Points, Minutes and Characters: How Pricing Actually Works

Vocallab pricing runs on three units that describe the same thing: your plan is sold in minutes, your balance is counted in points, and every generation is charged from characters. Confusing? Only until you see how they connect — and then the pricing page stops being a guess.

This guide lines the three up, works through the exact formula with real numbers, and shows why the same script can cost the same points but still produce a different length of audio depending on the voice. By the end you'll be able to price a script before you generate it, not after. Full plan details live on the pricing page if you want to follow along.

Every AI voice platform has to solve the same problem: text goes in and audio comes out, and somebody has to decide what a fair charge looks like. Vocallab's answer uses three layers that describe the same underlying thing from three different angles.

Here is the whole rule. 1 point ≈ 1 second of finished audio. But the exact charge for a generation isn't measured from the clock — it's calculated from the script itself: ⌈characters ÷ 17⌉ points in the app. Paste 1,700 characters and the charge is exactly 100 points (1,700 ÷ 17 = 100), whether the voice reads it briskly or takes its time.

This is the part that trips people up, and it's also the reason the pricing page makes more sense once you understand it. You're charged for the text, not for the audio. Two voices reading the identical script cost the identical number of points and still hand you back two different file lengths.

Work through it with real numbers and the logic holds up.

Every plan's headline number is a points allowance wearing a minutes costume. Here's what each tier actually holds.

Translate that into work you actually recognize. A 3,000-point Lite month is roughly 50 minutes of finished audio — about ten 5-minute videos, or one short audiobook chapter once you count the re-takes. Pro's 12,000 points cover about 200 minutes, enough for a weekly video series with room to retry a bad take. Max's 150,000 points run to more than 41 hours of finished audio a month — a production volume, not a hobby budget.

Every regeneration costs its points again. If you're not happy with a read and generate the same 500-point script three times before you land on the one you keep, that's 1,500 points spent, not 500. Nobody budgets for this the first time, and it's the single most common reason a plan that looked generous on paper runs out mid-month.

Monthly subscription points do not roll over. Whatever you don't use resets with the next billing cycle, so a quiet month doesn't bank credit for a busy one. Top-up points work differently: +20 minutes for $5 (1,200 points), +60 minutes for $12 (3,600 points), +200 minutes for $30 (12,000 points) — and top-up points never expire. They also work on any plan, including Free.

That difference decides the right shape for irregular work. If your volume swings — a busy launch month, then a quiet one after — a smaller plan plus a top-up when you need it beats paying for a big plan that mostly sits idle.

Two situations charge differently from the ÷17 rule above, and both are deliberate rather than a trick. Generating through the API charges ⌈characters ÷ 15⌉ points — a premium over the app's ÷17. That same 8,500-character script that costs 500 points in the app costs ⌈8,500 ÷ 15⌉ = 567 points through the API: about 13% more for identical text, and worth knowing before an invoice tells you.

Speech-to-text is priced differently again, because there's no text going in to count. Transcription is billed at 15 points per minute of audio, by duration rather than characters. A 30-minute meeting recording costs 30 × 15 = 450 points to transcribe, regardless of how many words end up in the transcript.

Free gives you 60 points, once, not monthly — about a minute of audio, sized for sampling rather than production. Voice Design produces three variations per run, so a single design session can use a real share of that trial before you've generated anything else. Free-tier output is for personal, non-commercial testing; commercial use needs a paid plan.

The numbers above only matter once you've picked a voice worth paying for. Here are six voices spanning a range of tones and paces — a useful reminder that the cost per script stays fixed no matter which one you cast. Browse the full voice library for hundreds more.