In terms of acoustic model self-evolution, it achieves accent adaptation, noise adaptation, and non-native pronunciation and accent adaptation for speech recognition. It also enabled speech clone with less than 10-second training speech data. In terms of language model self-evolution, it achieves ten-thousand-level hot-word self-learning within seconds, domain-transfering of language model based on sparse knowledge representation; it also achieves intent expansion and intent recognition based on large  language model prompt engineering. In terms of AI agent self-evolution, based on a general large language model, it achieves task-oriented self-learning, model self-optimization, domain knowledge self-learning, and general model data extraction.