The official WebVTT specification is not yet final and as a result it has been in flux for years. Some players use slight variations of the format (as various versions were proposed at different times). Our original version of the .vtt format was created according to the 2012 specifications, and it's available under the .vtt extension, on our list of caption outputs - older HTML5 players may still require this version.
We also have a newer .vtt file, compliant with the 02/25/2014 version of the WebVTT specification. It's available as the .2014.vtt output type, on our list of caption outputs.
The main difference between the two formats is the syntax for the align: tag (which was different in 2012).
We also generate the .noalign.vtt format, which doesn't store any of the information related to alignment.
Most players can handle the .vtt or the .2014.vtt outputs, but you can check all three output types to see which one works properly with your player.
All three formats use UTF-8 encoding.