Modified Recording Logic

New streamlit approach has been added on to the model for modification of the audio input response

app.py

@@ -19,6 +19,7 @@ from twilio.rest import Client
 import logging
 import whisper
 import speech_recognition as sr
+from audiorecorder import audiorecorder
 #model = whisper.load_model("base")
 
 
@@ -830,7 +831,7 @@ if st.session_state["generated_questions"]:
                 "question_played": True,
                 "question_start_time": time.time(),
                 "record_phase": "audio_playing",
-                "recorded_text": ""
+                # "recorded_text": ""
             })
 
         st.markdown(f"**Q{idx + 1}:** {question}")
@@ -881,7 +882,78 @@ if st.session_state["generated_questions"]:
             if remaining > 0:
                 st.markdown(f"<h4 class='timer-text'>🎙️ {remaining} seconds to answer...</h4>", unsafe_allow_html=True)
                 
-                audio_value = st.audio_input("🎤 Tap to record — then stop when done", key=f"audio_{idx}")
+                # audio_value = st.audio_input("🎤 Tap to record — then stop when done", key=f"audio_{idx}")
+                audio = audiorecorder("🎙️ Start Recording", "⏹️ Stop Recording")
+                if len(audio) > 0 and "response_file" not in st.session_state:
+                    if np.max(audio) > 0.01:
+                        wav_path = f"response_{idx}.wav"
+                        with open(wav_path, "wb") as f:
+                            f.write(audio.tobytes())
+                        st.session_state["response_file"] = wav_path
+                        st.session_state["record_phase"] = "listening"
+                        st.success("✅ Audio recorded. Please confirm to proceed.")
+                        st.audio(wav_path, format="audio/wav")
+                        st.rerun()
+                    else:
+                        st.warning("⚠️ Silence detected. Please try again.")
+                        st.rerun()
+
+                else:
+                    time.sleep(1)
+                    st.rerun()
+
+            else:
+                if not st.session_state.get("response_file"):
+                    st.warning("⚠️ No audio captured. Moving to next question.")
+                    st.session_state["answers"].append({"question": question, "response": "[No response]"})
+                    st.session_state.update({
+                        "record_phase": "idle",
+                        "question_played": False,
+                        "current_question_index": idx + 1
+                    })
+                    if st.session_state["current_question_index"] == len(st.session_state["generated_questions"]):
+                        evaluate_answers()
+                        st.session_state["show_summary"] = True
+                    st.rerun()
+        elif st.session_state["record_phase"] == "listening":
+            st.success("🎧 Review your recorded response below:")
+            st.audio(st.session_state["response_file"], format="audio/wav")
+
+            if st.button("⏹️ Confirm & Next"):
+                recognizer = sr.Recognizer()
+                try:
+                    with sr.AudioFile(st.session_state["response_file"]) as source:
+                        audio = recognizer.record(source)
+                        transcript = recognizer.recognize_google(audio)
+                except sr.UnknownValueError:
+                    transcript = "[Could not understand audio]"
+                except sr.RequestError:
+                    transcript = "[Google API error]"
+                except Exception as e:
+                    transcript = f"[Transcription failed: {e}]"
+
+                st.session_state["answers"].append({
+                    "question": question,
+                    "response_file": st.session_state["response_file"],
+                    "response": transcript
+                })
+
+                st.session_state.update({
+                    "record_phase": "idle",
+                    "recording_started": False,
+                    "question_played": False,
+                    "question_start_time": 0.0,
+                    "current_question_index": idx + 1,
+                    "response_file": None,
+                    "audio_waiting": True
+                })
+
+                if st.session_state["current_question_index"] == len(st.session_state["generated_questions"]):
+                    evaluate_answers()
+                    st.session_state["show_summary"] = True
+                st.rerun()
+                        
+                """
                 if audio_value and "response_file" not in st.session_state:
                     wav_path = f"response_{idx}.wav"
                     with open(wav_path, "wb") as f:
@@ -904,7 +976,7 @@ if st.session_state["generated_questions"]:
                     st.session_state["record_phase"] = "listening"
                     st.success("✅ Audio uploaded. You may now confirm your answer.")
                     #st.audio(wav_path, format="audio/wav")
-                    """
+                    ""
                     if st.button("⏹️ Confirm & Next"):
                         try:
                             with st.spinner("🧠 Transcribing your answer..."):
@@ -916,7 +988,7 @@ if st.session_state["generated_questions"]:
                         except Exception as e:
                                 st.error(f"❌ Transcription error: {e}")
                                 transcript = "[Transcription error]"
-                        """
+                        ""
                             
                     st.session_state["answers"].append({
                         "question": question,
@@ -991,7 +1063,7 @@ if st.session_state["generated_questions"]:
                     evaluate_answers()
                     st.session_state["show_summary"] = True
                 st.rerun()
-
+                """
 
 # === Summary Display ===