data-preparation
Routes VITS dataset preparation, filelist cleaning, and monotonic-alignment build tasks.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 11
- Forks
- 3
01 前言 Front Matter
- name
- data-preparation
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Routes VITS dataset preparation, filelist cleaning, and monotonic-alignment build tasks.
- disable-model-invocation
- true
02 资源文件 1
- 参考文档references/troubleshooting.md1.2 KB
03 SKILL.md 正文
Data preparation
Use this route for custom corpus setup, filelist cleaning, and build-time prerequisites that must happen before training or inference.
Use this route when
- You need to clean raw text filelists into the repo's
.cleanedformat. - You need to match the LJ Speech or VCTK filelist layouts.
- You need to verify dataset sampling rates or speaker-id columns.
- You need to build the monotonic-alignment extension before importing
models. - You need to diagnose
phonemizer,espeak, or filelist parsing issues.
Do not use this route when
- You are actually launching training; use
training. - You are generating audio from a checkpoint; use
inference. - You only need a quick environment check; use
../../scripts/check_install.pyfrom the root skill.
Read first
../../references/configuration.mdfor the filelist formats and config-field mapping.../../references/workflows.mdfor the data-preparation command flow.../../references/troubleshooting.mdfor cross-cutting failures.references/troubleshooting.mdin this sub-skill for filelist- and cleaner-specific failures.
Bundled helpers
../../scripts/preprocess_text.py— clean one or more filelists.../../scripts/build_monotonic_align.py— build the nested extension layout expected bymonotonic_align.../../scripts/check_install.py— confirm the repo imports before you move to training or inference.
Common workflow
- Confirm the dataset layout and audio sampling rate.
- Decide whether the repo's provided cleaned filelists are already enough.
- Run
../../scripts/preprocess_text.pyonly for custom raw filelists. - Build
monotonic_alignwith../../scripts/build_monotonic_align.pybefore any model import. - Re-run
../../scripts/check_install.pywhen the data or build step changes.
Ownership boundaries
- Include text normalization, filelist column mapping, and dataset-layout validation here.
- Exclude training launch, checkpoints, and synthesis outputs; route those to sibling sub-skills.
04 同仓库更多 Skills
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
obsidian-clipper-template-creator
Guide for creating templates for the Obsidian Web Clipper. Use when you want to create a new clipping template, understand available variables, or format clipped content.
python-api
Programmatic Object-Detection-Metrics API guidance for BoundingBox, BoundingBoxes, Evaluator, enums, coordinate conversion, AP/mAP semantics, and source-style import diagnostics.
file-evaluation
Folder-based PASCAL VOC AP/mAP evaluation from ground-truth and detection text files.
object-detection-metrics
Guides legacy Object-Detection-Metrics PASCAL VOC AP/mAP evaluation from detection text folders and direct Python API classes.
nx-workspace-patterns
Configure and optimize Nx monorepo workspaces. Use when setting up Nx, configuring project boundaries, optimizing build caching, or implementing affected commands.