How to write .vtt and .srt when running whisper in python?

[jamiecm17]

I'm trying to export .vtt and .srt caption files. Using command line, this happens automatically. However, there is no file output when running whisper in VSCode. Is there an additional command or argument I need to pass to output .srt and .vtt files from VSCode?

[glangford]

If you want to run it in the same way as the command line then you can supply arguments by changing launch.json, see this Stack Overflow article.

https://stackoverflow.com/questions/57889703/in-visual-studio-code-how-to-pass-arguments-in-launch-json

If you want to see how whisper writes .srt files for example, see here

whisper/whisper/utils.py

Line 102 in a6b36ed

class WriteSRT(ResultWriter):

[tflax32]

Here is some block code you can try running directly in your own env. This utilizes the pysrt library to generate an srt file:

import whisper
import pysrt

model = whisper.load_model("base")
audio_file = ("YOUR AUDIO FILE PATH")
result = model.transcribe(audio_file, language='en')

subs = pysrt.SubRipFile()
sub_idx = 1

for i in range(len(result["segments"])):
    start_time = result["segments"][i]["start"]
    end_time = result["segments"][i]["end"]
    duration = end_time - start_time
    timestamp = f"{start_time:.3f} - {end_time:.3f}"
    text = result["segments"][i]["text"]
    
    sub = pysrt.SubRipItem(index=sub_idx, start=pysrt.SubRipTime(seconds=start_time), 
                           end=pysrt.SubRipTime(seconds=end_time), text=text)
    subs.append(sub)
    sub_idx += 1
    
subs.save("output.srt")

[mjysci]

You can use whisper.utils.get_writer to get subtitles directly from model.transcribe result.

Here's an example.

[sovdevs]

sounds great but have been having import issues with this

[glangford]

@sovdevs We need more information to be able to help. Can you say which version of whisper you are using and some sample code with the import error you are getting?

[glangford]

...also, does it work outside VSCode but break if you are running in VSCode (like the original poster)

[sovdevs]

Thanks for getting back.
openai-whisper @ git+https://github.com/openai/whisper.git@b91c907694f96a3fb9da03d4bbdc83fbcd3a40a4
VS code locally thru virtualenv.
To write file I am currently using

from whisper.utils import write_srt
        with open(output_file_path, "w", encoding="utf-8") as srt_file:
              write_srt(result["segments"], file=srt_file)

which works well enough as an alternative.

However, in Production, I get this however

    from whisper.utils import write_srt
ImportError: cannot import name 'write_srt' from 'whisper.utils

[glangford]

This may be the issue (from your post in the other thread); write_srt has been removed. You may be using two different versions of whisper locally vs production (?)

• Starting today, I haven't been able to run "from whisper.utils import write_vtt" - any ideas why? #886 (reply in thread)