Spaces:
Running on Zero
Running on Zero
🚨 Request: Proper Kurdish Language Support (Not Token Support)
#31
by yawner - opened
It’s frustrating to see that Kurdish continues to be overlooked or poorly supported across TTS and LLM systems, including Qwen3-TTS.
Kurdish is spoken by 30+ million people, yet current support is either:
- Non-existent
- Low quality / inaccurate
- Confused with Arabic, Persian, or Turkish phonetics and vocabulary
This is not just a technical gap — it’s a data and prioritization issue.
Key Problems:
- ❌ No clear support for major dialects (Kurmanji, Sorani)
- ❌ Incorrect pronunciation due to training overlap with neighboring languages
- ❌ Lack of proper datasets and linguistic handling
- ❌ No official roadmap or acknowledgment from model providers
Why This Matters:
Kurdish has historically been underrepresented and often misclassified online.
AI systems are now reinforcing that gap instead of correcting it.
What We Need:
- ✅ Explicit Kurdish language support (separate from Arabic/Persian/Turkish)
- ✅ Dialect-aware TTS (at minimum: Kurmanji & Sorani)
- ✅ Proper phoneme modeling and pronunciation
- ✅ Collaboration with native speakers for dataset creation
Opportunity (for Qwen / others):
This is a high-leverage, low-competition win.
Supporting Kurdish properly would:
- Unlock millions of new users
- Build strong community loyalty
- Position your model as globally inclusive — not just mainstream-language optimized
Right now, it feels like Kurdish is invisible in AI.
It shouldn’t be.
Let’s fix that.