heimoshuiyu/whisper-fastapi
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

strip text on tsv and srt output

Browse Source
This commit is contained in:
黑墨水鱼
2024-09-04 18:03:01 +08:00
parent 1a5dbc65e0
commit b4fb0f217b
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
whisper_fastapi.py
View File

@@ -78,7 +78,7 @@ def tsv_writer(generator: Generator[dict[str, Any], Any, None]):
for i, segment in enumerate(generator): for i, segment in enumerate(generator):
start_time = str(round(1000 * segment["start"])) start_time = str(round(1000 * segment["start"]))
end_time = str(round(1000 * segment["end"])) end_time = str(round(1000 * segment["end"]))
text = segment["text"] text = segment["text"].strip()
yield f"{start_time}\t{end_time}\t{text}\n" yield f"{start_time}\t{end_time}\t{text}\n"
@@ -90,7 +90,7 @@ def srt_writer(generator: Generator[dict[str, Any], Any, None]):
end_time = format_timestamp( end_time = format_timestamp(
segment["end"], decimal_marker=",", always_include_hours=True segment["end"], decimal_marker=",", always_include_hours=True
) )
text = segment["text"] text = segment["text"].strip()
yield f"{i}\n{start_time} --> {end_time}\n{text}\n\n" yield f"{i}\n{start_time} --> {end_time}\n{text}\n\n"
@@ -99,7 +99,7 @@ def vtt_writer(generator: Generator[dict[str, Any], Any, None]):
for i, segment in enumerate(generator): for i, segment in enumerate(generator):
start_time = format_timestamp(segment["start"]) start_time = format_timestamp(segment["start"])
end_time = format_timestamp(segment["end"]) end_time = format_timestamp(segment["end"])
text = segment["text"] text = segment["text"].strip()
yield f"{start_time} --> {end_time}\n{text}\n\n" yield f"{start_time} --> {end_time}\n{text}\n\n"