跳到正文
cccskills
免费在 GitHub 公开

espnet

Use when working with ESPnet speech AI toolkit: installation, ESPnet2 recipes and training, pretrained inference/model zoo, ESPnet3 stage workflows, and repository testing.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill
查看安装方法

包含的文件（40）

  • SKILL.md3.9 KB
  • references/repo-provenance.md1.4 KB
  • references/repo-routing-metadata.json487 B
  • references/task-surface.md2.1 KB
  • references/troubleshooting.md2.9 KB
  • scripts/check_import_surface.py2.0 KB
  • sub-skills/development-and-testing/references/ci-command-map.md1.2 KB
  • sub-skills/development-and-testing/references/testing-and-contributing.md1.7 KB
  • sub-skills/development-and-testing/references/troubleshooting.md1.4 KB
  • sub-skills/development-and-testing/scripts/select_espnet_tests.py2.1 KB
  • sub-skills/development-and-testing/SKILL.md1.1 KB
  • sub-skills/espnet2-training/references/configuration-and-components.md1.3 KB
  • sub-skills/espnet2-training/references/gpu-distributed-training.md1.6 KB
  • sub-skills/espnet2-training/references/training-cli-reference.md2.1 KB
  • sub-skills/espnet2-training/references/troubleshooting.md1.6 KB
  • sub-skills/espnet2-training/scripts/inspect_training_cli.py1.7 KB
  • sub-skills/espnet2-training/scripts/make_dry_run_command.py1.5 KB
  • sub-skills/espnet2-training/SKILL.md1.2 KB
  • sub-skills/espnet3-workflows/references/configuration.md1.1 KB
  • sub-skills/espnet3-workflows/references/stage-runner.md1.2 KB
  • sub-skills/espnet3-workflows/references/troubleshooting.md1.2 KB
  • sub-skills/espnet3-workflows/scripts/inspect_espnet3_stages.py2.5 KB
  • sub-skills/espnet3-workflows/SKILL.md1.1 KB
  • sub-skills/inference-and-model-zoo/references/inference-api-reference.md2.2 KB
  • sub-skills/inference-and-model-zoo/references/model-zoo-and-packaging.md1.5 KB
  • sub-skills/inference-and-model-zoo/references/troubleshooting.md1.3 KB
  • sub-skills/inference-and-model-zoo/scripts/check_model_files.py2.2 KB
  • sub-skills/inference-and-model-zoo/scripts/inspect_inference_entrypoints.py3.1 KB
  • sub-skills/inference-and-model-zoo/SKILL.md1.1 KB
  • sub-skills/installation-and-diagnostics/references/install-matrix.md4.2 KB
  • sub-skills/installation-and-diagnostics/references/troubleshooting.md2.2 KB
  • sub-skills/installation-and-diagnostics/scripts/check_espnet_environment.py4.1 KB
  • sub-skills/installation-and-diagnostics/SKILL.md1.1 KB
  • sub-skills/recipes-and-data/references/espnet2-recipes.md2.3 KB
  • sub-skills/recipes-and-data/references/kaldi-data-formats.md1.8 KB
  • sub-skills/recipes-and-data/references/troubleshooting.md1.6 KB
  • sub-skills/recipes-and-data/references/utility-cli-reference.md1.5 KB
  • sub-skills/recipes-and-data/scripts/make_tiny_kaldi_fixture.py1.6 KB
  • sub-skills/recipes-and-data/scripts/validate_kaldi_data_dir.py3.8 KB
  • sub-skills/recipes-and-data/SKILL.md1.2 KB

SKILL.md（原文）

安装前可以查看交给 AI 代理的指令内容。

ESPnet Repo Skill

Use this operating skill when a task involves ESPnet, the end-to-end speech processing toolkit for ASR, TTS, speech translation, enhancement/separation, speaker tasks, diarization, SLU, SVS, SpeechLM, and related audio workflows.

This skill is self-contained. Do not assume the source checkout used to create it is available unless the user is explicitly editing an ESPnet checkout. Prefer the bundled references and scripts here over reopening original docs, examples, or tools.

First response pattern

  1. Classify the request: install/diagnose, prepare recipe data, train/configure, run inference/model-zoo, use ESPnet3 stages, or develop/test ESPnet.
  2. Read the matching sub-skill before detailed commands.
  3. Use bundled helper scripts when they fit. They are safe by default: no downloads, no package installs, no training, no uploads.
  4. Keep backend claims honest. CPU parser/import checks do not prove CUDA, distributed, FlashAttention, k2, or recipe-scale training.

Minimal package check

python -c "import espnet2, espnet3; print('ESPnet imports ok')"

For source development or editable use, install only the selected task extras, e.g. pip install -e ".[asr]" for ASR or pip install -e ".[tts]" for TTS. Avoid .[all] unless the user truly needs many optional task families.

Route map

User intentRead nextWhy
Install ESPnet, choose extras, diagnose imports or optional toolsinstallation-and-diagnosticsBase dependencies, extras, host tools, CUDA probes, safe environment checker.
Create/adapt ESPnet2 recipes, validate data/, understand wav.scp/segments, stage commands, tokenization utilitiesrecipes-and-dataKaldi-style data layouts, task script stage flow, validation helpers, utility CLIs.
Configure/train ESPnet2 models, use --print_config, dry-run configs, tune components, resume/fine-tune, GPU/distributed flagsespnet2-trainingTrain modules, Task config semantics, dry-run command generation, training troubleshooting.
Run pretrained or local inference, use from_pretrained, ModelDownloader, streaming ASR, enhancement, TTS, packaginginference-and-model-zooInference classes/CLIs, model file pairing, model zoo, packaging checks.
Use ESPnet3 System/Hydra stage runner, configs, --stages, demo/publication boundariesespnet3-workflowsESPnet3 stage order, required config flags, safe stage inspection.
Modify ESPnet source, choose tests, debug CI, add recipe/module tests, follow contribution conventionsdevelopment-and-testingFocused pytest/CI command selection, style, recipe PR policy, maintainer troubleshooting.

Shared runtime files

Safety boundaries

  • Ask before full recipes, dataset acquisition, model downloads, Gradio demos, uploads, long training, or broad CI.
  • CUDA is optional unless the user's selected workflow requires GPU runtime.
  • Model-zoo tasks are network/cache dependent; validate local file paths before downloads.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护，适用 Google 的隐私权政策和服务条款。

同一仓库的技能

3d-resnets-pytorch

免费

Routes 3D ResNets PyTorch video action-recognition workflows across training, inference, and data preparation.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill3392026年9月3日 更新

3ddfa

免费

Guide 3DDFA Python inference, geometry rendering, training/evaluation, and optional C++ ONNX workflows for 3D dense face alignment.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill3392026年9月3日 更新

3ddfa-v2

免费

Routes 3DDFA_V2 face-alignment setup, still-image demos, video tracking, and ONNX benchmarking workflows.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill3392026年9月3日 更新

ab3dmot

免费

Operate AB3DMOT 3D multi-object tracking workflows for KITTI and nuScenes data, tracking, evaluation, and visualization.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill3392026年9月3日 更新

accelerate

免费

Use Hugging Face Accelerate for PyTorch training-loop migration, distributed launch/configuration, DeepSpeed/FSDP/TPU backend setup, big-model inference/offload, checkpointing, tracking, and troubleshooting.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill3392026年9月3日 更新

acme

免费

Route Acme reinforcement-learning framework tasks across core loops, replay/data, JAX agents, and TensorFlow/Sonnet agents.

中文简介正在准备中。当前显示原文说明。

VectorSpaceLab/AREX-Skill3392026年9月3日 更新

查看 VectorSpaceLab 的全部技能

报告此技能的问题