Update app.py

app.py

@@ -27,7 +27,7 @@ print(f"Using {device} device")
 pipe = pipeline(
     "automatic-speech-recognition",
     model="openai/whisper-base",  # Smaller model for CPU
-    torch_dtype=torch.float32,
+    torch_dtype=torch.float16,
     device=device,
 )
 vocos = Vocos.from_pretrained("charactr/vocos-mel-24khz")
@@ -72,7 +72,7 @@ def load_model(page_name, repo_name, exp_name, model_cls, model_cfg, ckpt_step):
 
 # Reduced model configurations
 F5TTS_model_cfg = dict(
-    dim=1024, depth=12, heads=8, ff_mult=2, text_dim=512, conv_layers=4
+    dim=1024, depth=12, heads=16, ff_mult=2, text_dim=512, conv_layers=4
 )
 E2TTS_model_cfg = dict(dim=1024, depth=12, heads=8, ff_mult=2)