changes main.py & test_srt.py issue #11

[hirendra84]

feat(subtitles): migrate subtitle generation from SRT to WebVTT

• Rename generate_srt_files() to generate_vtt_files()
• Change output extension from .srt to .vtt
• Add WEBVTT header and WebVTT-compliant formatting
• Use dot-separated timestamps and speaker cue tags
• Remove segment numbering
• Update api, main, gradio demo, and tests
• Add VTT-specific test assertions

[shigabeev]

Hey @hirendra84 thank you for you PR. Sorry for a late reply. I'll check it and come back here to see if I can merge it.

[shigabeev]

Thanks for picking this up, @hirendra84. Direction is right, but this can't merge as-is — I think you may have missed a commit.

The diff only touches main.py and tests/test_srt.py. generate_vtt_files() doesn't exist yet: langswap/translation_pipeline.py:187 still defines generate_srt_files() and still writes SRT (numbered cues, comma timestamps, .srt filenames). So main.py raises AttributeError at the last stage and the new test fails. langswap/api.py:132, langswap/api.py:193 and gradio_demo.py:141 also still call the old name.

The point of #4 is the speaker tags, not the rename — that's the whole reason VTT is worth moving to. The data is already there: TranslatedTextedSegment.speaker (langswap/pipeline_models/models.py:34) is populated by the ASR clients, it's just never written out. So the change belongs in generate_srt_files():

• WEBVTT\n\n header
• drop the cue numbers
• , → . in timestamps
• wrap text as <v {segment.speaker}>…
• .srt → .vtt in the two get_file() calls
• Could you push that plus the api.py / gradio_demo.py call sites? Worth renaming tests/test_srt.py → test_vtt.py too.

One note on the test: asserting SPEAKER_00 passes even if the speaker path is broken, since every ASR client defaults to that literal. Build the fixture with two distinct speakers and assert both appear — otherwise the test doesn't test the feature.