All models
lightweight

Kokoro-82M

Apache-2.0

82M parameters, preset voices and CPU inference

Kokoro is an 82-million-parameter speech model with preset voices. Listen to the same three scripts across eight English voices below, or open the voice library for hosted settings, sample provenance and a direct studio link. CPU inference is supported; speed depends on your hardware and runtime.

Hear Kokoro-82M

Kokoro-82M · Bella · Same three scripts across the directory.

Hear all 8 voices with matching studio presetsRun it yourselfTry your own text

Best for: Lightweight narration and app voices when preset voices are enough.

Permissive model license: Model terms allow commercial deployment. Reference voices and training acknowledgements remain separate; the hosted English presets are identified individually.

Code: Apache-2.0. Weights: Apache-2.0. License sources checked 2026-09-20. Source 1Source 2

Before you choose: No voice cloning. Language and voice quality vary; test names, numbers, and longer passages.

CPU inference is supported. Speed depends on processor, text length, and runtime; the old speed estimate is not a standardized benchmark.

Details reviewed 2026-09-19. Hardware figures are estimates unless a benchmark is linked.

Compare with similar

Compare all →
PiperListen side-by-side →MeloTTSListen side-by-side →KittenTTSListen side-by-side →

Choose an alternative to hear both models together.

Voices

8 voices
Female
Belladefault
american
Use Bella in the paid studio →
Nicole
american
Use Nicole in the paid studio →
Emma
british
Use Emma in the paid studio →
Sky
american
Use Sky in the paid studio →
Male
Michael
american
Use Michael in the paid studio →
Adam
american
Use Adam in the paid studio →
George
british
Use George in the paid studio →
Onyx
american
Use Onyx in the paid studio →

The scripts

Neutral
“The quick brown fox jumps over the lazy dog near the riverbank.”
Emotional
“I can't believe you actually did it. This is incredible!”
Numbers & Dates
“On March 14th, 2025, the team raised $4.2 million at a 38% margin.”

Run Kokoro-82M

pip install kokoro

First audio example

Based on the author’s API. Install the documented dependencies and choose a compatible Python environment first.

# Install system dependency espeak-ng if required by your OS.
# pip install kokoro soundfile
from kokoro import KPipeline
import soundfile as sf
pipeline = KPipeline(lang_code="a")
for i, (_, _, audio) in enumerate(pipeline("Hello from OpenSpeech.", voice="af_bella")):
    sf.write(f"speech-{i}.wav", audio, 24000)

Sources & setup details

Kokoro-82M FAQ

Do I need a GPU to run Kokoro-82M?

Kokoro-82M supports CPU inference. Speed depends on your processor, runtime, and text.

Can Kokoro-82M clone voices?

No — Kokoro-82M uses preset voices. Browse the voice-cloning category for models that clone from reference audio.

What license is Kokoro-82M released under?

Kokoro-82M is released under the Apache-2.0 license. Always check the repository for the exact terms — some models license code and weights separately.

What languages does Kokoro-82M support?

Kokoro-82M supports English, Japanese, Chinese, Spanish, French, Hindi, Italian, Portuguese.

Is there a hosted Kokoro-82M API?

Yes. OpenSpeech Cloud offers Kokoro-82M in its paid English preset-voice studio and API, using prepaid credits. See the Cloud documentation for current availability.