WeChat Style DistillerStyle Distiller
HomeTourPipelinePersonasFeaturesValidationSimilarityMatch QuizPrompt TestPersona TunerAI LogMethodsPrivacyInstall

Synthetic Persona Gallery

Six pre-generated personas demonstrating diverse linguistic patterns. Each was created from 150+ synthetic messages and analyzed with the original pipeline.

Bilingual Early Adopter

Tech-savvy, EN/ZH code-switching 40%, emoji-heavy, casual tone

Avg Length
25 chars
Code-Switch
40%
Emoji Rate
0.30/msg
Style
casual_tech

Concise Mandarin

ZH only, short messages (8-15 chars), rare emoji, formal punctuation

Avg Length
12 chars
Code-Switch
0%
Emoji Rate
0.05/msg
Style
formal_zh

Expressive English

EN only, long messages (40-80 chars), reaction emoji, casual punctuation

Avg Length
60 chars
Code-Switch
100%
Emoji Rate
0.25/msg
Style
expressive_en

Balanced Bilingual

EN/ZH 50/50, medium length, moderate emoji, professional tone

Avg Length
35 chars
Code-Switch
50%
Emoji Rate
0.15/msg
Style
professional

Emoji Enthusiast

ZH primary, heavy emoji (50%+ messages), short bursts, playful

Avg Length
18 chars
Code-Switch
10%
Emoji Rate
0.60/msg
Style
playful_zh

Technical Researcher

EN-heavy code-switching, long analytical messages, minimal emoji, jargon

Avg Length
55 chars
Code-Switch
70%
Emoji Rate
0.08/msg
Style
academic

Personal project by Sunchuangyu Huang · 2026

Privacy-first NLP tool. Your data stays on your machine.

View on GitHub