Update app.py

app.py

@@ -14,6 +14,12 @@ import matplotlib.pyplot as plt
 from huggingface_hub import login
 import os
 from dotenv import load_dotenv
+import whisper
+import sounddevice as sd
+import queue
+import tempfile
+import scipy.io.wavfile as wav
+
 
 SPREADSHEET_ID = "1CsBub3Jlwyo7WHMQty6SDnBShIZMjl5XTVSoOKrxZhc"
 RANGE_NAME = 'Sheet1!A1:E'
@@ -379,65 +385,145 @@ def calculate_overall_sentiment(sentiment_scores):
         overall_sentiment = "NEUTRAL"
     return overall_sentiment
 
-def process_real_time_audio():
-    recognizer = sr.Recognizer()
-    microphone = sr.Microphone()
-
-    st.write("Adjusting microphone for ambient noise... Please wait.")
-    with microphone as source:
-        recognizer.adjust_for_ambient_noise(source,duration=2)
-
-    st.write("Listening for audio... Speak into the microphone.")
-    while True:
-        try:
-            with microphone as source:
-                audio = recognizer.listen(source, timeout=15, phrase_time_limit=20)
-
-
-
-            st.write("Transcribing audio...")
-            transcribed_text = recognizer.recognize_google(audio)
-            st.write(f"You said: {transcribed_text}")
-
-            if 'stop' in transcribed_text.lower():
-                st.warning("Stopping the speech recognition process.")
+# def process_real_time_audio():
+#     recognizer = sr.Recognizer()
+#     microphone = sr.Microphone()
+
+#     st.write("Adjusting microphone for ambient noise... Please wait.")
+#     with microphone as source:
+#         recognizer.adjust_for_ambient_noise(source,duration=2)
+
+#     st.write("Listening for audio... Speak into the microphone.")
+#     while True:
+#         try:
+#             with microphone as source:
+#                 audio = recognizer.listen(source, timeout=15, phrase_time_limit=20)
+
+
+
+#             st.write("Transcribing audio...")
+#             transcribed_text = recognizer.recognize_google(audio)
+#             st.write(f"You said: {transcribed_text}")
+
+#             if 'stop' in transcribed_text.lower():
+#                 st.warning("Stopping the speech recognition process.")
+#                 break
+
+#             st.markdown("### *Sentiment Analysis*")
+#             sentiment_label, sentiment_score = analyze_sentiment(transcribed_text)
+#             st.write(f"Sentiment: {sentiment_label}")
+#             st.write(f"Sentiment Score: {sentiment_score}")
+
+#             closest_objection = None
+#             response = None
+
+#             add_to_sentiment_history(transcribed_text, sentiment_label, sentiment_score, closest_objection, response)
+#             st.markdown("### *Recommendations*")
+#             recommendations = query_crm_data_with_context(transcribed_text)
+#             for i, rec in enumerate(recommendations, start=1):
+#                 if isinstance(rec, dict) and 'Product' in rec and 'Recommendations' in rec:
+#                     st.markdown(f"- *{rec['Product']}*: {rec['Recommendations']}")
+#                 else:
+#                     st.markdown(f"- {rec}")
+
+#             st.markdown("### *Objection Handling*")
+#             closest_objection, response = find_closest_objection(transcribed_text)
+#             st.write(f"Objection: {closest_objection}")
+#             st.write(f" Response: {response}")
+
+#             update_google_sheet(
+#                 transcribed_text=transcribed_text,
+#                 sentiment=f"{sentiment_label} ({sentiment_score})",
+#                 objection=f"Objection: {closest_objection} | Response: {response}",
+#                 recommendations=str(recommendations),
+#                 overall_sentiment=f"{sentiment_label}"
+#             )
+
+#         except sr.UnknownValueError:
+#             st.warning("Could not understand the audio.")
+#         except Exception as e:
+#             st.error(f"Error: {e}")
+#             break
+model = whisper.load_model("base")
+
+# Queue for streaming audio
+audio_queue = queue.Queue()
+
+def audio_callback(indata, frames, time, status):
+    """Callback function to continuously receive audio chunks."""
+    if status:
+        st.warning(f"Audio Status: {status}")
+    audio_queue.put(indata.copy())
+
+def transcribe_audio_stream():
+    """Continuously captures microphone input, transcribes, and processes the speech."""
+    samplerate = 16000
+    duration = 5  # Adjust duration for processing chunks in real-time
+
+    # Set up the microphone stream
+    with sd.InputStream(samplerate=samplerate, channels=1, callback=audio_callback):
+        st.write("Listening... Speak into the microphone.")
+
+        while True:
+            try:
+                # Collect audio chunks
+                audio_chunk = []
+                for _ in range(int(samplerate / 1024 * duration)):  # Collect chunks for `duration` seconds
+                    audio_chunk.append(audio_queue.get())
+
+                # Convert to NumPy array
+                audio_data = np.concatenate(audio_chunk, axis=0)
+
+                # Save the chunk as a temporary WAV file
+                with tempfile.NamedTemporaryFile(delete=True, suffix=".wav") as temp_audio:
+                    wav.write(temp_audio.name, samplerate, np.int16(audio_data * 32767))
+
+                    # Transcribe using Whisper
+                    result = model.transcribe(temp_audio.name)
+                    transcribed_text = result["text"]
+
+                st.write(f"You said: {transcribed_text}")
+
+                if 'stop' in transcribed_text.lower():
+                    st.warning("Stopping speech recognition.")
+                    break
+
+                # Sentiment Analysis
+                st.markdown("### *Sentiment Analysis*")
+                sentiment_label, sentiment_score = analyze_sentiment(transcribed_text)
+                st.write(f"Sentiment: {sentiment_label}")
+                st.write(f"Sentiment Score: {sentiment_score}")
+
+                # Add to history
+                add_to_sentiment_history(transcribed_text, sentiment_label, sentiment_score, None, None)
+
+                # Recommendations
+                st.markdown("### *Recommendations*")
+                recommendations = query_crm_data_with_context(transcribed_text)
+                for rec in recommendations:
+                    if isinstance(rec, dict) and 'Product' in rec and 'Recommendations' in rec:
+                        st.markdown(f"- *{rec['Product']}*: {rec['Recommendations']}")
+                    else:
+                        st.markdown(f"- {rec}")
+
+                # Objection Handling
+                st.markdown("### *Objection Handling*")
+                closest_objection, response = find_closest_objection(transcribed_text)
+                st.write(f"Objection: {closest_objection}")
+                st.write(f"Response: {response}")
+
+                # Update Google Sheets
+                update_google_sheet(
+                    transcribed_text=transcribed_text,
+                    sentiment=f"{sentiment_label} ({sentiment_score})",
+                    objection=f"Objection: {closest_objection} | Response: {response}",
+                    recommendations=str(recommendations),
+                    overall_sentiment=f"{sentiment_label}"
+                )
+
+            except Exception as e:
+                st.error(f"Error: {e}")
                 break
-
-            st.markdown("### *Sentiment Analysis*")
-            sentiment_label, sentiment_score = analyze_sentiment(transcribed_text)
-            st.write(f"Sentiment: {sentiment_label}")
-            st.write(f"Sentiment Score: {sentiment_score}")
-
-            closest_objection = None
-            response = None
-
-            add_to_sentiment_history(transcribed_text, sentiment_label, sentiment_score, closest_objection, response)
-            st.markdown("### *Recommendations*")
-            recommendations = query_crm_data_with_context(transcribed_text)
-            for i, rec in enumerate(recommendations, start=1):
-                if isinstance(rec, dict) and 'Product' in rec and 'Recommendations' in rec:
-                    st.markdown(f"- *{rec['Product']}*: {rec['Recommendations']}")
-                else:
-                    st.markdown(f"- {rec}")
-
-            st.markdown("### *Objection Handling*")
-            closest_objection, response = find_closest_objection(transcribed_text)
-            st.write(f"Objection: {closest_objection}")
-            st.write(f" Response: {response}")
-
-            update_google_sheet(
-                transcribed_text=transcribed_text,
-                sentiment=f"{sentiment_label} ({sentiment_score})",
-                objection=f"Objection: {closest_objection} | Response: {response}",
-                recommendations=str(recommendations),
-                overall_sentiment=f"{sentiment_label}"
-            )
-
-        except sr.UnknownValueError:
-            st.warning("Could not understand the audio.")
-        except Exception as e:
-            st.error(f"Error: {e}")
-            break
   
 def generate_sentiment_pie_chart(sentiment_history):
     if not sentiment_history: