update(9)

app.py

@@ -17,48 +17,15 @@ agent = NewsReporterAgent()
 
 # --- 2. Define Gradio Logic Handlers ---
 # These functions orchestrate the agent's actions based on UI events.
-def run_initial_generation(audio_data, image_path):
+def run_initial_generation(audio_path, image_path):
     """Handles the first step: processing inputs and generating the initial report."""
-
-    temp_audio_path = None
-    TARGET_SR = 16000 # The model's required sample rate
-
-    if audio_data is not None:
-        # --- Start of New Audio Processing Logic ---
-
-        # Load audio regardless of source (path or tuple)
-        if isinstance(audio_data, str):
-            waveform, sample_rate = librosa.load(audio_data, sr=None, mono=False)
-        elif isinstance(audio_data, tuple):
-            sample_rate, waveform = audio_data
-            # Ensure waveform is float for librosa
-            waveform = waveform.astype(np.float32)
-
-        # 1. Convert to Mono if it's stereo
-        if waveform.ndim > 1:
-            waveform = librosa.to_mono(waveform.T)
-
-        # 2. Resample to the target sample rate (16kHz) if necessary
-        if sample_rate != TARGET_SR:
-            waveform = librosa.resample(y=waveform, orig_sr=sample_rate, target_sr=TARGET_SR)
-
-        # 3. Normalize and convert to 16-bit integer format for saving
-        waveform_int16 = (waveform * 32767).astype(np.int16)
-
-        # Save the processed audio to a standardized file
-        os.makedirs("temp_audio", exist_ok=True)
-        temp_audio_path = "temp_audio/processed_audio.wav"
-        wavfile.write(temp_audio_path, rate=TARGET_SR, data=waveform_int16)
-
-        # --- End of New Audio Processing Logic ---
-
-    if not temp_audio_path and not image_path:
+    if not audio_path and not image_path:
         return "Please provide an audio or image file.", None, gr.update(visible=False), None, None, None
 
-    state = AgentState(audio_path=temp_audio_path, 
+    state = AgentState(audio_path=audio_path, 
                        image_path=image_path, 
                        news_report=[])
-
+    
     state.update(agent.transcribe_audio(state))
     state.update(agent.describe_image(state))
     state.update(agent.create_report(state))
@@ -71,6 +38,7 @@ def run_initial_generation(audio_data, image_path):
 
 
 
+
 def run_revision(feedback, current_state):
     """Handles the revision step based on user feedback."""
     if not feedback or not feedback.strip():