Update app.py

app.py

@@ -5,6 +5,13 @@ import time
 
 count = 0
 
+"""
+For Fairseq models, use the following name format: tts_models/<lang-iso_code>/fairseq/vits. You can find the language ISO codes here and learn about the Fairseq models here.
+https://dl.fbaipublicfiles.com/mms/tts/all-tts-languages.html
+"""
+api_rus = TTS(f"tts_models/rus/fairseq/vits", gpu=False)
+api_eng = TTS(f"tts_models/eng/fairseq/vits", gpu=False)
+api_spa = TTS(f"tts_models/spa/fairseq/vits", gpu=False)
 
 def audio_tts(txt, language, audio_file):
     global count
@@ -15,13 +22,17 @@ def audio_tts(txt, language, audio_file):
         count = 0
 
     # TTS with on the fly voice conversion
-    api = TTS(f"tts_models/{language}/fairseq/vits", gpu=False)
-    api.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
+    if language == "rus":
+        api_rus.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
+    elif language == "eng":
+        api_eng.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
+    elif language == "spa":
+        api_spa.tts_with_vc_to_file(txt, speaker_wav=audio_file, file_path="ouptut.wav")
     return "ouptut.wav"
     
 
 demo = gr.Interface(fn=audio_tts, inputs=[gr.Textbox(label="Input text TTS", value="Привет! Я Макс."),
-                                          gr.Textbox(label="Language", value="rus"),
+                                          gr.Dropdown(choices=["rus", "eng", "spa"],label="Language", value="rus"),
                                           gr.Audio(source="upload", type="filepath", label="Input audio")],
                     outputs=gr.Audio(source="upload", type="filepath", label="Output audio"))