What "unlimited text to speech" actually means (and where most tools quietly break the promise)

Search "unlimited text to speech" and you'll find a dozen tools claiming exactly that. We tested them. Here's what "unlimited" turns out to mean in practice — and the one thing almost none of them actually deliver.

The two kinds of "limit" that get hidden

Most tools that say "unlimited" are only unlimited on one axis. Read the fine print and you'll usually find one of these:

• Unlimited generations, capped characters per run. You can convert as many times as you want — but each conversion still tops out around 2,000–2,500 characters, so a real document still needs manual splitting.
• Unlimited free tier, until it isn't. A generous weekly character allowance (some tools advertise 20,000+ characters/week) that resets — fine for short tasks, but not actually unlimited for a heavy user or a long document.
• Unlimited listening, limited downloading. You can play the audio in-browser as many times as you like, but MP3/WAV export is capped, credit-based, or behind a paid tier.

Why this is possible at zero cost

Cloud-based "natural AI voice" tools pay a provider (Azure, Google, ElevenLabs) per character generated — that's a real, ongoing cost, which is exactly why they need a cap or a subscription to stay in business. HearYourDocs runs the voice engine in your browser, using WebAssembly, not a server call. There's no per-character bill to recoup, so there's no reason to cap you.

The trade-off, stated plainly: the voice is a formant-synthesis engine (clear, functional, closer to a classic screen reader) rather than a premium neural voice. That's the honest cost of genuinely unlimited and free.

How the claims stack up