Fish Audio does not run a free tier here. See what you can test before signing up, what a $9 Starter pack covers, and how credits are spent.
Short answer: this site does not give away generations. There is no free plan, no allowance that appears after you sign up, and no watermark-limited trial. Creating an account gives you access to the workspace, not to credits.
That is a deliberate choice rather than a missing feature. Text to speech and voice cloning run on capacity that costs money every time a line is generated, so the model here is simple: you buy a pack of credits once, you spend them on the audio you actually need, and nothing renews in the background.
Most tools that advertise free AI voices are really offering a capped trial: a few hundred characters, one voice, and a queue that never seems to move. This is not that. What you get instead is a published conversion rate between money and characters, and a one-time price you can plan around.
You do not have to pay to make an informed decision either. Before buying anything you can:
Starter is the entry point: 1,000 credits for a one-time $9, which works out to roughly 1,000,000 characters of generated speech. It includes:
None of that is a teaser tier. What the larger packs add is volume, queue priority, API access, and commercial usage rights, not a different quality of voice.
One credit converts to 1,000 characters. A 60 second voiceover of roughly 150 words is about 900 characters, so it costs about one credit. A ten minute narration of roughly 1,500 words lands near 9,000 characters, or about 9 credits. Under that rule of thumb the Starter pack covers a full audiobook chapter, a batch of explainer videos, or a few dozen short social clips before you need to think about topping up.
Because characters are the unit, the cheapest improvement to your balance is editing. Trim repeated phrasing, cut filler, and generate only when the script is final. Reusing one cloned voice across a series also saves time: the voice is set up once, and every later line is a straight generation.
If a script is long, it is worth splitting it into paragraphs before you generate. It keeps individual costs visible and makes it easy to regenerate a single paragraph instead of a whole chapter.
Personal projects are fine on Starter. Commercial usage rights start with the Creator pack at $29 for 4,000 credits, and the Studio pack at $99 for 15,000 credits extends that to higher volume, API access, and the fastest queue. All three are one-time purchases, and purchased credits stay valid for 12 months, so nothing expires while you are between projects.
If you are unsure which pack fits, start at $9 and measure what a real script costs you. The voice library and the Studio workflows pages show what larger projects look like in practice.
Not on this site. Every generation spends credits from a purchased pack. You can evaluate the product for free by browsing voices, reading the guides, and comparing packs, but audio generation starts at the $9 Starter pack.
Credits are valid for 12 months from purchase. They do not renew, and there is no subscription to cancel.
Starter covers 1,000 credits and personal projects. Creator covers 4,000 credits and adds commercial usage rights, which is what you need for client work and monetized content. Studio adds API access and the fastest generation queue on top of that.