Free text to speech with no account: what "free" usually costs you

Search for free text to speech and you will find a hundred tools, all of them free, none of them free in the same way. One gives you a monthly credit balance. One gives you unlimited audio with a voice reading its own web address over the top. One gives you everything you want and forbids you from using it for anything that earns money. One asks only for an email address, which is the price.

None of that is dishonest, exactly. It is just that “free” has stopped carrying information, and the word you actually need — free to do what? — never appears on the button.

The five things “free” is usually hiding

Trial credits dressed as a free plan. A monthly character or credit allowance is the most common shape. It is a real free tier, but it is a budget, and budgets run out mid-project. The number to look for is not the allowance but the reset: a monthly pool means one heavy afternoon can cost you the other twenty-nine days.

Mandatory attribution, in the title. Several of the best-known platforms let you generate freely and then require credit — and “credit” is often stricter than it sounds. ElevenLabs’ published terms, as of September 2026, require free-tier and signed-out content to attribute them in the title of whatever you publish, by including “elevenlabs.io” or “11.ai” in it. Not in the description, not in a credits roll at the end. That is a different proposition from a line in the end credits, and it is worth reading the exact wording rather than assuming the usual convention applies.

A commercial-use ban. This is the one that catches people latest and hardest, because it is invisible until someone asks. As of September 2026, ElevenLabs’ published terms state that the free plan carries no commercial licence and “cannot be used for any commercial purpose” — the audio is yours to make and not yours to monetise. Attribution and commercial rights are separate questions, and satisfying the first does not resolve the second.

The detail worth knowing, because almost nobody checks it: upgrading later does not fix what you already made. Their terms are explicit that content created outside a paid subscription — before or after — cannot be used commercially. So the sensible-sounding plan of trying a tool free, building something with it, and paying once the project earns money does not actually work. The audio from the free period stays non-commercial permanently, and re-generating it on a paid plan means re-timing every subtitle against new audio, because synthesis is not deterministic and the second render will not match the first.

An audio watermark. Less common than it was, but still around: a spoken tag or a tone laid over the output. Easy to spot, impossible to remove cleanly, and it makes the free tier a demo rather than a tool.

Your email address, and the text you pasted. An account is a price. So is a text box that ships everything typed into it to a server that keeps it. If the thing you are converting is a draft script, a client’s copy, or anything you would not post publicly, where the text goes matters more than what the voice sounds like.

What to check before you rely on one

Five questions, in the order that they tend to bite:

1. What is the limit, and what does it reset against? Per request, per day, per month, or per account lifetime. A daily cap and a monthly cap of the same size behave completely differently.
2. Can you use the output commercially? Look for the words “commercial licence” in the plan description, not on the marketing page.
3. Is attribution required, and where exactly? In the title, in the description, in the video itself, in a credits file — all four are things someone asks for, and the title is the one that changes how your work looks to everyone who sees it.
4. What comes out? An MP3 is the minimum. Whether you can also get timed subtitles decides whether the audio is usable on a video platform without a second tool and a manual sync pass.
5. What happens to the text? Stored, logged, used for training, or discarded. If the answer is not written down anywhere, assume the first three.

What this tool does, stated plainly

There is no account, no email field, and no paid tier to be moved toward. That is not generosity — it is that there is nothing here to sell, so there is no funnel to protect.

The limits are real and worth stating rather than burying: 5,000 characters per request, and 20,000 characters per day. The per-request cap keeps a single synthesis responsive. The daily cap exists because synthesis costs something to run and an open text box on the public internet attracts automated abuse within days. Neither number is a lever to make a paid plan look better, because there isn’t one.

For comparison, and dated because these change: as of September 2026 the free-tier allowances at the large commercial platforms sit in the region of ten thousand characters per month. A daily allowance of twenty thousand is a different kind of number, and if your work is bursty — a batch of video captions on a Sunday — it is the one that matters.

No watermark is added. No attribution is required. The audio and the subtitle files that come out are yours, and you do not need permission to put them in something that earns money.

What to actually optimise for

If you are converting a paragraph for your own listening, all of these tools are fine and you should use whichever loads fastest.

If you are producing something — a video, a course, a series of clips — the deciding factors are narrower than the marketing suggests. Can you use the output for what you are actually doing? Does the limit reset on a schedule that matches how you work? And does the tool give you timings, or just audio? Audio without timings means a manual sync pass every time, and that cost recurs on every single clip, long after the question of which voice sounded nicest has stopped mattering.