inno-kokoro 0.2.0

Zero-shot voice tuner for Kokoro-82M.

A 5-30 second reference clip in, a stock-shaped [510, 1, 256] voice pack out in about 0.3 seconds.

Usage

pip install inno-kokoro
inno-kokoro --fetch /models # -> /models/model.safetensors, e.g. in a Dockerfile; skips if present
inno-kokoro my_ref.wav am_me # -> voices/am_me.pt + voices/am_me_test.wav
inno-kokoro my_ref.wav am_me --fmax 300 # override pitch ceiling

Currently only available for English; prefixes work like the stock packs e.g: af_, am_, bf_, bm_.

from inno_kokoro.enroll import Tuner, enroll, read
from kokoro import KPipeline

tuner = Tuner()  # huggingface cache; or Tuner("/models/model.safetensors") for a path of your own
pack, _ = enroll(*read("my_ref.wav"), tuner)
pipe = KPipeline(lang_code="a")
wav = next(pipe("Hello from a tuned voice.", voice=pack)).audio

For best results, the reference audio should be:

• 3-second minimum, up to the first 30-seconds
• Single speaker (english).
• Reasonably clear of audio artifacts

Voice pack generation time:

• about 0.05 s per second of reference on CPU
• 0.1 to 0.3 s total on a GPU (after the model is loaded).

Identity benchmarking

LibriSpeech test-clean, F5-TTS cross-sentence split: 1127 utterances, 39 held-out speakers.

Reference in, new sentence out, scored against the speaker's real recording. Normalized scores compare the render between sounding like a stranger (0) and a second sample of the same benchmark speaker (1).

RTF on an RTX 4060 Ti.

UTMOS scoring stays high which reflects the priority placed on maintaining Kokoro's voice quality.

Inno can match about a third of the way to most identities, and avoids copying recording artifacts etc.

License and Attributions

Apache-2.0. The speaker encoder is CC BY-SA 3.0.

Full model card available on HuggingFace: remsky/kokoro-inno-clone-tuner.

Only clone voices you have permission to clone, even the shallow cloning provided by this model and technique.