Update app.py

app.py

@@ -3,6 +3,7 @@ import streamlit as st
 from dotenv import load_dotenv
 import base64
 import requests
+import time
 
 # Load environment variables from the .env file
 load_dotenv()
@@ -46,13 +47,26 @@ def translate_text(text, target_language):
             return text
     return text  # If English, return the same text
 
-# Initialize Streamlit UI
-st.set_page_config(page_title="Multilingual Text-to-Speech", page_icon="🔊")
-st.header("Multilingual Text-to-Speech Demo")
+# Function to send a request to the Hugging Face API with retry on 503
+def generate_speech(model_id, text):
+    url = f"https://api-inference.huggingface.co/models/{model_id}"
+    headers = {"Authorization": f"Bearer {HUGGINGFACE_API_TOKEN}"}
+    payload = {"inputs": text}
 
-# Get user input and language selection
-user_input = get_text()
-selected_language = select_language()
+    while True:
+        response = requests.post(url, headers=headers, json=payload, stream=True)
+
+        if response.status_code == 200:
+            return response.content  # Return the audio data if successful
+        elif response.status_code == 503:
+            # Parse the estimated wait time from the response and wait
+            response_data = response.json()
+            estimated_time = response_data.get("estimated_time", 10)
+            st.info(f"Model is loading, please wait {estimated_time} seconds...")
+            time.sleep(estimated_time)  # Wait before retrying
+        else:
+            st.error(f"Error: {response.status_code} - {response.text}")
+            return None
 
 # Function to play audio in Streamlit
 def play_audio(audio_bytes):
@@ -65,6 +79,14 @@ def play_audio(audio_bytes):
     """
     st.markdown(audio_html, unsafe_allow_html=True)
 
+# Initialize Streamlit UI
+st.set_page_config(page_title="Multilingual Text-to-Speech", page_icon="🔊")
+st.header("Multilingual Text-to-Speech Demo")
+
+# Get user input and language selection
+user_input = get_text()
+selected_language = select_language()
+
 # Create a button for generating speech
 submit = st.button('Generate Speech')
 
@@ -79,19 +101,10 @@ if submit and user_input:
             st.write(f"Original Text: {user_input}")
             st.write(f"Translated Text: {translated_text}")
 
-        # Make a direct request to the API for speech synthesis
-        url = f"https://api-inference.huggingface.co/models/myshell-ai/MeloTTS-English"
-        headers = {"Authorization": f"Bearer {HUGGINGFACE_API_TOKEN}"}
-        payload = {"inputs": translated_text}
-
-        # Requesting raw response with the option to manually handle the audio
-        response = requests.post(url, headers=headers, json=payload, stream=True)
+        # Generate speech with retry if the model is loading
+        audio_data = generate_speech("myshell-ai/MeloTTS-English", translated_text)
 
-        if response.status_code == 200:
-            audio_data = response.content
-            # Assuming the response is in audio/wav or another format
+        if audio_data:
             play_audio(audio_data)
-        else:
-            st.error(f"Error: {response.status_code} - {response.text}")
 elif submit:
     st.warning("Please enter some text.")  # Warning for empty input