Update app.py

app.py

@@ -17,7 +17,7 @@ MODEL_NAMES = [
 ]
 
 MAX_TXT_LEN = 800
-MODELS_DIRECTORY = os.path.join(os.path.dirname(__file__), "models")
+MODELS_DIRECTORY = "models"
 
 modelInfo = [
     ["vits-male", "best_model_65633.pth", "config-0.json",
@@ -36,54 +36,32 @@ modelInfo = [
      "https://huggingface.co/Kamtera/persian-tts-female-tacotron2/resolve/main/"]
 ]
 
-for model in modelInfo:
-    model_name, model_filename, config_filename, model_url = model
-    model_directory = os.path.join(MODELS_DIRECTORY, model_name)
-    if not os.path.exists(model_directory):
-        os.makedirs(model_directory)
-    print("|> Downloading: ", model_directory)
-    download_url(model_url + model_filename, model_directory, "best_model.pth")
-    download_url(model_url + config_filename, model_directory, "config.json")
+class PersianTTS:
+    def __init__(self):
+        self.model_manager = ModelManager(MODELS_DIRECTORY)
 
+        for model in modelInfo:
+            model_name, model_filename, config_filename, model_url = model
+            self.model_manager.download_model(model_name, model_filename, config_filename, model_url)
 
-def tts(text: str, model_name: str):
-    if len(text) > MAX_TXT_LEN:
-        text = text[:MAX_TXT_LEN]
-        print(f"Input text was cutoff since it went over the {MAX_TXT_LEN} character limit.")
+    def tts(self, text: str, model_name: str):
+        if len(text) > MAX_TXT_LEN:
+            text = text[:MAX_TXT_LEN]
+            print(f"Input text was cutoff since it went over the {MAX_TXT_LEN} character limit.")
 
-    # synthesize
-    model_directory = os.path.join(MODELS_DIRECTORY, model_name)
-    model_path = os.path.join(model_directory, "best_model.pth")
-    config_path = os.path.join(model_directory, "config.json")
-    synthesizer = Synthesizer(model_path, config_path)
-    if synthesizer is None:
-        raise NameError("model not found")
-    wavs = synthesizer.tts(text)
-    # return output
-    with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
-        synthesizer.save_wav(wavs, fp)
-        return fp.name
+        # synthesize
+        model_path, config_path = self.model_manager.get_model_paths(model_name)
+        synthesizer = Synthesizer(model_path, config_path)
+        if synthesizer is None:
+            raise NameError("model not found")
+        wavs = synthesizer.tts(text)
+        # return output
+        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as fp:
+            synthesizer.save_wav(wavs, fp)
+            return fp.name
 
 
-description = """
-This is a demo of persian text to speech model.
 
-**Github : https://github.com/karim23657/Persian-tts-coqui  **
-
-Models can be found here:  <br>
-
-|Model|Dataset|
-|----|------|
-|[vits female (best)](https://huggingface.co/Kamtera/persian-tts-female-vits)|[persian-tts-dataset-famale](https://www.kaggle.com/datasets/magnoliasis/persian-tts-dataset-famale)|
-|[vits male1 (best)](https://huggingface.co/Kamtera/persian-tts-male1-vits)|[persian-tts-dataset-male](https://www.kaggle.com/datasets/magnoliasis/persian-tts-dataset-male)|
-|[vits female1](https://huggingface.co/Kamtera/persian-tts-female1-vits)|[ParsiGoo](https://github.com/karim23657/ParsiGoo)|
-|[vits male](https://huggingface.co/Kamtera/persian-tts-male-vits)|[persian-tts-dataset](https://www.kaggle.com/datasets/magnoliasis/persian-tts-dataset)|
-|[glowtts female](https://huggingface.co/Kamtera/persian-tts-female-glow_tts)|[persian-tts-dataset-famale](https://www.kaggle.com/datasets/magnoliasis/persian-tts-dataset-famale)|
-|[glowtts male](https://huggingface.co/Kamtera/persian-tts-male-glow_tts)|[persian-tts-dataset](https://www.kaggle.com/datasets/magnoliasis/persian-tts-dataset)|
-|[tacotron2 female](https://huggingface.co/Kamtera/persian-tts-female-tacotron2)|[persian-tts-dataset-famale](https://www.kaggle.com/datasets/magnoliasis/persian-tts-dataset-famale)|
-
-
-"""
 article = ""
 examples = [
     ["و خداوند شما را با ارسال روح در جسم زندگانی و حیات بخشید", "vits-male"],
@@ -95,8 +73,10 @@ examples = [
     ["در گلو ماند خس او سالها، چیست آن خس مهر جاه و مالها", "vits male1 (best)"],
 ]
 
+persian_tts = PersianTTS()
+
 iface = gr.Interface(
-    fn=tts,
+    fn=persian_tts.tts,
     inputs=[
         gr.Textbox(
             label="Text",