Add AI voice assistant and Armaan voice data; update TTS core and dependencies

- Added `ai_voice_assistant.py` for new assistant functionality
- Updated `requirements.txt` and `setup.py` for new dependencies
- Enhanced `tortoise/do_tts.py` with improvements
- Added new Armaan voice samples and model files in

ai_voice_assistant.py

@@ -0,0 +1,23 @@
+import subprocess
+from playsound import playsound
+import os
+
+def get_ai_response(prompt):
+    # Replace this later with LLaMA integration
+    return f"Echo: {prompt}"
+
+def generate_voice(text):
+    subprocess.run(f'python tortoise/do_tts.py --text "{text}" --voice random', shell=True)
+
+def play_response():
+    playsound("tortoise/results/generated_audio.wav")
+
+if __name__ == "__main__":
+    while True:
+        user_input = input("\nYou: ")
+        if user_input.lower() == "exit":
+            break
+        response = get_ai_response(user_input)
+        print("AI:", response)
+        generate_voice(response)
+        play_response()

requirements.txt

@@ -18,7 +18,7 @@ appdirs
 nbconvert==5.3.1
 tornado==4.2
 pydantic==1.9.1
-deepspeed==0.8.3
+#deepspeed==0.17.0
 py-cpuinfo
 hjson
 psutil

setup.py

@@ -28,7 +28,7 @@ setuptools.setup(
         'scipy',
         'librosa',
         'transformers==4.31.0',
-        'tokenizers==0.14.0',
+        'tokenizers==0.13.0',
         'scipy==1.13.1'
         # 'deepspeed==0.8.3',
     ],

tortoise/do_tts.py

@@ -4,8 +4,8 @@ import os
 import torch
 import torchaudio
 
-from api import TextToSpeech, MODELS_DIR
-from utils.audio import load_voices
+from tortoise.api import TextToSpeech, MODELS_DIR
+from tortoise.utils.audio import load_voice, load_voices
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()