OmniVoice Igbo (experimental)
Experimental research checkpoint -- not part of the main WaxalNet release suite.
OmniVoice finetuned on an adaptive-CER-threshold (<=0.161, the language's own median) cleaned Igbo subset (336.9h), 5000 steps.
Languages: Igbo (ig)
License note
This model is finetuned from k2-fsa/OmniVoice, whose pretrained weights are licensed CC-BY-NC-4.0 (non-commercial) due to constraints from its Emilia training data. This finetune inherits that restriction: research use only, not for commercial use.
Training recipe
- Base: k2-fsa/OmniVoice (Qwen3-0.6B backbone + Higgs Audio v2 codec)
- Data: WAXAL Igbo corpus, cleaned via the project's Phase 1 pipeline, with an adaptive CER ceiling (floored at 0.08, capped at 0.3, relaxed to the language's own median CER) to avoid over-pruning low-ASR-confidence but otherwise valid clips.
- 5000 training steps (bf16, lr 1e-5), OmniVoice's default finetune recipe.
This is a fast research iteration, not a polished release -- quality has not been formally evaluated against the project's main VoxCPM2 models.
- Downloads last month
- 45