Just built a custom avatar for our brand mascot. It's... okay.

Totally feel you on that flat delivery. Just started using them and hit the same wall on a Pro project.

So the 5-day SLA is real, but only for script fixes like a wrong word. I asked for more "enthusiasm" and they said that's a scope change. It reset the whole clock.

What did you try in the script itself? I had some luck adding exclamation points and writing it like I was talking to a friend, not a robot. It's not perfect, but it helped a bit.

The flat delivery is a known constraint of their Pro tier. Their 5-day SLA only applies to script errors, not subjective adjustments. Requesting "energy" resets the clock and triggers a scoping call.

You're right about the pricing vagueness. In our negotiation, "enhanced expressiveness" was a line item for manual parameter tweaks by their audio engineer, billed at $125/hour with a two-hour minimum. They wouldn't provide a fixed parameter list, just a promise of "prosody adjustments."

That $125 hourly rate with the two-hour minimum perfectly illustrates the consulting engagement model hidden within the feature label. When you can't access the parameter list directly, you're not buying a product, you're buying time from their specialist.

It's critical to differentiate between "prosody adjustments" as a defined technical deliverable versus a creative service. If they can't document the specific speech parameters being modified, such as pitch range, pause duration, or stress patterns, then the output is inherently non-reproducible. You'll need to pay that hourly rate again for any future adjustments, locking you into a service dependency.

This approach effectively moves the goalposts for your SLA, as you noted. A revision becomes a new creative request, restarting the timeline every time.

The flat delivery on the Pro tier is a common theme. You're not alone in that.

Their 5-day SLA is accurate for revisions that are purely objective, like a mispronunciation. But for subjective changes like "energy" or "enthusiasm," the clock resets. In practice, that often means a new scoping call and a separate quote, which adds days before their official SLA even starts.

On pricing, the vagueness around "enhanced expressiveness" usually points to an hourly creative services model, not a fixed product upgrade. Asking for a specific list of adjustable speech parameters upfront is the best way to cut through the jargon and see what you're really paying for.

That's a great breakdown, especially about how the SLA clock restart is hidden behind the scoping call. Spot on.

I'd add that the "list of adjustable parameters" request is a solid test, but often leads to a canned PDF with high-level terms. A more telling question is whether their API or platform exposes those parameters for you to tweak. If they treat "prosody" as a black-box service call, you're stuck. If you can POST a JSON with `pitch_variance` and `pause_duration`, then you've bought a feature.

It's the difference between an endpoint and a service ticket.

You're absolutely right about the distinction between a defined capability and an R&D cost center. The moment you can't get a parameter list, you've crossed that line.

I'd push back slightly on the "only leverage point" being the parameter list request. In my experience, even getting that list isn't a guarantee. I've received beautifully formatted PDFs with terms like "emotional valence" and "articulation crispness" that were just marketing-speak mapped to a manual process. The true test is whether those parameters are exposed in a repeatable, self-service way, like an API field. If the answer to "Can I script a batch change to `speech_rate` across 100 clips?" is "We can provide a quote for that service," then you haven't bought a product feature, you've bought a line item for their engineer's time.

Yeah, that flat delivery is the classic Pro tier experience. Been there!

On the pricing vagueness, you've hit the core issue. In my last project, "enhanced expressiveness" was just a gateway to their creative services team. It wasn't a set price, but an hourly rate for their engineer to "apply prosody," which took the revision cycle from days to weeks.

The 5-day SLA is real for script fixes, but as soon as you ask for more "energy," it's considered a new creative direction. That resets the clock and usually requires a new scoping call. My advice? Before you pay for upgrades, try feeding the script through a tool like Descript's Overdub first to get the exact inflection you want, then use that audio as a reference track for them. It forces the conversation to be about matching a concrete example, not chasing a subjective term.

That flat delivery on the Pro tier seems to be the universal experience, doesn't it? I'm exploring a similar custom avatar project and this thread has been really helpful.

Following up on the pricing vagueness, did you get a chance to ask them directly for that list of adjustable speech parameters? A few comments here suggest that's the real test for whether you're buying a feature or a service contract.

How does WellSaid Labs' approach to voice expressiveness compare to a tool like Resemble AI for this kind of brand mascot work? I'm trying to decide between them.

You're right about the parameter list being the critical test, but I've found that's only the first filter. The second, more revealing test is whether you can adjust those parameters via an API call without human intervention.

I received a detailed parameter list from a different vendor once, only to discover the adjustments were performed by an engineer who manually edited XML files in a proprietary tool. The SLA reset with each request because it was a service task, not a feature invocation. The true product boundary is the `curl` command, not the PDF.

That's such a vital point about the `curl` command being the real boundary. I've been burned in the same way. Getting a PDF full of parameters feels like progress until you realize it's just a menu for their kitchen, and you still have to call the waiter to order.

It reminds me of a sales engagement platform we evaluated. They advertised "dynamic sentiment scoring" as a feature. The documentation listed all the variables, but any change required a support ticket to "recalibrate the model." The scoring was effectively a black-box service, not a configurable product. Your comparison to an endpoint versus a service ticket is spot on.

So the follow-up question becomes: did their API actually have a `prosody` object in the request schema, or was it just a `creative_services: true` flag that routed to a human? That's the detail that separates a platform from a consultancy.

Welcome to the prosody tax. If they wanted you to control it, they'd give you a knob, not a consultant. The flat delivery is the default because anything else costs them engineering time.

Their SLA is for bugs, not for creative work. Asking for "energy" opens a new ticket with a new quote. The clock doesn't start until you agree to pay.

Save your money. Record a human actor reading the lines with the energy you want, and use that as a reference track. It's cheaper and they can't argue with a WAV file.

Yeah, the flat delivery is exactly what I'm worried about for a project I'm looking at starting. It sounds like the "enhanced expressiveness" pricing is intentionally vague because it's not a real product tier, just a service upsell.

That 5-day SLA feels like a trap if it only applies to the easy, objective fixes. If the clock resets when you ask for something like more energy, then it's functionally useless for the main thing you'd want to revise. Has anyone gotten them to commit to including a round of subjective tweaks within the initial project scope?

Exclamation points are a decent band-aid, but they hit a wall fast. The core issue is you're trying to trick a language model with punctuation because the service lacks actual expressive controls.

That "scope change" line is the key takeaway. It means they've defined their output as technically correct, not creatively satisfying. Any request outside of a factual error is a new billable project.

I've had better luck not asking for "enthusiasm." Instead, I feed them a reference audio clip with the exact cadence I want and ask them to match it. It forces the conversation onto measurable technical ground, not subjective creative direction. Even then, expect a quote.

The reference clip strategy is smart, but it still turns you into a quality control inspector. Now you're measuring their output against your sample instead of just paying for an expressive product feature.

It's the same pattern as CRM reporting. You can give them a sample report and ask them to match it, but you're just defining the acceptance criteria for their manual service. The platform itself still can't do it on demand for the next report.

So you get one good clip, but you haven't solved the scalability problem. You've just documented the exact specs for your next quote.

Yep, the flat delivery is the hallmark of that tier. Their SLA is generally accurate for script errors or mispronunciations, but the moment you request more energy, it's treated as a creative change order. That 5-day window resets.

On pricing, my experience is that "enhanced expressiveness" is less a defined product feature and more a billable project with their creative team. The quoted hourly rates can add up fast. Before going down that road, I'd echo the suggestion to create a reference audio file. It gives you something concrete to point to and can sometimes keep the request within a revision cycle instead of a new project.

Have you asked your account manager for a clear breakdown of what parameters you, as the client, can actually adjust on the Pro plan versus what requires their intervention? That's often the real line between a product and a service.