Forum Discussion
Generate closed captions from tts could be better
Yes! Totally agree. Thank you for posting about this folks.
I also constantly have this issue. The SL guidance provided around auto-generated closed captioning says that the captions are split according to pauses in the spoken audio. In my experience this is absolutely NOT the case. The splits seem to be completely random, and ignore the very obvious audio pauses and text punctuation.
As Jurgen and Sarah have mentioned, I too spend not just hours but DAYS fixing the location of the splits in long courses to more logically marry with the flow of the spoken phrases and sentences. This is especially an issue in closed captions for the inbuilt Text-to-Speech feature.
It's so important for the visual presentation of the closed captions to model logical phrasing in spoken language, not just from an accessibility POV re. equitable access to information, but also for course participants whose first language may not be what they are reading on the screen.
SL, is there a simple way to ensure that closed caption balloons are split according to the natural pauses in the audio and punctuation in the text provided for TTS?
Your help would be much appreciated. This is such a time-consuming design flaw that I feel like I'm missing something really obvious!
Thanks.
Hello Liz,
I'm sorry to hear about your experience with the closed captions editor.
Getting the caption placeholders to match the corresponding audio each time is tricky since each audio file differs. What might sound like pauses to us might be interpreted differently by a machine. This is why we recommend tweaking the timing of captions, adding more caption placeholders, or splitting captions into smaller chunks as a way of fine-tuning the captions.
We also understand that the function might not be working as intended for you, so if you'd like to share a copy of your project file here or in private by opening a support case, we'll be glad to take a closer look to test the behavior you're experiencing. We'll delete it when we're done testing!
