Update utils.py

utils.py

@@ -5,14 +5,19 @@ import os
 import tiktoken
 import json
 import re
-from gtts import gTTS
 import tempfile
 import requests
 from bs4 import BeautifulSoup
+from TTS.api import TTS
+import torch
 
 groq_client = Groq(api_key=os.environ["GROQ_API_KEY"])
 tokenizer = tiktoken.get_encoding("cl100k_base")
 
+# Initialize TTS models
+tts_maria = TTS("tts_models/en/ljspeech/tacotron2-DDC")
+tts_sarah = TTS("tts_models/en/ljspeech/glow-tts")
+
 class DialogueItem(BaseModel):
     speaker: Literal["Maria", "Sarah"]
     text: str
@@ -95,11 +100,9 @@ def generate_script(system_prompt: str, input_text: str, tone: str, target_lengt
     return dialogue
 
 def generate_audio(text: str, speaker: str) -> str:
-    if speaker == "Maria":
-        tts = gTTS(text=text, lang='en', tld='com')  # US English for Maria
-    else:  # Sarah
-        tts = gTTS(text=text, lang='en', tld='co.uk')  # UK English for Sarah
-    
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as temp_audio:
-        tts.save(temp_audio.name)
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_audio:
+        if speaker == "Maria":
+            tts_maria.tts_to_file(text=text, file_path=temp_audio.name)
+        else:  # Sarah
+            tts_sarah.tts_to_file(text=text, file_path=temp_audio.name)
         return temp_audio.name