Kokoro model, setup and specifications

Find your Kokoro voice

Eight English voices, the same three scripts. Listen for pacing, pronunciation and the tone your recording needs, then carry that exact voice into the paid studio.

Listening is free. Custom recordings use prepaid credits from $5, with a $0.10 minimum per recording. English presets at normal speed; no voice cloning.

Bella

American English · Female preset

af_bella

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Bella in the paid studio →

Nicole

American English · Female preset

af_nicole

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Nicole in the paid studio →

Michael

American English · Male preset

am_michael

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Michael in the paid studio →

Adam

American English · Male preset

am_adam

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Adam in the paid studio →

Emma

British English · Female preset

bf_emma

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Emma in the paid studio →

George

British English · Male preset

bm_george

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use George in the paid studio →

Sky

American English · Female preset

af_sky

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Sky in the paid studio →

Onyx

American English · Male preset

am_onyx

The quick brown fox jumps over the lazy dog near the riverbank.

I can't believe you actually did it. This is incredible!

On March 14th, 2025, the team raised $4.2 million at a 38% margin.

Use Onyx in the paid studio →

How to choose from the samples

Start with the neutral sentence to compare pacing without changing the words. Use the emotional sentence to hear how each preset handles excitement. The numbers example is a separate pronunciation check: dates, currency and percentages may need rewriting in your own script. A longer sample does not necessarily mean clearer speech.

For an American-English delivery, compare Bella, Nicole, Michael, Adam, Sky and Onyx. For British-English delivery, compare Emma and George. These are selectable presets, not controls for copying a real person’s voice. Samples help you choose a preference; they are not a ranking or an accuracy guarantee.

What will your recording cost?

Kokoro costs 15 credits per 1,000 characters, rounded up, with a 10-credit minimum for each recording. One credit is $0.01. A 100-character recording is $0.10; a 1,000-character recording is $0.15. The minimum credit-pack purchase is $5. Switching these voices does not change the rate.

The studio accepts up to 2,000 characters and 4,000 UTF-8 bytes per recording. It shows the selected voice and price before you generate. New recordings remain privately downloadable for seven days. Calculate the price of your actual script.

Same settings from sample to studio

These 24 examples were generated on September 19, 2026 using the studio’s pinned Replicate adapter, each named voice ID, and speed 1. They replace older Kokoro examples whose checkpoint provenance was incomplete. Provider outputs can vary between runs; the link preserves the voice and settings, not identical audio bytes.

Audio format, duration and file integrity were checked for all samples. Additional Bella checks covered a longer passage and the 2,000-character limit. A full human pronunciation review is still pending. Review the audible examples against your own requirements before creating paid audio.

Sample provenance and sources

Provider: jaaari/kokoro-82m · WAV · 24 kHz · speed 1

Pinned adapter version: f559560eb822dc509045f3921a1921234918b91739db4bf3daab2169b71c7a13

The model card lists Apache-2.0. Model, provider and content terms still apply; this page does not grant rights to scripts or imply endorsement by the model authors.

Compare Kokoro with Chatterbox Turbo · Run Kokoro yourself