What SAMI actually is
SAMI stands for Synchronized Accessible Media Interchange, and it is Microsoft's answer to the same problem SRT solves — it just solved it with HTML. A .smi file is a web page with SYNC tags:
<SAMI>
<BODY>
<SYNC Start=1000><P Class=ENCC>Hello there
<SYNC Start=4000><P Class=ENCC>
<SYNC Start=5500><P Class=ENCC>Second line
<SYNC Start=8250><P Class=ENCC>
</BODY>
</SAMI>
Each SYNC opens a caption at a millisecond offset. There is no end time — the caption runs until the next SYNC. The blocks are screen-clears: the intent is for the caption to disappear between lines.
How the end times are derived
Because SAMI never states when a caption ends, the end of one cue is the start of the next. Blank SYNC blocks are dropped rather than emitted as empty cues, so a caption runs until the next real caption appears. That produces the same on-screen result the original file would have had in Windows Media Player.
Why so few tools open .smi
SAMI was built for Windows Media Player, and it died with it. Nothing made after roughly 2010 reads the format: not YouTube, not Vimeo, not Premiere, not DaVinci, not a browser's video element. Converting to SRT is the only way to use the subtitles in anything modern.
The one thing SAMI did better than SRT is multi-language: several language tracks could live in one file, separated by CSS class. That is genuinely useful and nobody has replaced it properly — the modern equivalent is one separate subtitle file per language.
Where .smi files come from
Old Windows Media content, Korean and Japanese video releases from the 2000s (the format was widely used there), some DVD rips, and training videos produced for Windows Media Services. Opening one in a text editor shows HTML soup, which is usually the moment someone starts looking for a converter.
After converting
The SRT opens everywhere. If the destination is a browser player, WebVTT is the native format. If the source turns out to have no usable timings at all, text to SRT can rebuild them from the text.