audio_pause_safe.py

import sounddevice as sd
import numpy as np
from scipy.signal import find_peaks

DURATION = 10
FS = 44100

print("🎤 Speak naturally for", DURATION, "seconds...")

# BLOCKING RECORD (no callbacks → no crash)
audio = sd.rec(
    int(DURATION * FS),
    samplerate=FS,
    channels=1,
    blocking=True
)

signal = audio.flatten()
energy = np.abs(signal)

# Normalize safely
max_val = np.max(energy)
if max_val > 0:
    energy = energy / max_val

# Detect speech peaks
peaks, _ = find_peaks(energy, height=0.05)

speech_events = len(peaks)
speech_rate = speech_events / DURATION

# Pause estimation
silence_threshold = 0.02
pause_ratio = np.mean(energy < silence_threshold)

print("\n--- Audio Behavior Metrics ---")
print("Speech events:", speech_events)
print("Speech rate (events/sec):", round(speech_rate, 2))
print("Silence ratio:", round(pause_ratio, 2))

audio_score = int(min(100, (pause_ratio * 100) + (speech_rate * 10)))
print("Audio Behavioral Load Score:", audio_score)