Collections

Bangla (Bangladesh) speech-to-text: a Voxtral Mini fine-tune with a live demo, plus FastConformer, Qwen3-ASR and Whisper models.

Models that predict reaction products by reasoning: Qwen3-8B and Qwen3.8-27B distilled from Kimi K3, then self-trained. With all data.

Bangla text-to-speech: an IndicF5 fine-tune for a Bangladeshi voice and an emotional Orpheus model.

Fine-tuned sentence-embedding models for retrieval and similarity.

Datasets for terminal agents, reward-hacking audits and Sinhala speech emotion.