Update app.py

app.py

@@ -5,6 +5,7 @@ import speech_recognition as sr
 from streamlit_webrtc import webrtc_streamer, AudioProcessorBase, WebRtcMode
 import numpy as np
 from gtts import gTTS
+import asyncio
 
 # Function to convert text to speech and return audio file
 def text_to_speech(text):
@@ -15,11 +16,20 @@ def text_to_speech(text):
     return audio_file
 
 # Function to convert speech to text using SpeechRecognition
-def speech_to_text(audio_file):
+def speech_to_text(audio_data):
     recognizer = sr.Recognizer()
-    with sr.AudioFile(audio_file) as source:
-        audio_data = recognizer.record(source)
-        return recognizer.recognize_google(audio_data)
+    with sr.AudioFile(BytesIO(audio_data)) as source:
+        audio = recognizer.record(source)
+        return recognizer.recognize_google(audio)
+
+class AudioProcessor(AudioProcessorBase):
+    def __init__(self):
+        self.audio_buffer = BytesIO()
+
+    def recv(self, frame):
+        audio_data = frame.to_ndarray().tobytes()
+        self.audio_buffer.write(audio_data)
+        return frame
 
 def main():
     st.title("Real-Time Audio Chat with AI")
@@ -36,42 +46,52 @@ def main():
             st.write("AI:")
         st.audio(message["audio"], format="audio/mp3")
 
-    # Upload audio input from the user
-    user_audio_file = st.file_uploader("Upload your voice message (in WAV format):", type=["wav"])
-
-    if user_audio_file is not None:
-        # Convert speech to text
-        user_text = speech_to_text(user_audio_file)
-        
-        # Display the converted text (optional)
-        st.write(f"Recognized Text: {user_text}")
-
-        # Convert user's speech input to audio for playback
-        user_audio = text_to_speech(user_text)
-
-        # Add user message (as audio) to chat history
-        st.session_state.chat_messages.append({
-            "role": "user",
-            "content": user_text,
-            "audio": user_audio.getvalue()
-        })
-
-        # Get AI response using responsr function
-        response = responsr(user_text)
-
-        # Convert AI response to speech
-        response_audio = text_to_speech(response)
-
-        # Add assistant's response (as audio) to chat history
-        st.session_state.chat_messages.append({
-            "role": "assistant",
-            "content": response,
-            "audio": response_audio.getvalue()
-        })
-
-        # Display the audio files for both user input and AI response
-        st.audio(user_audio, format="audio/mp3")
-        st.audio(response_audio, format="audio/mp3")
+    # Initialize WebRTC audio streamer
+    webrtc_ctx = webrtc_streamer(
+        key="speech-to-text",
+        mode=WebRtcMode.SENDRECV,
+        audio_processor_factory=AudioProcessor,
+        rtc_configuration={"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]},
+        media_stream_constraints={"audio": True, "video": False},
+        async_processing=True,
+    )
+
+    if webrtc_ctx.state.playing:
+        processor = webrtc_ctx.audio_processor
+        if processor and processor.audio_buffer.getvalue():
+            # Convert audio buffer to speech
+            audio_data = processor.audio_buffer.getvalue()
+            user_text = speech_to_text(audio_data)
+            
+            # Convert the user's speech input to audio for playback
+            user_audio = text_to_speech(user_text)
+
+            # Add user message (as audio) to chat history
+            st.session_state.chat_messages.append({
+                "role": "user",
+                "content": user_text,
+                "audio": user_audio.getvalue()
+            })
+
+            # Get AI response using the responsr function
+            response = responsr(user_text)
+
+            # Convert AI response to speech
+            response_audio = text_to_speech(response)
+
+            # Add assistant's response (as audio) to chat history
+            st.session_state.chat_messages.append({
+                "role": "assistant",
+                "content": response,
+                "audio": response_audio.getvalue()
+            })
+
+            # Display the audio files for both user input and AI response
+            st.audio(user_audio, format="audio/mp3")
+            st.audio(response_audio, format="audio/mp3")
+
+            # Clear the audio buffer
+            processor.audio_buffer = BytesIO()
 
 if __name__ == "__main__":
     main()