Logo Lanfrica

gabar-tech/cosyvoice2-amharic

Domain:

natural language processing

Record type:

model
Creator:
gab
Host:
Amharic fine-tune of CosyVoice2-0.5B, with inference code, paired demos, evaluation results, and audit files. --- language: - am license: cc-by-sa-4.0 base_model: FunAudioLLM/CosyVoice2-0.5B base_model_relation: finetune pipeline_tag: text-to-speech datasets: - google/WaxalNLP - mozilla-foundation/common_voice_25_0 tags: - text-to-speech - amharic - ethiopia - voice-cloning - cosyvoice --- # CosyVoice2 Amharic This release fine-tunes the text-to-token and token-to-mel stages of CosyVoice2-0.5B (FunAudioLLM / Alibaba, Apache-2.0) for Amharic. It keeps the base model's reference-audio voice cloning, which uses about ten seconds of audio. The model produces intelligible Amharic; frequent hesitation sounds are its main quality limitation. The tokenizer represents Ge'ez text without unknown tokens, but the stock checkpoint did not produce intelligible Amharic in our tests. The vocoder, speech tokenizer and speaker encoder are unchanged. `amharic_text.py` is the text normalizer used for training labels and at inference. ## Hear it Each pair uses the same sentence, reference audio, settings and seed. | # | Test case | Stock CosyVoice2 | + Gabar fine-tune | |:-:|---|:-:|:-:| | 1 | Ordinary prose | | | | 2 | Prose, ፥ punctuation | | | | 3 | Prose (ejectives ቡ/ጅ) | | | | 4 | Numbers + ዓ.ም. date abbreviation | | | | 5 | ዶ/ር title abbreviation | | | | 6 | Question intonation | | | | 7 | Technical prose | | | | 8 | Mixed punctuation + question | | | The players render on Hugging Face only. On GitHub the clips are in `demo/` — stock: 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8; fine-tuned: 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8. The weights (`llm.pt`, `flow.pt`, ~2 GB) are on Hugging Face; everything else is mirrored on GitHub. Texts: `demo/sentences.txt`. Reference voice: `demo/reference.wav`, a consenting member of the Gabar team ( ). Both models got the text after `amharic_text.normalize`, cross-lingual mode (reference audio only), seed 1234. No outputs were selected from multiple attempts. Known weaknesses are under Limitations. ## Files in this re …