Add 3 new demo scenarios + Coming Soon badge + lazy generation

New demos: Wake Word (Alexa + beep), Synthetic/Robotic Voice,
Two Different Speakers. Total 6 demo scenarios. Demos generate
on first app startup instead of build time. Added Batch Analysis
Coming Soon badge below Upload section.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

app.py

@@ -298,7 +298,10 @@ def render_analyzer_tab():
     demo_files = {
         "Clean Exam (No Fraud)": "demo_clean_exam.wav",
         "Reading Pattern + Long Pause": "demo_reading_fraud.wav",
-        "Multiple Speakers + Whisper": "demo_coaching_fraud.wav",
+        "Coaching (Whispered Prompts)": "demo_coaching_fraud.wav",
+        "Wake Word (Alexa + Beep)": "demo_wake_word.wav",
+        "Synthetic / Robotic Voice": "demo_synthetic_voice.wav",
+        "Two Different Speakers": "demo_two_speakers.wav",
     }
     available_demos = {
         name: os.path.join(samples_dir, fname)

generate_demos.py

@@ -85,6 +85,103 @@ def generate_coaching_fraud():
     )
 
 
+def generate_wake_word():
+    """Scenario 4: Wake word - someone asks Alexa mid-exam."""
+    import numpy as np
+
+    main1 = _tts("I think the most important factor in choosing a career is job satisfaction.")
+    main2 = _tts("Because if you enjoy what you do, you'll be more productive and motivated.")
+    main3 = _tts("Sorry, what was I saying? Oh yes, career satisfaction is crucial for long term happiness.")
+    main4 = _tts("And that's why I believe people should follow their passion when choosing a career.")
+
+    # "Alexa" wake word spoken quietly
+    alexa_call = _tts("Alexa, what is career satisfaction", lang="en") - 12
+
+    # Synthetic beep tone (1kHz, 0.3s) to simulate device response
+    sample_rate = 16000
+    t = np.linspace(0, 0.3, int(sample_rate * 0.3))
+    beep = (np.sin(2 * np.pi * 1000 * t) * 0.15 * 32767).astype(np.int16)
+    beep_seg = AudioSegment(beep.tobytes(), frame_rate=sample_rate, sample_width=2, channels=1)
+
+    return (
+        main1
+        + _silence(1000)
+        + main2
+        + _silence(1500)
+        + alexa_call
+        + _silence(400)
+        + beep_seg
+        + _silence(2000)
+        + main3
+        + _silence(800)
+        + main4
+    )
+
+
+def generate_synthetic_voice():
+    """Scenario 5: Synthetic/robotic voice with beep tones."""
+    import numpy as np
+
+    # Use slow TTS to sound more robotic
+    robo1 = _tts("The answer to question number one is as follows.", slow=True)
+    robo2 = _tts("Climate change is caused by greenhouse gas emissions from human activities.", slow=True)
+    robo3 = _tts("The main sources include transportation, industry, and agriculture.", slow=True)
+    robo4 = _tts("In conclusion, reducing emissions requires global cooperation.", slow=True)
+
+    # Generate beep tones between segments (synthetic indicator)
+    sample_rate = 16000
+    def make_beep(freq, dur):
+        t = np.linspace(0, dur, int(sample_rate * dur))
+        tone = (np.sin(2 * np.pi * freq * t) * 0.2 * 32767).astype(np.int16)
+        return AudioSegment(tone.tobytes(), frame_rate=sample_rate, sample_width=2, channels=1)
+
+    beep_hi = make_beep(880, 0.15)
+    beep_lo = make_beep(440, 0.15)
+    double_beep = beep_hi + _silence(100) + beep_lo
+
+    return (
+        double_beep
+        + _silence(300)
+        + robo1
+        + _silence(400)
+        + robo2
+        + double_beep
+        + _silence(500)
+        + robo3
+        + _silence(400)
+        + robo4
+        + _silence(200)
+        + double_beep
+    )
+
+
+def generate_two_speakers():
+    """Scenario 6: Two distinct speakers taking turns (impersonation/proxy)."""
+    # Speaker A - normal pace, English
+    a1 = _tts("Hello, my name is John and I'm here to take the speaking test today.")
+    a2 = _tts("I think that learning a second language is very important in today's world.")
+    a3 = _tts("For example, it helps you communicate with people from different countries.")
+
+    # Speaker B - different accent (UK English) to sound like a different person
+    b1 = _tts("Right, so the next topic is about technology in education.", lang="en-uk")
+    b2 = _tts("Technology has completely changed the way students learn and interact with content.", lang="en-uk")
+    b3 = _tts("Online courses and digital tools make education more accessible to everyone.", lang="en-uk")
+
+    return (
+        a1
+        + _silence(1000)
+        + a2
+        + _silence(800)
+        + a3
+        + _silence(2500)  # Longer pause as speakers switch
+        + b1
+        + _silence(700)
+        + b2
+        + _silence(600)
+        + b3
+    )
+
+
 def main():
     os.makedirs(OUTPUT_DIR, exist_ok=True)
 
@@ -92,6 +189,9 @@ def main():
         "demo_clean_exam.wav": generate_clean_exam,
         "demo_reading_fraud.wav": generate_reading_fraud,
         "demo_coaching_fraud.wav": generate_coaching_fraud,
+        "demo_wake_word.wav": generate_wake_word,
+        "demo_synthetic_voice.wav": generate_synthetic_voice,
+        "demo_two_speakers.wav": generate_two_speakers,
     }
 
     for filename, generator in demos.items():