I have some audio of an interview carried out in multiple languages. When I used auto-language, it got most of the words from both of the languages, but there were long stretches missing from both. When I transcribed exclusively in language A and exclusively in language B, it got all of the words. So I uploaded two versions of the file and made two transcriptions. Ideally users shouldn't have to upload redundant media, though. Since different transcripts can be made of the same audio with different settings, allow multiple transcriptions per file.