text stringlengths 64 117 | label stringclasses 10
values |
|---|---|
Melancholic Chinese ballad with soft acoustic guitar and emotional male vocals about resembling someone from the past | c-pop |
Smooth Chinese R&B duet featuring gentle harmonies and minimalist production about unconditional feelings | c-pop |
Dark Chinese trap-influenced track with heavy bass and auto-tuned vocals exploring emotional addiction | c-pop |
Catchy pop track with playful lyrics about dating drama featuring upbeat production and confident female vocals | pop |
Sultry electronic pop with breathy vocals and dark synths about intense desire and dangerous attraction | pop |
Smooth R&B with layered harmonies and trap-influenced beats about relationship complications | r&b |
Powerful anthem with fierce vocals celebrating female empowerment over hip-hop influenced production | r&b |
Experimental bilingual track mixing Chinese and English with edgy electronic production and attitude | c-pop |
Energetic cover with powerful vocals and rock-influenced arrangement building to explosive chorus | pop |
Synth-pop perfection with dreamy production and lyrics about a complicated summer romance | pop |
Ethereal pop with nostalgic production and wistful vocals about memories of past love | pop |
Emotional ballad with soaring vocals and orchestral production about desperate love | pop |
Vulnerable pop ballad with piano and strings about giving love one final chance | pop |
Hyperpop anthem with glitchy production and autotuned vocals celebrating hedonistic partying | pop |
Playful pop with witty wordplay and bouncy production about crushing on someone | pop |
Electronic pop with dramatic production about awakening from emotional manipulation | pop |
High-energy K-pop with video game inspired sounds and confident group vocals | k-pop |
Solo K-pop debut with trap beats and fierce attitude showcasing individual star power | k-pop |
Smooth R&B-pop with sultry vocals about possessive romance over minimalist beats | pop |
Dreamy pop with West Coast vibes and catchy hooks about young love in California | pop |
Fierce collaboration with heavy bass and confident dual vocals about irresistible chemistry | k-pop |
Emotional pop with raw vocals processing heartbreak over stripped-down production | pop |
K-pop solo track celebrating Seoul nightlife with electronic beats and confident delivery | k-pop |
Disco-pop with retro production and empowering lyrics about knowing your worth | pop |
Latin-pop collaboration with reggaeton influences and fierce female empowerment message | latin |
High-energy K-pop with aggressive choreography beats and powerful group chants | k-pop |
Dark R&B with moody production featuring male vocals about toxic attraction | r&b |
80s-influenced synth ballad with emotional vocals about lost love and regret | pop |
Haunting K-pop solo with orchestral production exploring toxic relationships | k-pop |
Flirty pop with cheeky lyrics and retro production about unexpected attraction | pop |
Dreamy pop with Y2K influences and breathy vocals creating nostalgic summer vibes | pop |
Ethereal indie-pop with reverb-heavy production and melancholic vocals about transcendent love | indie |
Atmospheric vintage-pop with cinematic strings and dramatic vocals about fame | pop |
Dreamy indie track with ambient guitars and intimate vocals creating late-night mood | indie |
Dark pop with haunting production and vulnerable vocals about mental health | pop |
Atmospheric British indie-rock with moody guitars and emotional male vocals about crime metaphors | indie |
Upbeat indie-rock with catchy guitars and romantic lyrics about finding perfect match | indie |
Melancholic indie-rock with ambient production exploring complicated relationships | indie |
Smooth R&B with intimate vocals and minimalist production about self-love journey | r&b |
Alternative R&B with trap influences and layered vocals about complicated love | r&b |
Rock anthem with Italian flair mixing aggressive guitars with seductive vocals | rock |
Emotional pop ballad with orchestral arrangement dedicated to someone special | pop |
Feel-good pop with throwback production celebrating natural beauty and confidence | pop |
Uplifting pop with sincere vocals and warm production about unconditional acceptance | pop |
Hard-hitting hip-hop with aggressive flow and heavy bass addressing competition | hip-hop |
Piano ballad with powerful vocals and emotional delivery about lost love | pop |
Inspirational pop with acoustic guitar about overcoming life's challenges | pop |
Soft rock ballad with gentle vocals and romantic lyrics about devotion | rock |
Emotional power ballad with soaring female vocals about overwhelming love | pop |
Orchestral pop ballad with powerful vocals and sweeping production about complete devotion | pop |
Viral indie-pop with unique male vocals and intimate production about double takes | indie |
Bilingual indie-pop with dreamy production and soft vocals creating intimate atmosphere | indie |
Mid-2000s R&B with signature production and smooth vocals about attraction | r&b |
Sultry electronic pop with breathy vocals and dark production about physical desire | pop |
Hyperpop anthem with distorted production and party-ready energy | pop |
Club-ready pop with pulsing beats and confident vocals celebrating success | pop |
Dark R&B with vengeful lyrics and cinematic production inspired by martial arts films | r&b |
Smooth jazz-influenced track with saxophone and groovy bassline creating chill vibes | r&b |
Classic R&B with smooth vocals and timeless production about love's impact | r&b |
Emotional R&B ballad with vulnerable vocals about heartbreak and moving on | r&b |
Korean rock ballad with emotional guitars and powerful vocals about waiting in the rain | k-rock |
Melancholic Korean rock with atmospheric production and emotional delivery about apologies | k-rock |
Classic jazz piano piece with elegant melodies evoking winter scenes in New York | jazz |
Smooth jazz standard with sophisticated arrangement and relaxed tempo | jazz |
Vintage jazz vocal with orchestral backing capturing romantic New England autumn | jazz |
Intimate jazz ballad with soft trumpet and vulnerable male vocals about falling too quickly | jazz |
Classic jazz standard with melancholic trumpet and gentle swing rhythm about seasonal change | jazz |
Cheerful pop with harmonious vocals and optimistic lyrics about feeling on top | pop |
Emotional K-pop ballad with delicate vocals and piano-driven melody about final farewells | k-pop |
Confident K-pop track with powerful choreography beats and self-assured lyrics about being the best | k-pop |
Fierce R&B-pop with attitude-filled vocals and bass-heavy production about confidence | r&b |
Nostalgic hyperpop with glitchy production and autotuned vocals about chasing euphoric moments | pop |
Sultry alternative R&B with smooth vocals and minimalist beats exploring physical attraction | r&b |
Dreamy R&B with ethereal production and layered vocals about the number's significance | r&b |
Upbeat K-pop collaboration with playful vocals and party-ready production asking where you at | k-pop |
Energetic pop-rock with catchy hooks and confident vocals about non-stop conversation | pop |
Bold hip-hop track with fierce female rap and trap beats about dual personalities | hip-hop |
Clever wordplay in alt-hip-hop style with catchy production about avoiding reality | hip-hop |
Moody indie rock with distinctive British vocals about late night drunk calls | indie |
Japanese rock anthem with powerful vocals and driving guitars about rebellion | rock |
Explosive dance-pop with infectious beat and party lyrics about lighting up the night | pop |
Latin-pop fusion with Colombian rhythms and bilingual vocals about truthful body language | latin |
Confident K-pop with Latin influences and fierce choreography celebrating identity | k-pop |
Caribbean-influenced pop with commanding vocals and dancehall beats about bad boys | pop |
Arena rock anthem with thunderous drums and inspirational lyrics about transformation | rock |
Emotional K-rock ballad with heartfelt vocals and guitar-driven melody about past love | k-rock |
Progressive house with country vocals building to euphoric drop about awakening | pop |
K-pop track mixing English and Korean with smooth production about chasing success | k-pop |
Powerful ballad with soaring female vocals and orchestral arrangement about nature | pop |
Uplifting pop-rock with youthful energy and positive message about living fully | pop |
Neo-soul masterpiece with conscious rap and live instrumentation about societal issues | r&b |
Classic 90s R&B with soulful vocals and hip-hop beats defining an era | r&b |
Energetic K-pop with catchy hook and synchronized choreography about attraction | k-pop |
Dreamy K-pop with minimalist R&B production and soft vocals creating chill vibes | k-pop |
Fierce hip-hop anthem with confident female rap and trap production about success | hip-hop |
West Coast hip-hop with smooth flow and G-funk production about lifestyle | hip-hop |
Conscious hip-hop with introspective lyrics and jazz-influenced beats about growth | hip-hop |
Latin reggaeton with infectious beat and Spanish vocals about dancing all night | latin |
Modern Latin trap with autotuned vocals and heavy bass about heartbreak | latin |
Moody indie rock with jangly guitars and introspective lyrics about youth | indie |
Song Genre Classification Text Dataset
Dataset Summary
Purpose: This dataset was created for multi-class text classification of song descriptions into music genres, developed as part of CMU 24-679 coursework to explore text augmentation techniques in NLP.
Quick Stats:
- 1,122 total samples (102 original + 1,020 augmented)
- 10 genre categories
- ~200 character descriptions per sample
- Balanced augmentation (10x per original)
Contact: maryzhang@cmu.edu
Dataset Composition
Features
text: String (song description, ~200 characters)label: String (genre category)
Genre Distribution
| Genre | Original | Augmented | Percentage |
|---|---|---|---|
| pop | 35 | 350 | 34.3% |
| k-pop | 15 | 150 | 14.7% |
| r&b | 15 | 150 | 14.7% |
| indie | 10 | 100 | 9.8% |
| rock | 8 | 80 | 7.8% |
| hip-hop | 7 | 70 | 6.9% |
| jazz | 5 | 50 | 4.9% |
| c-pop | 4 | 40 | 3.9% |
| latin | 4 | 40 | 3.9% |
| k-rock | 3 | 30 | 2.9% |
Data Splits
- original: 102 manually written song descriptions
- augmented: 1,020 synthetically augmented descriptions (10x augmentation per original)
Data Collection Process
Collection Methodology
Song descriptions created between January-February 2025:
- Based on 100+ popular songs from various streaming platforms
- Covers music from 1980s-2020s
- International representation (English, Korean, Chinese, Spanish)
- Manual description writing without using copyrighted lyrics
Selection Criteria
- Popular/recognizable songs across genres
- Diverse cultural and temporal representation
- Clear genre categorization
- No explicit content or offensive material
Preprocessing and Augmentation
Preprocessing Pipeline
- Standardized to ~200 character descriptions
- Removed artist names and song titles from text
- Ensured grammatical consistency
- Verified genre labels
Augmentation Techniques
Each original generated 10 augmented variants using:
- EDA: Synonym replacement, random deletion/swap/insertion (3 variants)
- Character Noise: Random drops and swaps (3 variants)
- Back-Translation: English → German → English pipeline (1 variant)
- T5 Paraphrasing: Neural text generation with T5-small (1 variant)
- Combined: EDA + character noise (2 variants)
Labels and Annotation
Labeling Schema
- pop: Mainstream pop music
- k-pop: Korean pop music
- r&b: Rhythm and blues
- indie: Independent/alternative
- rock: Rock and soft rock
- hip-hop: Rap and hip-hop
- jazz: Jazz standards
- c-pop: Chinese pop music
- latin: Latin/Spanish music
- k-rock: Korean rock
Annotation Process
- Manual labeling based on known genre classifications
- Single-label assignment (no multi-genre)
- Verified against music platform categorizations
Intended Use and Limitations
Intended Use Cases
- Text classification model training
- Studying text augmentation effectiveness
- NLP educational projects
- Genre classification research baseline
Limitations
- Limited to ~200 character descriptions
- Subjective genre boundaries
- Western music bias despite international songs
- Augmented samples may contain grammatical errors
- Small original dataset size (102 samples)
Out-of-Scope Uses
- Production music recommendation systems
- Detailed music analysis (lyrics, audio)
- Multi-label genre classification
- Commercial applications without additional data
Ethical Considerations
Representation
- Attempted balance across major global music markets
- Acknowledges genre classification subjectivity
- May not equally represent all music communities
Privacy
- No personal information included
- No copyrighted lyrics reproduced
- Descriptions are original creations
Cultural Sensitivity
- Genre labels may oversimplify cultural music traditions
- "World music" avoided as a category due to its problematic nature
- Users should be aware of Western-centric genre definitions
AI Usage Disclosure
AI-Assisted Components
- Augmentation: Back-translation using Helsinki-NLP models
- Paraphrasing: T5-small model for text generation
- Documentation: README structure refined with AI assistance
- Descriptions: Written with human knowledge, no lyrics copied
Human Oversight
- All original descriptions manually written
- Genre labels manually assigned and verified
- Augmentation quality manually reviewed
- No copyrighted material reproduced
Usage Example
from datasets import load_dataset
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
# Load dataset
dataset = load_dataset("maryzhang/hw1-24679-text-dataset-augmented")
# Prepare data
X_train = dataset['augmented']['text']
y_train = dataset['augmented']['label']
X_test = dataset['original']['text']
y_test = dataset['original']['label']
# Train classifier
vectorizer = TfidfVectorizer(max_features=1000)
X_train_vec = vectorizer.fit_transform(X_train)
X_test_vec = vectorizer.transform(X_test)
clf = LogisticRegression()
clf.fit(X_train_vec, y_train)
accuracy = clf.score(X_test_vec, y_test)
print(f"Accuracy: {accuracy:.2f}")
Citation
bibtex@dataset{zhang2025songgenre, author = {Mary Zhang}, title = {Song Genre Classification Text Dataset}, year = {2025}, publisher = {Hugging Face}, note = {CMU 24-679 Homework 1}, url = {https://huggingface.co/datasets/maryzhang/hw1-24679-text-dataset-augmented} }
License
This dataset is released under the MIT License.
Contact
Dataset created by Mary Zhang for CMU 24-679. For questions or issues, please contact maryzhang@cmu.edu.
- Downloads last month
- 32