@Lorgar Yep. Took me over 2 months but it’s done. AI really is soemthing.
Awesome job, congrats.
Can you outline the process? What was the hardest part?
@Lorgar , you did the Gynferno right? You should redo yours like this. I enjoyed your version but it would be better with great AI voices, like sampling Sandman's actual voice. That would be funny as hell when he says "And cheeeeers!"
Also, is it possible to monetize it? There are 2 audiobooks of Dante's Inferno on audible, this could be a third.
@UncleIroh @Lorgar I didn’t intend for it to be monetized. I even say that in introduction. The Inferno itself is public domain, however the voices belong to who ever owns their rights.
So if someone sampled Sandmans voice, Sandman would own those rights by default and someone else couldn't monetize his voice?
Or is it the case like where celebrities, voice actors and studios explicitly make someone's voice their legal IP?
@UncleIroh @Lorgar I don’t know the exact law but the studio would own the rights to the character’s voice.
Sean Schemmel, the English voice actor for Goku, once said that he can’t lend out Goku’s voice for things like Team Four Star because it would be against his contract. Which would mean Funimtion owns his voice acting for Goku.
So to avoid legal troubles you’d need clips of the voice actor speaking normally not in any character, which is probably illegal anyways. 🤷
Sounds about right.
@UncleIroh @Lorgar thank you for the praise. It was tedious more than anything.
Step 1. Finding instances of the characters speaking with little background noise. I was extremely fortunate LoK has a lot of that. Then it was screen shotting the clips and putting them into Eleven Labs to synthesize them.
Step 2. Copying the entirety of The Inferno and parsing it by who speaks when.
Example:
Canto 14 part 5 - Raziel
Canto 14 part 6 - Kain
………………………………………………………………
@UncleIroh @Lorgar
Step 3. Eliminate all background noises in Reaper (a DAW). Done with a noise gate or manually.
Step 4. Equalize volume for each character to every other instance in the poem, first by each character to themselves in every canto, then each character to themselves in all cantos.
Step 5. Put the entire poem together in sequential order, adjusting volumes along the way. Render it.
Step 6. Make speaker credits, introduction, & thumbnail. combine & upload.
Yeah that seems to be one of the trickiest parts, finding enough quality audio material for 11Labs to do its thing.
In my brief experiment with 11Labs this was a major stumbling block.
@UncleIroh If you’re getting the source from video games there may be options to turn sound effects and music volume to zero so that the only sound is from dialogue.
I was able to make Android 18’s voice this way using voice lines from DBFZ
Nice, that's put the thought back in my head to have another go.
@UncleIroh @Mr_Mister I'm not sure it's ok to steal the voice of someone. But I agree that I should redo an audio of my book (and write the next parts!)
@Lorgar Well that’s why I’m not going to monetize this. I’m also not going to do anything illegal or malicious with these voices.
I think I've heard TFM say he's OK with people using his material and voice, could be wrong about that.
Sandman would probably be up for it if you asked.
@Mr_Mister WTF