算法艺术创作
algorithmic-art
通过代码和可调参数创作生成艺术与交互式视觉效果。
text-processing【设计与创意】Use PaddleSpeech punctuation restoration, text frontend, G2P, text normalization, MFA, and tokenizer preparation workflows safely.
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Use this sub-skill for paddlespeech text --task punc, punctuation restoration models, text frontend preparation, G2P, text normalization, MFA/rhythm-tag orientation, and SentencePiece tokenizer recipe planning.
references/punctuation-and-frontends.md and scripts/prepare_punctuation_input.py.../speech-to-text/SKILL.md, then use punctuation restoration here.../text-to-speech/SKILL.md for synthesis.../deployment-serving/SKILL.md after choosing the punctuation model/config.python scripts/prepare_punctuation_input.py --text 今天的天气真不错啊你下午有空吗我想约你一起去吃饭
paddlespeech text --task punc --model ernie_linear_p3_wudao_fast --input 今天的天气真不错啊你下午有空吗我想约你一起去吃饭
The punctuation executor cleans unsupported characters and asserts that the remaining text is non-empty. Validate before running a model download.
references/punctuation-and-frontends.md covers punctuation CLI/API, ERNIE model choices, text frontend tools, G2P, TN, MFA, and SentencePiece recipe boundaries.references/troubleshooting.md covers unsupported input, model/tokenizer downloads, AIStudio/PaddleNLP mismatch, MFA/Kaldi side effects, and job parsing.scripts/prepare_punctuation_input.py validates and cleans punctuation input without loading PaddleSpeech models.