2("LibavSMASHAudioSource", &hp->ai, audio_get_frame, audio_free, fmUnordered, NULL, 0, hp, core);
     if (!node) {
         free_handler(&hp);
         vsapi->mapSetError(out, "lsmas: failed to create the LibavSMASH audio node.");

VapourSynth/lsmashsource.c

@@ -76,7 +76,7 @@ VS_EXTERNAL_API(void) VapourSynthPluginInit2(VSPlugin* plugin, const VSPLUGINAPI
         "clip:vnode;", vs_libavsmashsource_create, NULL, plugin);
     vspapi->registerFunction("LibavSMASHAudioSource",
         "source:data;track:int:opt;skip_priming:int:opt;layout:data:opt;rate:int:opt;decoder:data:opt;ff_loglevel:int:opt;drc_scale:"
-        "float:opt;ff_options:data:opt;",
+        "float:opt;ff_options:data:opt;skip_tail:int:opt;",
         "audio:anode;", vs_libavsmashaudiosource_create, NULL, plugin);
     vspapi->registerFunction("LWLibavSource",
         "source:data;stream_index:int:opt;cache:int:opt;cachefile:data:opt;" COMMON_OPTS