KitsuMate/omnivoice-onnx

OmniVoice ONNX for KitsuMate

Provider-neutral OmniVoice model profiles curated for the KitsuMate ONNX Unity runtime.

Profiles

• cpu-merged-int4: compact bidirectional merged backbone using symmetric 4-bit weight-only quantization with 128-weight blocks and the standard FP32 Higgs codec.
• portable-merged-fp32: unquantized merged graph using standard ONNX opset 17 operators. It is intended for any ONNX Runtime execution provider that can successfully create all sessions; it is not tied to CUDA.
• codec-fp32: portable Higgs Audio V2 codec for the FP32 profile.

The upstream FP16 semantic codec is intentionally not mirrored: stock ONNX Runtime 1.24.4 rejects its mixed FP32/FP16 LayerNormalization contract during session creation. Keeping only the working FP32 codec avoids presenting an artifact as portable when it is not.

The upstream split INT4 and CPU-FP16 language graphs are causal and failed speech-recognition review. The compact profile is instead quantized directly from the working merged FP32 graph with ONNX Runtime 1.24.4 MatMulNBits, preserving bidirectional attention. Whisper checks recovered the requested words for auto voice, instructed/tagged voice design, and voice cloning.

The runtime implements 32-step iterative masked decoding, classifier-free guidance, auto voice, voice cloning, voice design, language conditioning, and OmniVoice inline controls. See omnivoice-manifest.json for exact roles, revisions, and hashes.

Attribution

The codecs are derived from onnx-community/OmniVoice-Onnx. The merged FP32 graph and its KitsuMate-quantized INT4 derivative are based on gluschenko/omnivoice-onnx. OmniVoice was created by the k2-fsa contributors; see k2-fsa/OmniVoice and ATTRIBUTION.md.

No provider compatibility is claimed merely from the profile name. A provider is supported only when it successfully creates every required session and completes inference.