计划执行与检查节点
executing-plans
按既定实施计划完成工作,保持过程检查和反馈。
inference-and-model-zoo【效率与协作】Use ESPnet pretrained and local inference APIs, model-zoo tags, streaming classes, model file checks, and packaging workflows.
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Use this sub-skill for pretrained model inference, local config/checkpoint inference, from_pretrained, ModelDownloader, streaming ASR, TTS vocoders, enhancement/separation, diarization, speaker embeddings, SVS, and packaging.
Route package/backend issues to ../installation-and-diagnostics/SKILL.md, data/recipe stage setup to ../recipes-and-data/SKILL.md, and training config creation to ../espnet2-training/SKILL.md.
Do not run full recipes, model downloads, training jobs, uploads, demo servers, broad CI, or host-mutating installers without explicit user approval. Keep local/private environment details out of public answers.