| # Demo clips | |
| Five ~6–8 s speech clips (16 kHz mono), one per language, for a quick check of the model (see the model card). | |
| Source: **FLEURS** (`google/fleurs`, validation split), Conneau et al. 2022 — licensed **CC-BY-4.0**. | |
| Files: `hi_in` Hindi · `ta_in` Tamil · `bn_in` Bengali · `te_in` Telugu · `mr_in` Marathi. | |