# Netease Lyrics Vocab > Build vocabulary study tables and study websites from NetEase Cloud Music playlist lyrics. Use when the user provides a NetEase playlist URL or playlist ID and wants to fetch songs, collect lyrics, extract English words, count frequency, record source indexes by song and lyric line, add word meanings, export a CSV/XLSX table, or generate an interactive local vocabulary study page with word and example sentence pronunciation. - Skill: `honeylemonade/netease-lyrics-vocab` (Agent Skill, multi-file: 7 files) - Install (CLI): `npx skillmds@latest add honeylemonade/netease-lyrics-vocab` - Raw SKILL.md: https://api.skillmd.com/api/skills/honeylemonade/netease-lyrics-vocab/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: DevOps & Infra - Author: Honeylemonade (https://skillmd.com/u/honeylemonade) - Updated: 2026-09-22 - Page: https://skillmd.com/skills/honeylemonade/netease-lyrics-vocab --- # NetEase Lyrics Vocab Use this skill to turn a NetEase Cloud Music playlist into a vocabulary study table. ## Workflow 1. Identify the playlist ID from the user's URL or numeric ID. 2. Run `scripts/netease_lyrics_vocab.py` to fetch playlist tracks, fetch lyrics, tokenize English words, count frequency, and export the table. 3. Run `scripts/enrich_dictionary.py` to add built-in dictionary columns: phonetic, part of speech, Chinese meaning, Chinese explanation, English definition, root/affix, and memory note. 4. Inspect the generated table and fill or improve `meaning_zh` and `explanation_zh` for important words. Prefer concise Chinese meanings in the lyric context. 5. When the user wants a learning interface, run `scripts/build_vocab_site.py` against the enriched CSV/XLSX table to generate a static HTML study site. 6. Return the output file path and summarize coverage: songs processed, lyrics found, unique words, generated table, generated site, and any failures. ## Script Run from the skill directory or pass the full script path: ```bash python3 scripts/netease_lyrics_vocab.py "https://music.163.com/#/playlist?id=123456" --output vocab.csv ``` Useful options: - `--format csv|xlsx`: Force output format. Defaults from the output extension. - `--max-songs N`: Limit songs for a quick sample. - `--min-count N`: Keep only words that appear at least N times. - `--include-stopwords`: Keep common function words such as `the`, `and`, `you`. - `--define english`: Try to add short English definitions from dictionaryapi.dev. Keep `meaning_zh` for Codex or the user to fill. The script uses public NetEase web endpoints. These endpoints are unofficial and can change or rate-limit. If live fetching fails, ask the user for exported lyric files or a song list, then adapt the script input rather than inventing data. ## Dictionary Enrichment Add built-in dictionary fields to the table before building the site: ```bash python3 scripts/enrich_dictionary.py vocab.xlsx --output vocab.enriched.xlsx ``` When regenerating a playlist table, preserve previously filled dictionary fields with: ```bash python3 scripts/enrich_dictionary.py vocab.xlsx --previous old-vocab.enriched.xlsx --output vocab.enriched.xlsx ``` The enrichment script reads and writes CSV/XLSX. It uses a small built-in Chinese meaning map for common words, local root/affix rules, and public English dictionary lookup for phonetics, part of speech, and English definitions. Use `--offline` to skip public dictionary lookup. For high-quality Chinese explanations and deeper etymology, use the enriched table as the base and let Codex or another model fill `meaning_zh`, `explanation_zh`, `root_affix`, and `memory_note`; the study page embeds the completed values and works offline. ## Study Website Generate a static learning page from a finished table: ```bash python3 scripts/build_vocab_site.py vocab.xlsx --output vocab-study-site/index.html --title "歌词单词学习" ``` The generated page embeds the vocabulary and dictionary data and works as a local file. It provides: - Search by word, meaning, definition, or song. - Frequency filtering and known-word marking stored in browser local storage. - Word cards with meaning, phonetic, part of speech, Chinese explanation, English definition, root/affix, memory note, source indexes, source examples, and songs. - Browser speech synthesis buttons for the word and each example sentence. The page prefers English system voices and uses slower rates for clearer study playback. - Deduplicated examples per word, so repeated lyric lines do not crowd the card while frequency counts remain complete. For more natural voices, generate cached model-based MP3 files before building the page: ```bash OPENAI_API_KEY=... python3 scripts/generate_tts_audio.py vocab.xlsx \ --site-dir vocab-study-site \ --voice cedar \ --include-examples \ --max-examples 3 python3 scripts/build_vocab_site.py vocab.xlsx \ --output vocab-study-site/index.html \ --audio-manifest vocab-study-site/audio_manifest.json ``` The page plays generated MP3 files first and falls back to browser speech synthesis if audio is missing. Show the AI-generated audio disclosure when sharing the page with others. If the user asks for a shareable or portable result, return both the table and the generated `index.html`. Do not commit generated study pages unless the user explicitly wants examples stored in the repository. ## Output Columns See `references/output-schema.md` for column definitions. Preserve these columns unless the user requests a different layout: `word`, `count`, `phonetic`, `part_of_speech`, `meaning_zh`, `explanation_zh`, `definition_en`, `root_affix`, `memory_note`, `source_indexes`, `source_examples`, `source_songs`, `songs`. ## Meaning Rules - Fill `meaning_zh` with compact Chinese meanings, not long dictionary entries. - Use the lyric context where possible, especially for slang, contractions, idioms, and phrasal verbs. - Leave ambiguous words with multiple likely meanings separated by semicolons. - Do not output full lyrics. Keep only short snippets already produced by the script. ## Quality Checks - Confirm the playlist ID and output path. - Check that `count` is sorted descending. - Spot-check several `source_indexes`, `source_examples`, and `source_songs` rows for one-to-one alignment. - Report songs with missing lyrics or failed fetches.