starred/ccextractor
1
0
Fork 0
mirror of https://github.com/CCExtractor/ccextractor.git synced 2026-02-04 05:44:53 +00:00
Code Issues 1.5k Packages Projects Releases 20 Wiki Activity

Improve the start and end timestamps of extracted burned in captions (#962)

Browse Source 
The start and end timestamps of extracted burned in captions are flawed
and off by a large difference. Also, the start time of the first burned
in caption extracted is always zero, which is not always the case. And
the extracted captions always appear in continuous timestamps.

This commit improves the start and end timestamps of the extracted
burned in captions and reduces the error significantly, bringing the
timestamps fairly close to the actual timings as they appear in the
media file.
This commit is contained in:
Saurabh Shah
2018-03-12 23:49:24 +05:30
committed by Carlos Fernandez Sanz
parent 801f9e8dc8
commit 86356ba4d2
1 changed files with 51 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

75
src/lib_ccx/hardsubx_decoder.c
View File

@@ -374,13 +374,16 @@ int hardsubx_process_frames_tickertext(struct lib_hardsubx_ctx *ctx, struct enco
int hardsubx_process_frames_linear(struct lib_hardsubx_ctx *ctx, struct encoder_ctx *enc_ctx)
{
// Do an exhaustive linear search over the video
int prev_sub_encoded = 1; // Previous seen subtitle encoded or not
int got_frame;
int dist;
int dist = 0;
int cur_sec,total_sec,progress;
int frame_number = 0;
int64_t begin_time = 0,end_time = 0,prev_packet_pts = 0;
char *subtitle_text=NULL;
char *prev_subtitle_text=NULL;
int64_t prev_begin_time = 0, prev_end_time = 0; // Begin and end time of previous seen subtitle
int64_t prev_packet_pts = 0;
char *subtitle_text=NULL; // Subtitle text of current frame
char *prev_subtitle_text=NULL; // Previously seen subtitle text
while(av_read_frame(ctx->format_ctx, &ctx->packet)>=0)
{
@@ -425,27 +428,42 @@ int hardsubx_process_frames_linear(struct lib_hardsubx_ctx *ctx, struct encoder_
progress = (cur_sec*100)/total_sec;
activity_progress(progress,cur_sec/60,cur_sec%60);
if(subtitle_text==NULL)
continue;
if(!strlen(subtitle_text))
continue;
char *double_enter = strstr(subtitle_text,"\n\n");
if(double_enter!=NULL)
*(double_enter)='\0';
//subtitle_text = prune_string(subtitle_text);
if((!subtitle_text && !prev_subtitle_text) || (subtitle_text && !strlen(subtitle_text) && !prev_subtitle_text)) {
prev_end_time = convert_pts_to_ms(ctx->packet.pts, ctx->format_ctx->streams[ctx->video_stream_id]->time_base);
}
end_time = convert_pts_to_ms(ctx->packet.pts, ctx->format_ctx->streams[ctx->video_stream_id]->time_base);
if(prev_subtitle_text)
if(subtitle_text) {
char *double_enter = strstr(subtitle_text,"\n\n");
if(double_enter!=NULL)
*(double_enter)='\0';
}
if(!prev_sub_encoded && prev_subtitle_text)
{
//TODO: Encode text with highest confidence
dist = edit_distance(subtitle_text, prev_subtitle_text, strlen(subtitle_text), strlen(prev_subtitle_text));
if(dist > (0.2 * fmin(strlen(subtitle_text), strlen(prev_subtitle_text))))
if(subtitle_text)
{
add_cc_sub_text(ctx->dec_sub, prev_subtitle_text, begin_time, end_time, "", "BURN", CCX_ENC_UTF_8);
encode_sub(enc_ctx, ctx->dec_sub);
begin_time = end_time + 1;
dist = edit_distance(subtitle_text, prev_subtitle_text, strlen(subtitle_text), strlen(prev_subtitle_text));
if(dist < (0.2 * fmin(strlen(subtitle_text), strlen(prev_subtitle_text))))
{
dist = -1;
subtitle_text = NULL;
prev_end_time = convert_pts_to_ms(ctx->packet.pts, ctx->format_ctx->streams[ctx->video_stream_id]->time_base);
}
}
if(dist != -1)
{
add_cc_sub_text(ctx->dec_sub, prev_subtitle_text, prev_begin_time, prev_end_time, "", "BURN", CCX_ENC_UTF_8);
encode_sub(enc_ctx, ctx->dec_sub);
prev_begin_time = prev_end_time + 1;
prev_subtitle_text = NULL;
prev_sub_encoded = 1;
prev_end_time = convert_pts_to_ms(ctx->packet.pts, ctx->format_ctx->streams[ctx->video_stream_id]->time_base);
if(subtitle_text) {
prev_subtitle_text = strdup(subtitle_text);
prev_sub_encoded = 0;
}
}
dist = 0;
}
// if(ctx->conf_thresh > 0)
@@ -460,15 +478,24 @@ int hardsubx_process_frames_linear(struct lib_hardsubx_ctx *ctx, struct encoder_
// {
// prev_subtitle_text = strdup(subtitle_text);
// }
prev_subtitle_text = strdup(subtitle_text);
if(!prev_subtitle_text && subtitle_text) {
prev_begin_time = prev_end_time + 1;
prev_end_time = convert_pts_to_ms(ctx->packet.pts, ctx->format_ctx->streams[ctx->video_stream_id]->time_base);
prev_subtitle_text = strdup(subtitle_text);
prev_sub_encoded = 0;
}
prev_packet_pts = ctx->packet.pts;
}
}
av_packet_unref(&ctx->packet);
}
add_cc_sub_text(ctx->dec_sub, prev_subtitle_text, begin_time, end_time, "", "BURN", CCX_ENC_UTF_8);
encode_sub(enc_ctx, ctx->dec_sub);
if(!prev_sub_encoded) {
add_cc_sub_text(ctx->dec_sub, prev_subtitle_text, prev_begin_time, prev_end_time, "", "BURN", CCX_ENC_UTF_8);
encode_sub(enc_ctx, ctx->dec_sub);
prev_sub_encoded = 1;
}
activity_progress(100,cur_sec/60,cur_sec%60);
}