Smart Turn — Hinglish

Audio-only turn detection tuned for Indian Hinglish, Indian-accented English, and Indian filler words. Compared live against stock Smart Turn v3.2.

Complete = the speaker has finished and a voice bot may reply. Incomplete = they are mid-thought; replying now interrupts them.

Record or upload (16 kHz mono, ≤8 s used)

Ours — Hinglish fine-tuned

Stock — Smart Turn v3.2

Examples — the filler/tag pairs are where the two differ

Hesitation markers (matlab, toh, woh, arre, yaani) signal mid-thought → incomplete. Tag particles (na, haan, ji, theek hai) hand over the turn → complete. They sound similar and label oppositely, which is why a model that has never heard them gets both directions wrong.