Update app.py

app.py

@@ -4,9 +4,8 @@ import google.generativeai as genai
 from pytube import Search
 import speech_recognition as sr
 import tempfile
-from pydub import AudioSegment
+from audio_recorder_streamlit import audio_recorder
 import numpy as np
-from streamlit_webrtc import webrtc_streamer, WebRtcMode, AudioProcessorBase
 
 # Load sentiment analysis model using PyTorch backend
 mood_classifier = pipeline("sentiment-analysis", framework="pt")
@@ -21,32 +20,41 @@ def detect_mood(text):
     else:
         return "neutral"
 
-def speech_to_text(audio_bytes):
+def speech_to_text():
     # Initialize recognizer
     r = sr.Recognizer()
     
     # Create a temporary file to store the recorded audio
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as fp:
-        # Save audio bytes to temporary file
-        fp.write(audio_bytes)
-        temp_filename = fp.name
+        # Add audio recording widget
+        audio_bytes = audio_recorder(
+            text="Click to record",
+            recording_color="#e8b62c",
+            neutral_color="#6aa36f"
+        )
         
-        # Read the audio file
-        with sr.AudioFile(temp_filename) as source:
-            # Adjust for ambient noise and record
-            r.adjust_for_ambient_noise(source)
-            audio = r.record(source)
+        if audio_bytes:
+            # Save audio bytes to temporary file
+            fp.write(audio_bytes)
+            temp_filename = fp.name
             
-            try:
-                # Use Google Speech Recognition
-                text = r.recognize_google(audio)
-                return text
-            except sr.UnknownValueError:
-                st.error("Could not understand the audio")
-                return None
-            except sr.RequestError:
-                st.error("Could not request results from speech recognition service")
-                return None
+            # Read the audio file
+            with sr.AudioFile(temp_filename) as source:
+                # Adjust for ambient noise and record
+                r.adjust_for_ambient_noise(source)
+                audio = r.record(source)
+                
+                try:
+                    # Use Google Speech Recognition
+                    text = r.recognize_google(audio)
+                    return text
+                except sr.UnknownValueError:
+                    st.error("Could not understand the audio")
+                    return None
+                except sr.RequestError:
+                    st.error("Could not request results from speech recognition service")
+                    return None
+    return None
 
 def get_song_recommendations(mood, api_key):
     try:
@@ -93,49 +101,34 @@ gemini_api_key = st.sidebar.text_input("Enter your Gemini API Key:", type="passw
 # Add option to choose between text and speech input
 input_method = st.sidebar.radio("Choose input method:", ["Text", "Speech"])
 
+# Initialize user_mood as None
+user_mood = None
+user_text = None
+
 if input_method == "Text":
     # Text input
     user_mood = st.sidebar.selectbox("Select your mood:", mood_options)
+    user_text = user_mood
 else:
     # Speech input
     st.write("📢 Tell me about your day...")
+    user_text = speech_to_text()
     
-    # Use streamlit-webrtc for audio recording
-    webrtc_ctx = webrtc_streamer(
-        key="speech-to-text",
-        mode=WebRtcMode.SENDONLY,
-        audio_receiver_size=1024,
-        media_stream_constraints={"audio": True, "video": False},
-    )
-    
-    if webrtc_ctx.audio_receiver:
-        audio_frames = webrtc_ctx.audio_receiver.get_frames(timeout=5)
-        if audio_frames:
-            audio_bytes = b"".join([frame.to_ndarray().tobytes() for frame in audio_frames])
-            spoken_text = speech_to_text(audio_bytes)
-            
-            if spoken_text:
-                st.write(f"You said: {spoken_text}")
-                user_mood = detect_mood(spoken_text)
-            else:
-                user_mood = None
-        else:
-            st.warning("No audio frames received. Please try again.")
-            user_mood = None
-    else:
-        user_mood = None
+    if user_text:
+        st.write(f"You said: {user_text}")
+        user_mood = detect_mood(user_text)
 
 # Playlist
 if 'playlist' not in st.session_state:
     st.session_state.playlist = []
 
 # Main content
-if (user_mood or spoken_text) and gemini_api_key:
-    mood = detect_mood(user_mood if user_mood else spoken_text)
-    st.write(f"🎭 Detected Mood: **{mood}**")
+if user_text and gemini_api_key:
+    detected_mood = detect_mood(user_text)
+    st.write(f"🎭 Detected Mood: **{detected_mood}**")
     
     st.write("🎵 Recommended Songs:")
-    recommendations = get_song_recommendations(mood, gemini_api_key)
+    recommendations = get_song_recommendations(detected_mood, gemini_api_key)
     if recommendations:
         st.write(recommendations)