ParaASR: Multi-Token Prediction for Fast and Long-Context LLM-Based Speech Recognition | Digital Library | PAMCET | PAMCET