Online subtitle conversion tool

[doodle]

Hello,

I just wanted to share an online web-based tool I created for subtitle conversion, shift and translation:
https://subtitleone.cc

100% free, unlimited, no queue - only showing ads to cover hosting costs.
Would love to hear any feedback.

Cheers

[oss]

congratulations you can link to us as to source of subtitles

[mrtinkles]

Just some quick feedback after an initial upload and conversion to test OCR.

For OCR conversion, ran a random check with Vob/Sub file for comparison. Typical error rates were experienced, tracked closely with those generated by SubtitleEdit, with the usual batch of I,l,i,1 cluster of errors. Other issues noted, were those typical gap occurrences due to spacing variations within italicized text strings.

Roughly, the same output in comparison to those generated by SubtitleEdit, unless manual user intervention.

One major issue as noted, was generation in .SSA output. Seems all parent info from source such as positional references were lost, and defaulted as in examples below:

[subtitleone.cc]
Dialogue: 0,0:01:47.26,0:01:50.71,Default,,0,0,0,,Boy, oh, boy,\Na Winchester 73!

Versus:

[Subtitle Extractor 1.0.3.1]
Dialogue: 0,0:01:47.26,0:01:50.72,Dialogue4,Unknown,0000,0000,0000,,{\an4\pos(431,66)}{\i1}Boy, oh, boy,{\i0}
Dialogue: 0,0:01:47.26,0:01:50.72,Dialogue4,Unknown,0000,0000,0000,,{\an4\pos(430,95)}{\i1}a Winchester '73!{\i0}

Note that italicized text was not identified, also occurred in .srt generated output.

1
00:01:47,261 --> 00:01:50,719
Boy, oh, boy,
a Winchester 73!

For reference, the image file uploaded was Handbrake/Inviska MKV extracted from Universal PSDH source.

With any software, including SubtitleEdit, Subtitle Composer or my personal favorite Subtitle Extractor, conversion and quality output without human interaction and post review has always been an issue with OCR conversions.

Anyway, hope this feedback helps a bit. Always chat offline if you like.

Thanks again!

Edit: After re-reading your webpage, maybe I misinterpreted regarding direct conversion, and require clarification.

Subtitle One currently supports converting text-based subtitle files such as ASS, VTT, SBV, and image-based subtitle files such as PGS SUP and VobSub IDX/SUB into SRT or VTT.

[doodle]

Thanks for your feedback.

I agree with all your assessments.

The OCR engine (Tesseract 5 LTSM) unfortunately doesn't recognise italic fonts. From what I gathered in the maintainers discussions it's not going to happen very soon. So I guess I'll just have to wait until they support it.

Subtitle One was indeed not intended to replace some more specialised offline tools. I intended for it to be able to carry out all the conversion process in the browser. This means I had to port every (non-JS) library we use into JS - which increased complexities. I might be able to extract positional reference from the source for SSA output, but this will take some time to implement.

All in all, the goal was to provide an alternative to other web-based tools such as Subtitle Tools rather than an alternative to offline tools and/or paid services. Being able to process everything in the browser without sending files to the server was also one of the design goals.

Hope that explains. Cheers!