Datasets:
id stringclasses 5
values | hinglish_text stringclasses 5
values | domain stringclasses 5
values | switch_pattern stringclasses 3
values | script stringclasses 1
value | status stringclasses 1
value | notes_for_review stringclasses 3
values |
|---|---|---|---|---|---|---|
HING-EXP-001 | Instagram pe sab founder hain, asli mein sab Jira mein phasay hain. | humor | domain_borrowing | roman | DRAFT-PREVIEW | Machine-expanded from canonical pattern 'domain_borrowing' (source label 'domain_borrowing', source record draft-patterns). Naturalness UNREVIEWED — approve/fix/reject like any draft. |
HING-EXP-003 | Birthday pe kya plan hai? Kahin bahar chalte hain na. | daily | intra_sentential_noun | roman | DRAFT-PREVIEW | Machine-expanded from canonical pattern 'intra_sentential_noun' (source label 'intra_sentential_noun', source record draft-patterns). Naturalness UNREVIEWED — approve/fix/reject like any draft. |
HING-EXP-004 | Yeh error fix nahi ho raha mere phone pe. | tech | domain_borrowing | roman | DRAFT-PREVIEW | Machine-expanded from canonical pattern 'domain_borrowing' (source label 'domain_borrowing', source record draft-patterns). Naturalness UNREVIEWED — approve/fix/reject like any draft. |
HING-EXP-006 | Ghabrahat ho rahi hai, pata nahi viva kaisa jayega. | emotional | hindi_dominant | roman | DRAFT-PREVIEW | Machine-expanded from canonical pattern 'hindi_dominant' (source label 'hindi_dominant', source record draft-patterns). Naturalness UNREVIEWED — approve/fix/reject like any draft. |
HING-EXP-014 | Ma'am, amount 48 hours mein aapke wallet mein show ho jayega. | customer-service | domain_borrowing | roman | DRAFT-PREVIEW | Machine-expanded from canonical pattern 'domain_borrowing' (source label 'domain_borrowing', source record draft-patterns). Naturalness UNREVIEWED — approve/fix/reject like any draft. |
Hinglish Code-Switching Expansion Engine — Preview
This is a free preview. The full commercial product (pattern taxonomy, Python expansion engine, domain vocabulary banks) is available on DataVendor:
Full dataset: [DataVendor listing — paste your listing URL here]
What this is
A complete Hinglish code-switching expansion framework: 8 canonical code-switching pattern families extracted from 63 native draft utterances, domain-matched vocabulary banks, and a Python generator engine (expand.py) that produces new utterances following real bilingual patterns.
Hinglish is spoken by 500M+ people but has almost zero structured training data. This isn't just a corpus — it's a generator that produces unlimited pattern-authentic Hinglish utterances across 5 domains: daily conversation, customer service, tech talk, emotional expression, and humor.
What's in this preview
preview.jsonl — 5 real machine-expanded preview records across 5 domains (one JSON object per line). Statuses are honestly marked DRAFT-PREVIEW — naturalness unreviewed.
Example records
| Domain | Pattern | Utterance |
|---|---|---|
| humor | domain_borrowing | "Instagram pe sab founder hain, asli mein sab Jira mein phasay hain." |
| daily | intra_sentential_noun | "Birthday pe kya plan hai? Kahin bahar chalte hain na." |
| tech | domain_borrowing | "Yeh error fix nahi ho raha mere phone pe." |
| emotional | hindi_dominant | "Ghabrahat ho rahi hai, pata nahi viva kaisa jayega." |
| customer-service | domain_borrowing | "Ma'am, amount 48 hours mein aapke wallet mein show ho jayega." |
Full product
| Contents | 8 pattern families (PATTERNS.md), Python expansion engine (expand.py), domain vocabulary banks, 20-record preview |
| Workflow | Native-speaker review → expand to 300+ utterances via engine |
| Price | $2,000 — commercial license |
| Where | [DataVendor listing — paste your listing URL here] |
Use cases: multilingual AI teams, code-switching researchers, Indian-market voice AI companies.
Honest status note
The expansion engine and pattern taxonomy are complete and functional. The 63 seed utterances are pending native-speaker review (scheduled); machine-expanded records are never sold as human-written — provenance is recorded per record.
License
This preview is free for evaluation. The full product is sold under a commercial license via DataVendor. Double Shot Film Studio retains ownership of the underlying creative IP.
- Downloads last month
- 37