update README

src/voxcpm/model/voxcpm.py

@@ -283,6 +283,7 @@ class VoxCPMModel(nn.Module):
                 else:
                     break
             else:
+                break   
                 
         decode_audio = self.audio_vae.decode(latent_pred.to(torch.float32)).squeeze(1).cpu()  
         decode_audio = decode_audio[..., 640:-640] # trick: trim the start and end of the audio