cielvox2-tts-0.6b-base-GGUF

CielVox 0.6B Base — GGUF

Local-first multilingual voice-cloning TTS model for the cielvox2-tts backend in stelnetxcis-create/cielvox2.

• 10 supported languages: zh en ja ko de fr ru pt es it
• discrete multi-codebook LM architecture with a separate 12 Hz tokenizer / codec
• runtime voice cloning from (ref_audio, ref_text) or pre-baked voice-pack GGUFs
• Apache-2.0 licence

This repo contains the talker / code-predictor / speaker-encoder model. It must be used together with the separate tokenizer / codec GGUF from Xenna/cielvox2-tokenizer-12hz.

Files

Quick Start

Build the runtime:

git clone https://github.com/stelnetxcis-create/cielvox2
cd cielvox2
cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j$(nproc) --target stelnettts-lib

Download the talker + tokenizer:

huggingface-cli download Xenna/cielvox2-tts-0.6b-base-GGUF \
    cielvox2-tts-12hz-0.6b-base-q8_0.gguf --local-dir .

huggingface-cli download Xenna/cielvox2-tokenizer-12hz \
    cielvox-tokenizer-12hz.gguf --local-dir .

Voice clone from a reference WAV:

./build/bin/stelnettts \
    --backend cielvox2-tts \
    -m cielvox2-tts-12hz-0.6b-base-q8_0.gguf \
    --codec-model cielvox-tokenizer-12hz.gguf \
    --voice clone.wav \
    --ref-text "Exact transcript of clone.wav" \
    --tts "Hello there" \
    --tts-output hello.wav

Use a baked voice-pack GGUF:

./build/bin/stelnettts \
    --backend cielvox2-tts \
    -m cielvox2-tts-12hz-0.6b-base-q8_0.gguf \
    --codec-model cielvox-tokenizer-12hz.gguf \
    --voice my-voice-pack.gguf \
    --tts "Hello there" \
    --tts-output hello.wav

When --voice points to a .wav, --ref-text is required. When --voice points to a .gguf, it is treated as a baked voice pack.

Quantisation Notes

• cielvox2-tts-12hz-0.6b-base-q8_0.gguf — recommended quantised deployment
• cielvox2-tts-12hz-0.6b-base-q4_k.gguf — smaller variant; quality regression on short utterances

Architecture

Implementation

The C++ runtime is implemented in src/cielvox2_tts.cpp using ggml graphs for the talker / code-predictor path and the companion tokenizer GGUF for codec encode/decode.

Related

• Companion tokenizer / codec GGUF: Xenna/cielvox2-tokenizer-12hz
• C++ runtime: stelnetxcis-create/cielvox2

License

Apache-2.0.