We love our transcript tool, and we sure hope you do, too!
It's a pretty amazing feature, and knowing that there's no person sitting behind a keyboard getting your transcript ready means that there is some serious smarts that went into getting that tool ready for the world to use (thanks IBM). It also keeps the folks here free to do other things, like keep the platform running smoothly for you.
But the big question mark here is HOW can you make the transcript tool work smoothly, in your favor, and without too many hiccups (or headaches)?
While the logic behind the logic isn't as easy as an equation, we do have some key points to keep in mind before you hit that button to process a transcript or a snip.
- Languages! As multilingual as AI is, it can still only handle so much. Check out which languages our transcription tool supports here, and think twice about hitting that button if your podcast has more than one language in it.
- Non-human "voices". We're not kitten. The tool can't transcribe the dog barking or the rooster crowing, so if it comes across this in the audio, it's going to give up.
- Cross-talk. We don't mean the occasional cross-talk, but if there's a lot of cross-talk or excessive overlap, the transcript will likely fail.
- Background noises. This one is huge. If your listener has to tune in closely to distinguish what the speaker is saying versus what is happening in the background, our transcription tool won't be able to do it.
- Audio quality. This is a given. If the audio quality is poor, then your transcript will be poor or even non-existent.
- 🎵 Music 🎸 🥁 That bass is pretty cool and we're really digging the drums, but the AI isn't quite as appreciative. If there's music mixed in through a good portion of your show, or if the hosts and guests are speaking at the same volume as the music, the transcription tool is going to throw in the towel, sadly.
All this said, press that button! Pop open a chat and let us know if you run into an issue and you think that the tool should have been able to transcribe your audio...we'd be happy to take a look at it for you.