Update tools.py

tools.py

@@ -13,7 +13,7 @@ from langchain_community.utilities import WikipediaAPIWrapper
 # import bitsandbytes as bnb
 # import torch.nn as nn
 # import torch
-import pyttsx3
+# import pyttsx3
 # from agents import get_agents_and_tasks
 # from langchain_google_genai import ChatGoogleGenerativeAI
 
@@ -61,31 +61,66 @@ import pyttsx3
 # pipe.enable_model_cpu_offload()
 
 
-def generate_speech(text, speech_dir='./outputs/audio', lang='en', speed=170, voice='default', num=0):
+# def generate_speech(text, speech_dir='./outputs/audio', lang='en', speed=170, voice='default', num=0):
+#     """
+#     Generates speech for given script.
+#     """
+#     engine = pyttsx3.init()
+    
+#     # Set language and voice
+#     voices = engine.getProperty('voices')
+#     if voice == 'default':
+#         voice_id = voices[1].id
+#     else:
+#         # Try to find the voice with the given name
+#         voice_id = None
+#         for v in voices:
+#             if voice in v.name:
+#                 voice_id = v.id
+#                 break
+#         if not voice_id:
+#             raise ValueError(f"Voice '{voice}' not found.")
+    
+#     engine.setProperty('voice', voice_id)
+#     engine.setProperty('rate', speed)
+#     # os.remove(os.path.join(os.path.dirname(os.path.abspath(__file__)), speech_dir, f'speech_{num}.mp3')) if os.path.exists(os.path.join(speech_dir, f'speech_{num}.mp3')) else None
+#     engine.save_to_file(text, os.path.join(os.path.dirname(os.path.abspath(__file__)), speech_dir, f'speech_{num}.mp3'))
+#     engine.runAndWait()
+
+def generate_speech(text, speech_dir='./outputs/speeches', lang='en', speed=1.0, num=0):
     """
-    Generates speech for given script.
+    Generates speech for the given script using gTTS and adjusts the speed.
     """
-    engine = pyttsx3.init()
+    # Ensure the speech directory exists
+    if not os.path.exists(speech_dir):
+        os.makedirs(speech_dir)
+    
+    # Generate speech
+    tts = gTTS(text=text, lang=lang)
     
-    # Set language and voice
-    voices = engine.getProperty('voices')
-    if voice == 'default':
-        voice_id = voices[1].id
+    # Save the speech to an MP3 file
+    speech_path = os.path.join(speech_dir, f'speech_{num}.mp3')
+    temp_path = os.path.join(speech_dir, f'temp_speech_{num}.mp3')
+    if os.path.exists(speech_path):
+        os.remove(speech_path)  # Remove existing file if it exists
+    
+    tts.save(temp_path)
+
+    # Adjust the speed of the speech
+    sound = AudioSegment.from_file(temp_path)
+    if speed != 1.0:
+        sound_with_altered_speed = sound._spawn(sound.raw_data, overrides={
+            "frame_rate": int(sound.frame_rate * speed)
+        }).set_frame_rate(sound.frame_rate)
+        sound_with_altered_speed.export(speech_path, format="mp3")
     else:
-        # Try to find the voice with the given name
-        voice_id = None
-        for v in voices:
-            if voice in v.name:
-                voice_id = v.id
-                break
-        if not voice_id:
-            raise ValueError(f"Voice '{voice}' not found.")
+        sound.export(speech_path, format="mp3")
     
-    engine.setProperty('voice', voice_id)
-    engine.setProperty('rate', speed)
-    # os.remove(os.path.join(os.path.dirname(os.path.abspath(__file__)), speech_dir, f'speech_{num}.mp3')) if os.path.exists(os.path.join(speech_dir, f'speech_{num}.mp3')) else None
-    engine.save_to_file(text, os.path.join(os.path.dirname(os.path.abspath(__file__)), speech_dir, f'speech_{num}.mp3'))
-    engine.runAndWait()
+    os.remove(temp_path)  # Remove the temporary file
+    # print(f"Speech saved to {speech_path}")
+
+# Example usage
+# generate_speech("Hello, this is a test speech.", speed=1.2, num=1)
 
 # class VideoGeneration(BaseModel):
 #     images_dir : str = Field(description='Path to images directory, such as "outputs/images"')