lyrics-translator

name	lyrics-translator
description	Translate lyrics while preserving musical constraints (syllable counts, rhyme scheme, syllable patterns). Use when the user wants to translate song lyrics from one language to another while maintaining singability.

Lyrics Translator

Translate lyrics between languages while preserving syllable counts, rhyme scheme, and word-level syllable patterns so the translation remains singable.

Workflow

Follow this 3-phase pipeline. Use the sub-skill scripts to verify at each step.

Phase 1: Extract Constraints

Before translating, analyze the source lyrics to extract targets:

1. Count syllables per line:

   python skills/syllable-counter/scripts/count_syllables.py "<line>" "<source_lang>"
   

2. Get syllable patterns (word-level distribution):

   python skills/syllable-pattern-analyzer/scripts/pattern_analysis.py "<source_lang>" "<line1>" "<line2>" ...
   

3. Detect rhyme scheme:

   from rhyme_analysis import detect_rhyme_scheme
   scheme = detect_rhyme_scheme(lines, source_lang)
   

Record the target syllable counts, patterns, and rhyme scheme.

Phase 2: Initial Translation

Translate all lines at once, considering all 3 constraints:

• Each line must have EXACTLY the target syllable count.
• Lines sharing a rhyme label must rhyme in the target language.
• Word-level syllable distribution should match the target pattern.
• Maintain poetic quality and emotional impact.

Key rules for Chinese targets:

• Each Chinese character = 1 syllable.
• Strip all punctuation when counting.

Phase 3: Iterative Refinement

For each line, verify and refine:

1. Check syllable count:

   python skills/syllable-counter/scripts/count_syllables.py "<translation>" "<target_lang>"
   

2. If count is wrong, adjust the translation:

   • Too long: remove adjectives, use shorter words, merge concepts.
   • Too short: add descriptive words, use longer alternatives.
   • Re-check after each adjustment. Repeat up to 10 times.

3. Check syllable patterns:

   python skills/syllable-pattern-analyzer/scripts/pattern_analysis.py "<target_lang>" "<translated_line>"
   

   • Skip refinement if average pattern similarity is already >75%.
   • Only accept pattern changes that improve similarity by >15%.

4. Verify rhyme preservation after pattern adjustments using the rhyme-analyzer.

Final Validation

Run the full validator on all translated lines:

python skills/lyrics-validator/scripts/validate.py "<target_lang>" '<target_syllables_json>' "<line1>" "<line2>" ...

Output Format

Present the final translation as:

1. <translated line 1>
2. <translated line 2>
...

Syllable accuracy: X/Y lines match
Pattern similarity: X%
Rhyme scheme: <scheme>

Sub-Skills Used

• syllable-counter - count syllables
• rhyme-analyzer - rhyme detection
• syllable-pattern-analyzer - word rhythm patterns
• lyrics-validator - constraint verification
• phonetic-analyzer - IPA and phonetic similarity

Configuration Reference

See references/config.md for language codes and defaults. See references/graph-pipeline.md for pipeline architecture details.