Update src/app.py

src/app.py

@@ -1,21 +1,15 @@
-import os
-os.environ["STREAMLIT_CONFIG_DIR"] = "/tmp/.streamlit"
-os.makedirs("/tmp/.streamlit", exist_ok=True)
-
-import tempfile
-from gtts import gTTS
+import streamlit as st
 from docx import Document
 from PyPDF2 import PdfReader
-import streamlit as st
-import base64
 from io import BytesIO
+from TTS.api import TTS
+from scipy.io.wavfile import write
+import numpy as np
 
-def text_to_speech(text):
-    tts = gTTS(text)
-    audio_buffer = BytesIO()
-    tts.write_to_fp(audio_buffer)
-    audio_buffer.seek(0)
-    return audio_buffer
+# Load offline TTS model once
+@st.cache_resource
+def load_tts_model():
+    return TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC", progress_bar=False, gpu=False)
 
 def convert_docx_to_text(docx_file):
     doc = Document(docx_file)
@@ -25,19 +19,26 @@ def convert_pdf_to_text(pdf_file):
     reader = PdfReader(pdf_file)
     return "\n".join([page.extract_text() or '' for page in reader.pages])
 
-def get_download_link(audio_data, filename="output.mp3"):
-    b64 = base64.b64encode(audio_data).decode()
-    href = f'<a href="data:audio/mp3;base64,{b64}" download="{filename}">Download {filename}</a>'
+def text_to_speech(text):
+    tts = load_tts_model()
+    wav = tts.tts(text)
+    buffer = BytesIO()
+    write(buffer, 22050, np.array(wav))  # Save as WAV
+    buffer.seek(0)
+    return buffer
+
+def get_download_link(audio_data, filename="output.wav"):
+    b64 = st.base64.b64encode(audio_data.getvalue()).decode()
+    href = f'<a href="data:audio/wav;base64,{b64}" download="{filename}">Download {filename}</a>'
     return href
 
 def main():
-    st.title("Text to Speech Converter (No File Write)")
+    st.title("Text to Speech Converter (Offline - Hugging Face Safe)")
 
     uploaded_file = st.file_uploader("Upload a text, docx, or pdf file", type=["txt", "docx", "pdf"])
 
     if uploaded_file:
         ext = uploaded_file.name.split('.')[-1].lower()
-        text = ""
 
         if ext == 'txt':
             text = uploaded_file.read().decode("utf-8")
@@ -53,12 +54,11 @@ def main():
             st.warning("No readable text found.")
             return
 
-        with st.spinner("Converting to speech..."):
+        with st.spinner("Generating speech..."):
             audio_buffer = text_to_speech(text)
-            audio_bytes = audio_buffer.read()
 
-        st.audio(audio_bytes, format="audio/mp3")
-        st.markdown(get_download_link(audio_bytes), unsafe_allow_html=True)
+        st.audio(audio_buffer, format="audio/wav")
+        st.markdown(get_download_link(audio_buffer), unsafe_allow_html=True)
 
 if __name__ == "__main__":
     main()