Update app.py

import os
import gradio as gr
from TTS.api import TTS
import torch

os.environ["COQUI_TOS_AGREED"] = "1"
device = "cuda" if torch.cuda.is_available() else "cpu"

print("Loading XTTS...")
try:
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
except:
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)

def clone(text, lang, ref):
out = "out.wav"
tts.tts_to_file(text=text, file_path=out, speaker_wav=ref, language=lang)
return out

LANGS = ["en","es","fr","de","it","pt","pl","tr","ru","nl","cs","ar","zh-cn","ja","ko","hu"]

with gr.Blocks() as demo:
t=gr.Textbox(label="Text")
l=gr.Dropdown(label="Lang", choices=LANGS, value="fr")
r=gr.Audio(label="Ref", type="filepath")
o=gr.Audio(label="Out")
gr.Button("Generate").click(clone, [t,l,r], [o])

demo.launch()

app.py

@@ -8,23 +8,22 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
 
 print("Loading XTTS...")
 try:
-        # Try direct ID loading which works on most versions
-        tts = TTS(model_name="tts_models/multilingual/multi-dataset/xtts_v2", progress_bar=False).to(device)
-except Exception as e:
-        print(f"Error loading by name: {e}")
-        # Fallback to general initialization which triggers download
-         tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
+    tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
+except:
+    tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)
 
 def clone(text, lang, ref):
-        out = "out.wav"
+    out = "out.wav"
     tts.tts_to_file(text=text, file_path=out, speaker_wav=ref, language=lang)
     return out
 
-LANGS = ["en","es","fr","de","it","pt","pl","tr","ru","nl","cs","ar","zh-cn","ja","ko","hu"
-        with gr.Blocks() as demo:
-            t=gr.Textbox(label="Text")
-                l=gr.Dropdown(label="Lang", choices=LANGS, value="fr")
-                    r=gr.Audio(label="Ref", type="filepath")
-                        o=gr.Audio(label="Out")
-                            gr.Button("Generate").click(clone, [t,l,r], [o])
-        demo.launch()]
+LANGS = ["en","es","fr","de","it","pt","pl","tr","ru","nl","cs","ar","zh-cn","ja","ko","hu"]
+
+with gr.Blocks() as demo:
+    t=gr.Textbox(label="Text")
+    l=gr.Dropdown(label="Lang", choices=LANGS, value="fr")
+    r=gr.Audio(label="Ref", type="filepath")
+    o=gr.Audio(label="Out")
+    gr.Button("Generate").click(clone, [t,l,r], [o])
+
+demo.launch()