🎤
Rap AI Model Catalog — Lyrics Generation, Rhyme Detection, and Flow Analysis
GPT-2 rap fine-tuned models + pronouncing rhyme analysis + all runnable code
1. Rap Lyrics Generation — input → model → output
Top pick: Elida-Sensoy/gpt2-rap-generator
GPT-2 fine-tuned on rap lyrics. Runs directly via pipeline().
from transformers import pipeline, set_seed
generator = pipeline("text-generation", model="Elida-Sensoy/gpt2-rap-generator")
set_seed(42)
results = generator("I walk the streets at night", max_length=200,
num_return_sequences=3, temperature=1.75, top_p=0.95,
do_sample=True, pad_token_id=50256)
2. Rhyme Detection
import pronouncing
print(pronouncing.rhymes("flow")) # → ['blow', 'go', 'know', 'show', ...]
print("hat" in pronouncing.rhymes("cat")) # → True
3. Phoneme & Stress Analysis
print(pronouncing.phones_for_word("rapping")) # → ['R AE1 P IH0 NG']
print(pronouncing.stresses_for_word("rapping")) # → ['10']
4. Generation + Rhyme Check Pipeline
Generate lyrics → extract last words → check if consecutive lines rhyme.
5. Syllable Count — Flow Measurement
import syllables
count = syllables.estimate("Money on my mind every single day") # → 10
6. Beat Analysis
import librosa
y, sr = librosa.load("beat.mp3")
tempo, beats = librosa.beat.beat_track(y=y, sr=sr)
print(f"BPM: {tempo:.0f}")
Model Catalog
| Model | Size | Feature |
|---|---|---|
| Elida-Sensoy/gpt2-rap-generator | ~500MB | Eminem-style rhyme density |
| flax-community/gpt2-rap-lyric-generator | ~500MB | 10K+ songs trained |
| weej16/rap-lyrics-qwen-0.6b | 600MB | Qwen3-based, latest |
| pronouncing (library) | 0MB | CMU dict rhyme detection |
Key Concepts
1
Lyrics generation — generate rap lyrics via pipeline("text-generation", model="gpt2-rap-generator")
2
Rhyme detection — pronouncing.rhymes("flow") for rhyme word list
3
Phoneme/stress analysis — pronouncing.phones_for_word() for pronunciation structure
4
Auto rhyme verification — auto-check if generated line-ending words rhyme
5
Beat analysis — extract BPM and beat positions with librosa.beat.beat_track()
Use Cases
Rap writing assistant — use AI-generated lyrics as a base to modify and develop
Rhyme quality measurement — auto-analyze rhyme density and patterns in lyrics
Rap education — visualize flow structure with syllable count and stress patterns