Update xtts.py

xtts.py

@@ -98,7 +98,7 @@ def playInHTML(url):
 
 def get_conditioning_latents(audio_path, **others):
     global model
-    speaker_wav, pt_file=download(audio_path)
+    speaker_wav, pt_file=download_sample(audio_path)
     try:
         if pt_file != None:
             (
@@ -115,7 +115,7 @@ def get_conditioning_latents(audio_path, **others):
         logging.debug(f'sample wav info saved to {pt_file}')
     return gpt_cond_latent,speaker_embedding
 
-def download(url):
+def download_sample(url):
     try:
         response = requests.get(url)
         if response.status_code == 200:
@@ -128,6 +128,13 @@ def download(url):
     except:
         return default_sample
 
+def download(url):
+    response = requests.get(url)
+    if response.status_code == 200:
+        with tempfile.NamedTemporaryFile(mode="wb", suffix=".wav", delete=True) as temp_file:
+            temp_file.write(response.content)
+            return os.path.abspath(temp_file.name)
+
 def trim_sample_audio(speaker_wav):
     global ffmpeg
     try:
@@ -171,9 +178,13 @@ def tts_play():
 
 @app.route("/clone")
 def clone():
+    get_tts()
     source=request.args.get('source')
     sample=request.args.get('sample')
-    get_tts()
+    if not source:
+        return "No source", 400
+
+    source=download(source)
     output=tempfile.mktemp(suffix=".wav", delete=False)[1]
     tts.voice_conversion_to_file(
         source_wav=source,