I actually hit the ground running with Vois Studio, which is allowing me to share my thoughts perhaps a bit sooner than I normally would. Still, bear that in mind that I'm still learning the tool and may uncover more along the way, but I was able to get up and running quickly and actually created real content so I feel like I'm reasonably qualified to speak to it even at this early stage.

My short review: Vois is a capable tool that has a few areas that need some polish and improvement, but is business-ready if you're able to work around those rough edges (I was). If I could give a more nuanced rating, I'd probably go 4.5 tacos to reflect that "almost there" status, but for a relatively young product, I was pleasantly surprised by how good it already is. So I'm happy to round up to 5 tacos to reflect that as well as the potential I see.

My longer review:

My specific use case is that I regularly create short podcasts for myself and my clients, so Vois' voice cloning and script features were very familiar to me. I had been leveraging a SaaS product (Wondercraft) to do this, and while it's not terribly expensive, I was paying for more than I ultimately needed, especially with their push to get me to upgrade to access their new AI tools.

While Vois does not have all the shiny, built-in AI functionality (note: they have rolled out the ability to use local AI agents through CLIs), the core capabiltiies are pretty solid. I cloned three voices within minutes, learned how to craft and annotate my script pretty easily, and it generated an almost-ready-to-use audio file on the first pass.

Further, the interface is pretty easy to figure out, though it lacks a few conveniences that I'd like to see added soon (more on that below). Still, I was able to go from download to having my first episode finished in about an hour.

Setup isn't too bad, though I ended up redoing my clones a few times. Vois is particular about the clip length (it must be under 60 seconds long), and I found that I needed a really clean recording to upload.

Here's my more detailed list of Pros (+), Cons (-), and Observations (/):

+ The ability to run an unlimited number number of generations is very nice. No need to monitor a credit allowance!
+ If you have any experience with an editor, Vois is pretty easy to figure out. It has it's own specific quirks, but the fact that I went from installation to a finished product in about an hour I think is telling
+ You can override pronunciations on a global or project-specific basis
+ There's a solid selection of pre-built voices to select from, along with the ability to clone voices (the number of clones being based on your plan)
+ There's an easy way to add some audio "personality" - a chuckle, sigh, etc. - into your recordings
+ Adding music is simple to do and, again, there's some built-in options or you can upload your own

/ Processing time was reasonable. As everything runs locally, your experience will depend on your hardware. I'm running on a 3 year old business-focused Windows machine - not a gaming rig - and Vois was acceptably fast
/ The tagging of the script is pretty easy though some keyboard shortcuts would be welcome rather than having to mouse-click everything
/ There's mention of an Ultra or Pro plan in the software but it's not currently available. However, if/when it is, there's apparently going to be an additional fee associated with it
/ If you are doing anything commercial with your recordings, you need to be on Tier 3

- Every run seemed to create at least one audio artifact - an unexplained pause, an odd pronounciation, or an unwanted word (on one point, Vois said "dot" when there was a period)
- The timeline interface is 85% great, but lacks a built-in scroll bar which was surprising
- There's no option to regenerate just one clip so Vois reprocesses everything when you have a change. To the positive, it recognizes unchanged clips and skips those, but to the negative, every time you regenerate, it resets all the timeline changes (moving around clips or shortening them all gets reset).
- Vois could benefit from more contextual recognition. Take "read" for example: "I read the book" vs. "I will read the book" both generated with a past tense pronunciation. I had to fake Vois out by switching it to "reed"

Still, I'm pleased with the purchase. It's a rare experience for me to find something on AppSumo that can fully replace an existing tool from the outset but it looks like Vois is going to be that for me. I do see room for further refinement, but where the product is at already is impressive and I see things only getting better from here.