Wordagen

Generate unique names for characters, online aliases, places, products, pets, brand names, hostnames or other realistic-sounding but original words.

Two generation methods available:

• Syllable-based: Fast generation using phonetic rules (default)
• Markov chains: More realistic words trained on dictionary data (auto-enabled by Markov options)

Installation

git clone https://github.com/j33433/wordagen.git
cd wordagen
python wordagen.py

Example Output

# Batch of unique names generated by Markov chains trained on US name lists
python wordagen.py --name --order=4 --count=10
Carletha Widden
Jeanmarilou Binnis
Millis Carottsch
Talita Butchel
Maryroseanna Frisker
Fletchelle Gloeck
Merlie Vanhollard
Marthey Battison
Laurenae Tucken
Latashida Santanas

# Spanish nonsense words (Hunspell dictionary with morphological expansion)
python wordagen.py --words=es --order=4 --count=20
exime         estomano      mación        faneguillar   anecer      
liviador      almente       ladrilar      fullecedor    mascaloide  
calencia      alfalfeta     huélano       sufrar        melodra     
indonero      percos        manos         emparecen     seminencia  

# Names starting with a prefix (--prefix automatically enables Markov mode)
python wordagen.py --words=names --prefix=joe --order=2 --count=10 --length=5-20
joemardoretta  joell          joellys        joemikaridy    joeston      
joelana        joelie         joellina       joeminicollia  joelanne     

# Words ending with common suffixes (--suffix automatically enables Markov mode)
python wordagen.py --suffix=ing --count=10
rementing     naling        glanning      twing         reminereping
istricarming  sumbetauding  rumming       drobberring   scaring     

# Words with both prefix and suffix
python wordagen.py --prefix=pre --suffix=ing --count=5 --order=4 --length=8-15
prewheretting   prewoodcurring  prewrapping     pretening       prewashinning 

# Syllable-based simple algorithm (default)
python wordagen.py --single
drubriebreat

# Latin tokens word-word-word (--order automatically enables Markov mode)
python wordagen.py --token --order=2 --words=la
sonis-calita-coris

# Word blends
python wordagen.py --blend=hippopotamus,godzilla --count=8 --order=4
hipilla
godzillamus
hippodzilla
godzillus
hippola
godotamus
godamus
hippopotilla

Parameters

Length Control (--length)

• --length=MIN-MAX: Range (e.g., 5-8)
• --length=N: Exact length (e.g., 10)

Markov Chain Options

Markov mode is automatically enabled when any of these options are used:

Order (--order, default: 2)

• 1: Most creative, often unpronounceable
• 2..3: Good balance
• 4..6: More realistic

Word Lists (--words, default: "en")

• Language dictionaries (Hunspell): en, es, fr, de, it, pt, ar, bg, ca, cs, etc.
• Special word lists: names, surnames, pet
• Custom URLs: https://example.com/wordlist.txt
• Use --list to see all available options (50+ languages supported via Hunspell)

Cutoff (--cutoff, default: 0.1)

• 0.0: Include all transitions (most random)
• 0.1: Filter rare patterns (balanced)
• 0.5: Conservative, predictable output

Prefix (--prefix)

• Start generated words with a specific prefix
• Example: --prefix=steve generates words like "stevenson", "stevie"

Suffix (--suffix)

• End generated words with a specific suffix
• Can be combined with --prefix
• Example: --suffix=ing generates words like "processing", "marketing"

Blend (--blend)

• Blend two words into portmanteau candidates
• Uses Markov chain probabilities to find natural splice points
• Example: --blend=smoke,fog generates words like "smog", "sfog", "smofog"
• Returns multiple candidates ranked by naturalness

Manual Markov Mode (--markov)

• Explicitly enable Markov chains with default settings
• Not needed if other Markov options are used

Python API

Syllable Generator

from syllable_generator import SyllableWordGenerator

generator = SyllableWordGenerator()
word = generator.generate(min_len=6, max_len=12)
words = generator.generate_batch(10, min_len=4, max_len=8)

Markov Generator

from markov_generator import MarkovWordGenerator

generator = MarkovWordGenerator(order=2, cutoff=0.1, words="en")
word = generator.generate(min_len=6, max_len=12)
words = generator.generate_batch(10)

Blend Generator

from markov_generator import MarkovWordGenerator
from blend_generator import BlendGenerator

markov = MarkovWordGenerator(order=2, words="en")
blender = BlendGenerator(markov)
blends = blender.blend("smoke", "fog", count=5)
# e.g. ['smog', 'smokfog', 'sfog', ...]

Performance Notes

• First run downloads word lists and builds chains (slower)
• Subsequent runs use cached files (much faster)
• Use --verbose to see initialization progress

Vibe Coded 🤖

Aider with Claude and ChatGPT

License

MIT