Refactor: Move translation and TTS code from notebook to separate scripts

my_translate.py

@@ -0,0 +1,26 @@
+import argostranslate.package
+import argostranslate.translate
+
+def translate_text(text, from_lang="en", to_lang="hi"):
+    """
+    Translate text using Argos Translate
+
+    Args:
+        text (str): Text to translate
+        from_lang (str): Source language code (default: "en")
+        to_lang (str): Target language code (default: "hi")
+
+    Returns:
+        str: Translated text
+    """
+    # Download language packs (e.g., English to Hindi)
+    argostranslate.package.update_package_index()
+    available_packages = argostranslate.package.get_available_packages()
+    package = next(filter(lambda x: x.from_code == from_lang and x.to_code == to_lang, available_packages))
+    argostranslate.package.install_from_path(package.download())
+    
+    translated_text = argostranslate.translate.translate(text, from_lang, to_lang)
+    # hindi_translation = argostranslate.translate.translate(text, "en", "hi")
+
+
+    return translated_text

my_tts.py

@@ -0,0 +1,38 @@
+from transformers import BarkModel, AutoProcessor
+import torch
+
+def text_to_speech(text, voice_preset="v2/hi_speaker_2"):
+    """
+    Convert text to speech using Bark model
+
+    Args:
+        text (str): Text to convert to speech
+        voice_preset (str): Voice preset to use for the speech synthesis
+
+    Returns:
+        torch.Tensor: Generated speech audio
+        sampling_rate (int): Sampling rate of the generated audio
+    """
+    # Check if CUDA is available and set device accordingly
+    device = "cuda:0" if torch.cuda.is_available() else "cpu"
+    
+    # Load the model and processor
+    model = BarkModel.from_pretrained("suno/bark-small")
+    processor = AutoProcessor.from_pretrained("suno/bark")
+
+    # Move model and inputs to the appropriate device
+    model = model.to(device)
+    inputs = processor(text=text, voice_preset=voice_preset)
+    for key, value in inputs.items():
+        inputs[key] = value.to(device)
+    
+    # prepare the inputs
+    inputs = processor(text, voice_preset=voice_preset)
+    for key, value in inputs.items():
+        inputs[key] = inputs[key].to(device)
+
+    # generate speech
+    speech_output = model.generate(**inputs)
+    sampling_rate = model.generation_config.sample_rate
+
+    return speech_output, sampling_rate