Aoi AI Voice
Japanese, storytelling. Type your text below and generate natural speech with Aoi — then download the MP3.
Speech studio
29 voices · 18 languages
Examples
Hear Aoi on real scripts
Each clip below was generated with a free-tier voice on the script shown. Play one to hear how this kind of writing lands, or load the script into the generator above and hear Aoi read it.
Phone greeting
IVR or voicemail line — the classic use for a consistent voice.
“Thank you for calling Northgate Supply. Our team is available Monday to Friday, eight until six. For an existing order, press one. For everything else, stay on the line and someone will be with you shortly.”
Product ad
Thirty seconds of ad copy, the hardest thing to read naturally.
“You already own the shoes. That's the annoying part. They're at the back of the cupboard, they still fit, and the only thing standing between you and Saturday morning is the fifteen minutes it takes to find them. We can't help with that. But everything after it, we can.”
Article to audio
Turn something written to be read into something worth listening to.
“Reading aloud is the oldest editing tool there is. The eye forgives a sentence that has no breath in it; the ear refuses. Which is why writers who read their own drafts out loud tend to produce shorter sentences, and why anyone about to publish something long should hear it first.”
About the Aoi voice
Aoi has more shape to her reading than Yuki — a little more rise and fall, a little more room between phrases. That makes her the better of the two for anything with a narrative in it, where the pauses are doing as much work as the words.
Why a Japanese voice, and not an English one
Aoi was trained on Japanese, which matters more than it sounds like it should. An English voice pointed at Japanese text will get the individual sounds roughly right and the rhythm completely wrong — stress in the wrong places, vowel lengths flattened, the melody of the sentence replaced with an English one. It is intelligible and it is immediately, unmistakably foreign.
Japanese is also read here by Yuki, Haruto — worth previewing side by side, since the difference between two voices in the same language is usually more useful than the difference between two languages.
The engine behind this voice
Kokoro is an 82-million-parameter open-weights model under Apache-2.0 — the cheapest capable engine available by roughly an order of magnitude, which is what makes this tier free at all. It takes up to 5,000 characters per request, adjusts from 0.75× to 1.5×, and returns MP3 or WAV. Longer text is split at sentence boundaries and joined back into one file.
More about Kokoro TTS →What Aoi costs
Nothing. Aoi is a free-tier voice: 5,000 characters a day without an account, 10,000 with a free one, and no credit card at any point. Audio downloads as MP3 or WAV and carries a commercial licence, so monetised videos and client work are both fine.
The daily allowance is stated plainly here because several sites in this category advertise "unlimited" and enforce a monthly cap. What is written above is what the server enforces.
Best for
What Aoi is good for
- Japanese storytelling, fiction and narrative video
- Audiobook-style Japanese passages and long descriptive prose
- Character-led or dialogue-heavy Japanese scripts
- Two-voice Japanese projects alongside Yuki
Limits
Where Aoi is the wrong choice
Her extra shape is a liability on flat informational text, where the rise and fall reads as unwarranted drama. Use Yuki for anything procedural, and Aoi where there is actually a story.
How it works
How to use Aoi
- 1
Paste your text
Type or paste the script you want Aoi to read. Longer text is split into parts automatically and joined back into one file.
- 2
Preview Aoi
The generator above already has this voice selected. Play the sample, or generate a sentence of your own text to hear it on your writing.
- 3
Generate and download
Audio comes back in seconds as MP3 or WAV, with a commercial licence on every tier — including the free one.
Questions