I've finally done something that I've wanted to do for a while, and I'm so happy with how it turned out.
Eleven Labs
I bought a subscription to this AI voice generating tool. By all accounts, they're one of the best on the market, but for me the price didn't seem too great so I shopped around a fair bit.
But, all of the looking around led me back Eleven Labs. For €27 a months you're given 100,000 credits, 1 credit is one character, and one of the stories I created was somewhere in the region of 22,000 characters.
I thought that would translate to about 10-15 minutes of audio, but in fact, it was 28 minutes, which is pretty cool.
Roughly 1 month should allow me to produce 5-7 short stories, which I have some decent plans for.
The best thing about Eleven Labs is that you can generate an entire audiobook narrated by one person, but pre-program numerous dialogue tags with different voices, to give it more of a radio-play feel.
The Plan
I'm going to upload these short stories to social media as 2-3 minutes segments, and then when the final part releases, I'll upload the uninterrupted one to YouTube.
But, with the audio, my dad said he'd take a look at it and put some ambient sounds, music, and maybe some sound effects in, which will take the whole thing to the next level.
I have some plans for creating imagery, but for the time being I'm just going to produce some AI artwork, because my idea will take time, and I'd rather just get this started sooner rather than later.
Credit Restraint
I find the time constraint a bit annoying, but in a way it's a bit of a blessing - if I were to try and look at it positively.
5-7 audio files per month means that I'll have to make sure at least that many are ready to go before the credits run out, and then have more ready to go in time for them to reset. Which means I'll produce these things as fast as I can, to ensure I make the most out of what I have available.
All in all, I do wish there was a bit more leeway, but at the same time, it's okay as is. At least for now.
The first story does have a few hammy moments due to the ai voices, but that's sort of the charm when it comes to them, and apart from those little moments and inconsistencies, it's actually fairly decent.
I must admit I've got such mixed feelings when it comes to AI voices reading things. On the one hand, it enables creativity for people who don't have the funds to pay actual voice actors or have access to a pool of people with the talent and right voices/accents to read out loud (doing it without lots of "ummm" and "aaaah" is quite a skill !). But on the other hand, it's killed a lot of jobs, and it's still on the wrong side of the uncanny valley. I have a particular dislike of the way YouTube is being filled up with AI-generated stuff, but a lot of that is down to the low quality and production values.
But if I decide to publish an audiobook starring the Einheriar, the imperfections of AI would be perfect for them, since they are basically autonomous robotic AI war machines 😁
I wish I had the money to hire some real voice actors to do it, but at the moment, it's a bit outside my reach. The main reason for doing so is that I don't really have much control over the tone or direction of the voices. Also, some of the AI ones kind of take away from the material a bit. But, as proof of concept, it's good.
It'll be interesting to see if a few months of this might attract enough of a following to make hiring actors and adding more production value possible in the future. That would definitely be the goal anyway.
That makes total sense ! Something to look at is your circle of friends, to see who has a good speaking voice with the right tone and accent. You might just have to do a lot of editing until they get into practice reading longer passages without error. Or perhaps you could use them for smaller dialogue parts with AI doing the main narration, blending human and machine.
!BBH
!PIZZA
You know what, that would be a great idea!
To be fair, I could swap out voices from this file at any stage in the edit, so it could be well worth looking into that on a very small scale.
The main narrator voice is actually pretty good, and it seems to get pronunciation fairly correct which was something I was afraid of.
That’s cool, I certainly like the text to speech type of things when it comes to the advances we’ve been seeing. Pretty remarkable stuff!