gradio

F5-TTS/src/f5_tts/infer/infer_cli_test.py

@@ -257,7 +257,7 @@ sway_sampling_coef = args.sway_sampling_coef or config.get("sway_sampling_coef",
 speed = args.speed or config.get("speed", speed)
 fix_duration = args.fix_duration or config.get("fix_duration", fix_duration)
 
-print("############nfe_step", nfe_step)
+print("############nfe_step", nfe_step, vocoder_name)
 
 
 # patches for pip pkg user

F5-TTS/src/f5_tts/infer/utils_infer.py

@@ -263,7 +263,8 @@ def load_model(
         vocab_char_map=vocab_char_map,
     ).to(device)
 
-    dtype = torch.float32 if mel_spec_type == "bigvgan" else None
+    #dtype = torch.float32 if mel_spec_type == "bigvgan" else None
+    dtype = torch.float32
     model = load_checkpoint(model, ckpt_path, device, dtype=dtype, use_ema=use_ema)
 
     return model