Update src/streamlit_app.py

src/streamlit_app.py

@@ -21,13 +21,18 @@ from pydantic import BaseModel, Field
 from typing import Optional
 import os
 try:
-    from gtts import gTTS
-except Exception as e:
-    st.error(f"error importing gtts: {e}")
+    from google.cloud import speech
+except Exception:
+    speech = None
 try:
     from google.cloud import texttospeech
 except Exception:
     texttospeech = None
+try:
+    from streamlit_mic_recorder import mic_recorder
+except Exception:
+    mic_recorder = None
+
 # --- Pydantic Models (from your code) ---
 
 class questions(BaseModel):
@@ -174,6 +179,54 @@ If a good followup question can be asked generate it but only if it is a genuine
 # --- MODIFIED Streamlit Audio/Visual Function ---
 
 import io # Make sure 'import io' is at the top of your file
+@st.cache_data
+def speech_to_text(audio_bytes):
+    """
+    Transcribes audio bytes using Google Cloud Speech-to-Text
+    and returns the transcribed text.
+    """
+    if speech is None:
+        st.warning("google-cloud-speech library not found, transcription is disabled.")
+        return None
+        
+    # Get the API key from the environment (where HF secrets put it)
+    api_key = os.environ.get("GOOGLE_API_KEY") 
+    
+    # Check if the key exists
+    if not api_key:
+        st.error("GOOGLE_API_KEY not found in secrets. Cannot initialize STT.")
+        return None
+        
+    # Pass the key explicitly to the client
+    client_options = {"api_key": api_key}
+    client = speech.SpeechClient(client_options=client_options)
+
+    # Configure the audio
+    # Note: streamlit-mic-recorder outputs WAV, which is LINEAR16
+    audio = speech.RecognitionAudio(content=audio_bytes)
+    config = speech.RecognitionConfig(
+        encoding=speech.RecognitionConfig.AudioEncoding.LINEAR16,
+        language_code="en-US",
+        sample_rate_hertz=16000 # This is a common sample rate
+    )
+
+    try:
+        # Detects speech in the audio file
+        st.info("Transcribing audio... (this may take a moment)")
+        response = client.recognize(config=config, audio=audio)
+        
+        if response.results:
+            transcript = response.results[0].alternatives[0].transcript
+            st.session_state.chat_history.append(f"**You:** {transcript}")
+            return transcript
+        else:
+            st.warning("Could not understand audio.")
+            return None
+            
+    except Exception as e:
+        st.error(f"Error during speech-to-text: {e}")
+        st.info("This usually means the 'Cloud Speech-to-Text API' is not enabled or your mic is not outputting the correct audio format.")
+        return None
 
 # --- REPLACED: Official Google Cloud TTS Function ---
 
@@ -404,14 +457,29 @@ if st.session_state.stage not in ['start', 'processing_resume']:
     # --- REPLACEMENT: Text Input Area ---
     user_text = None # Initialize user_text
     is_disabled = (st.session_state.stage == 'finished')
-
-    with st.form(key="answer_form", clear_on_submit=True):
-        answer = st.text_input("Your answer:", disabled=is_disabled)
-        submit_button = st.form_submit_button(label="Submit Answer", disabled=is_disabled)
+    
+    if mic_recorder is None:
+        st.error("streamlit_mic_recorder library failed to import. Voice input is disabled.")
+        st.info("Please add 'streamlit-mic-recorder' to your requirements.txt")
+    
+    elif is_disabled:
+        st.info("Interview is finished. Start a new interview to speak.")
+        
+    else:
+        st.write("Your turn to speak:")
+        audio_bytes_dict = mic_recorder(
+            start_prompt="Start Recording ⏺️",
+            stop_prompt="Stop Recording ⏹️",
+            key='recorder'
+        )
         
-        if submit_button and answer:
-            user_text = answer
-            st.session_state.chat_history.append(f"**You:** {user_text}")
+        if audio_bytes_dict:
+            # The component returns a dictionary, get the bytes
+            audio_bytes = audio_bytes_dict['bytes']
+            
+            with st.spinner("Transcribing your answer..."):
+                # Use our NEW Google Cloud STT function
+                user_text = speech_to_text(audio_bytes)
     # --- END OF REPLACEMENT ---