Update README.md

README.md

@@ -60,7 +60,7 @@ To tackle flat emotional expression, GLM-TTS uses a **Group Relative Policy Opti
 
 Evaluated on `seed-tts-eval`. **GLM-TTS_RL** achieves the lowest Character Error Rate (CER) while maintaining high speaker similarity.
 
-| Model | CER $\downarrow$ | SIM $\uparrow$ | Open-source |
+| Model | CER ↓ | SIM ↑ | Open-source |
 | :--- | :---: | :---: | :---: |
 | Seed-TTS | 1.12 | **79.6** | 🔒 No |
 | CosyVoice2 | 1.38 | 75.7 | 👐 Yes |