Update app.py

app.py

@@ -14,6 +14,19 @@ import re
 import time
 from streaming_stt_nemo import Model
 
+Female_language_dict = {
+ 'English-Jenny (Female)': 'en-US-JennyNeural',
+ 'English-Ana (Female)': 'en-US-AnaNeural',
+ 'English-Aria (Female)': 'en-US-AriaNeural',
+ 'English-Michelle (Female)': 'en-US-MichelleNeural',
+ 'English (Australia)-Natasha- (Female)': 'en-AU-NatashaNeural',
+ 'English (Canada)-Clara- (Female)': 'en-CA-ClaraNeural',
+ 'English (UK)-Libby- (Female)': 'en-GB-LibbyNeural',
+ 'English (UK)-Maisie- (Female)': 'en-GB-MaisieNeural',
+ 'English (UK)-Sonia- (Female)': 'en-GB-SoniaNeural',
+ 'English (Ireland)-Emily- (Female)': 'en-IE-EmilyNeural',
+}
+
 default_lang = "en"
 
 engines = { default_lang: Model(default_lang) }
@@ -54,9 +67,19 @@ def model(text):
 async def respond(audio):
     user = transcribe(audio)
     reply = model(user)
-    voices = await VoicesManager.create()
-    voice = voices.find(Gender="Female", Locale="en-GB")
-    communicate = edge_tts.Communicate(reply, random.choice(voice)["Name"])
+    
+    #        Random voice choise from Manager base on locale and gender
+    #
+    #voices = await VoicesManager.create()
+    #voice = voices.find(Gender="Female", Locale="en-GB")
+    #communicate = edge_tts.Communicate(reply, random.choice(voice)["Name"])
+    #
+    #        Or the following as one selected Voice for the character.
+    
+    language_code = 'English (UK)-Maisie- (Female)'
+     
+    voice = language_dict.get(language_code, "default_voice")
+    communicate = edge_tts.Communicate(reply, voice)
     
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
         tmp_path = tmp_file.name
@@ -87,10 +110,18 @@ async def generate1(prompt):
         if not response.token.text == "</s>":
             output += response.token.text
 
-    voices = await VoicesManager.create()
-    voice = voices.find(Gender="Female", Locale="en-GB")
-    
-    communicate = edge_tts.Communicate(output, random.choice(voice)["Name"])
+    #        Random voice choice from VoiceManager base on locale and gender
+    #
+    #voices = await VoicesManager.create()
+    #voice = voices.find(Gender="Female", Locale="en-GB")
+    #communicate = edge_tts.Communicate(reply, random.choice(voice)["Name"])
+    #
+    #        Or the following as one selected voice for the character.
+        
+    language_code = 'English (UK)-Maisie- (Female)'
+        
+    voice = language_dict.get(language_code, "default_voice")
+    communicate = edge_tts.Communicate(output, voice)
       
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
         tmp_path = tmp_file.name
@@ -103,6 +134,11 @@ with gr.Blocks(css="style.css") as demo:
                      ### <center>An Artificial Intelligence Assistant just for YOU:
                      ### <center>Now you can Ask for a spell to Hermione</center>
                      """)
+    #           Use for the Selection of voice for Hermione
+    #with gr.Row():
+    #    language_code = gr.Dropdown(choices=list(Female_language_dict.keys()), label="Select Voice for Hermione")
+    #
+    
     with gr.Row():
         input = gr.Audio(label="Your Voice Chat", type="filepath", interactive=True, sources="microphone", waveform_options=False)
         output = gr.Audio(label="Hermione", type="filepath", interactive=False, autoplay=True, elem_classes="audio")