Forum Discussion

NikiStoyles's avatar
NikiStoyles
Community Member
2 months ago

Closed Caption issues with AI Text to Speech

Hi all 

I am loving the new AI Text to Speech voices but have found an issue with the Closed Captions. The Closed Caption placeholders are not lining up or splitting where expected. 

Image 1 is how the closed captions are displaying for the voice to text.  

Image 2 or 3 are both examples of how I would expect the captions to display, with the placeholders aligned to speech. Either of these would be acceptable. 

Is anyone else experiencing this? 

Not sure if this is a bug or an enhancement request or if I'm missing something? 

Thanks all! 

Niki Stoyles

 

  • NikiStoyles's avatar
    NikiStoyles
    Community Member

    Hello Eric. as commenter Laura stated, the example you posted actually makes our point - that the closed captions are not aligning with the natural breaks in the speech. 

    I've attached a sample file in this post. The voice I am using is Matilda. The red arrows in the image here show where I would expect the captions to break but as you can see they are breaking at random locations. 

     

  • KelleyConrad's avatar
    KelleyConrad
    Community Member

    I see the same thing with my closed captions. I would expect them to break at naturally stops and pauses in the audio, especially since the AI should know where those are based on the notes. I've also found some text gets left out of the captions completely - usually a phrase at the end of a sentence or a sentence near the end. I have to manually fix all the captions based on the issue you identified and the missing text.

    • EricSantos's avatar
      EricSantos
      Staff

      NikiStoyles KelleyConrad 

      Hello Niki and Kelley,

      Sorry to hear about the trouble with the closed caption synchronization. I created the attached Storyline file and tried adding AI audio with closed captions. Fortunately, the captions are synchronized with the AI audio.

      Can you reproduce this issue consistently, or does it happen when using certain AI voices? What AI voice are you using? As a reference, I used the AI voice, "Roger."

      If the issue persists, we can try running initial troubleshooting to isolate the problem. If you're already working on your local drive, try running these steps to fix possible installation issues

      We'd be glad to investigate further as needed. You may share your Storyline project here or privately in a support case, where our Support Engineers will work with you one-on-one. We'll delete it when we're done testing.

      • KelleyConrad's avatar
        KelleyConrad
        Community Member

        I've had the same issue no matter which voice I use. Even in your example, the captions don't break at the natural breaking points. For example, your third caption starts the next sentence near the end. It would be better for the captions to end at the end of the period and then begin the next caption with the next sentence. Most of my clients want it to work that way, so I have to manually clean them up. 

         

        The next time I have an example of a dropped caption, I'll share it here, too.

  • BenBoozer's avatar
    BenBoozer
    Community Member

    I'm going to chime in with the same issue. Love the new voices but having to do a lot of cleanups of closed captions. Using Laura.

  • NikiStoyles's avatar
    NikiStoyles
    Community Member

    Thanks for the reply and I'm glad its not just me! I'm spending more time cleaning up these closed captions than I did when I recorded using my own voice 😬 

    • IrisSchlabitz's avatar
      IrisSchlabitz
      Community Member

      I experienced the same issue with various AI voices that I have tried. And I had to spend some time too, to adjust the correct captions, pauses, etc. This should be fixed!

  • Hello everyone,

    Thank you for sharing your experiences with the AI text-to-speech closed captions!

    I've added more voices and TTS to my attached sample project, and I'd like to invite you to test how the closed captions synchronize with the AI audio. Here's the Review 360 link to my course.

    I might be wrong, but I have tested Roger, Laura, and Sarah and do not see a marked asynchronicity in these voices. Feel free to share your sample Storyline files with AI audio so we can investigate this issue further. You may also share it privately by sending it along in a support case. We'll delete it when we're done testing.

    • IrisSchlabitz's avatar
      IrisSchlabitz
      Community Member

      Hi Eric,

      I just checked your file, and already found this: 

      You see, the voice actually breathes in, and a new sentence starts. But the caption does not adhere to that fact. I would think that the caption would normally break at the end of the sentence, this would make reading the captions much easier and smoothly. This is, I think, what everyone means when posting about this bug?

       

      • NikiStoyles's avatar
        NikiStoyles
        Community Member

        Hi Iris and Eric

        Iris your example is exactly what I'm referring to! its almost quicker to add the closed captions from scratch rather than all the cutting and pasting to line up the captions withthe audio. 

        Eric I did upload a file for your review where the captions (using the voice Matilda) are not aligned to the audio. It is attached to my post from October 21st.   

  • JurgenLepla's avatar
    JurgenLepla
    Community Member

    Also doesn't help that for every caption you need to edit, you need to click the top bar, twice!

    Would be better if it was in the right click menu for the audio, just as the Edit TTS is. 

  • KateKromer's avatar
    KateKromer
    Community Member

    Captions are aligning to the audio clips for me, but turning off automatic CC generation does not work. I like to add closed captions manually since I find the VO content breaks up illogically when using the auto feature. Clsoed captions are part of the learning experience and knowledge transfer and it's especially critical for hearing-impaired audiences. In addition, when there are edits to VO and I need to re-generate audio, I want to be able to maintain my custom CC. Unfortunately, the AI TTS generator is consistently overwriting that. I've even tried exporting my vtt files to preserve my work, but when I reimport the original vtts, they aren't "taking" - Storyline maintains the automatically-generated CC. Is the team aware of these issues and is a fix underway?