Train AI voice

Is there anybody we can contact at the church about an AI project I'm trying to do?

I had been thinking of having AI READ some older text (Text to speech) - one example would be reading talks from Journal of Discourses, or general conferences that we don't have an audio source of.
Maybe a small miracle but when I couldn't figure out which direction to go, this project got released.
https://swivid.github.io/F5-TTS/

This is all done locally on my computer, all I need is voice to train it with.
I tried my voice and it has been okay. I actually did download some Lloyd Newell reading Moses and wasn't high enough quality.
I have found some voice clips online that are pretty good, but that is somebody else's voice. (It sounds similar to Elder Patrick Kearon's voice, which I think people would enjoy listening to).

I guess what I'm looking for is if the church has any "official" voice that I could use to train AI with? Anybody at the church I would reach out to discuss this idea?
I want to do it the correct way.

Nice! I think there's a lot of opportunity when it comes to older content that was never recorded.

The Church is already using AI voices in a few places, such as magazines. For example, if you listen to any of the articles in the October 2024 Liahona, at the end of the audio it will say that the voice was generated using AI. There are different voices being used on different articles, and the quality is pretty good (for an untrained ear).

I think it's unlikely that the Church would want "official" AI voices floating around outside of Church control – for example, voices that imitate apostles or other Church leaders – as it could be used dangerously in the hands of someone with an agenda against the Church. (That said, I suspect it's only a matter of time before it happens, given that there's so much publicly-available audio and video from recent general conferences that could be used for training.)

The official way to contact product or development teams at the Church is by sending feedback through a Church app (like Gospel Library) or using the feedback link at the bottom of the Church website.

If you're just wanting permission to create and share audio for content that's copyrighted by the Church, with synthetic audio that doesn't imitate Church leaders, you can submit a request through Permissions.ChurchofJesusChrist.org. Anything published before 1929 (like the Journal of Discourses) is public domain, so it can be used without permission, but more recent content is still in copyright.

I guess going back.

Do we know what program the church is using to generate its AI voice over?

I listened to some from that Oct Liahona, seemed to have a couple of voice options. It seems like Eleven Labs is one of the more popular ones... and sample voices seemed similar.

tubaloth wrote: Sat Dec 14, 2024 8:59 am Do we know what program the church is using to generate its AI voice over?

If you're talking about the Gospel Library app, I'm pretty sure the app is passing it to the device's text-to-speech.

One of the major functions of that feature is to reduce the amount of data required to be downloaded as text uses far less than sound.

Yes the church was doing that, but seems like they shifted over to AI. They seemed to be using it in the new articles for the Liahona. (They tag at the end of the audio).

Just curious what program they're using.