Umbaji/cey-25-mms-tts-mozilla-kdh

MMS-TTS Tem / Kotokoli — Mozilla Fine-tuned

Fine-tuned facebook/mms-tts-kdh on Mozilla Data Collective data for Tem / Kotokoli (kdh).

Training Statistics

Metric	Value
Training samples	8
Validation samples	2
Best val mel-L1	4.7781
Dataset source	Mozilla Data Collective

Usage

from transformers import VitsModel, VitsTokenizer
import torch, torchaudio

model     = VitsModel.from_pretrained("Umbaji001/cey-25-mms-tts-mozilla-kdh")
tokenizer = VitsTokenizer.from_pretrained("Umbaji001/cey-25-mms-tts-mozilla-kdh")

inputs = tokenizer("your text here", return_tensors="pt")
with torch.no_grad():
    waveform = model(**inputs).waveform[0]

torchaudio.save("out.wav", waveform.unsqueeze(0), model.config.sampling_rate)

Training

• Loss: Mel-spectrogram L1
• Optimizer: AdamW (lr=2e-4, betas=(0.8, 0.99))
• Scheduler: ExponentialLR γ=0.999
• Epochs: 6 | Effective batch: 16

Fine-tuned: 2026-02-25 — Mozilla Data Collective + Eyaa-Tom