Spaces:
Sleeping
Sleeping
File size: 3,951 Bytes
c5c9261 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 | """
Generate test audio samples for the Voice Detection API.
Creates both human-like and AI-like audio patterns for testing.
"""
import numpy as np
import soundfile as sf
from scipy import signal
import os
def create_human_voice_sample(duration=3.0, sample_rate=16000):
"""
Create a synthetic human-like voice sample.
Uses multiple harmonics and natural variations.
"""
t = np.linspace(0, duration, int(sample_rate * duration))
# Fundamental frequency (typical human voice range: 85-255 Hz)
f0 = 150 # Hz
# Create harmonics (what makes it sound voice-like)
audio = np.zeros_like(t)
for harmonic in range(1, 8):
frequency = f0 * harmonic
amplitude = 1.0 / harmonic # Decreasing amplitude for higher harmonics
# Add slight frequency modulation (vibrato)
vibrato = 5 * np.sin(2 * np.pi * 5 * t)
audio += amplitude * np.sin(2 * np.pi * (frequency + vibrato) * t)
# Add amplitude envelope (attack, sustain, release)
envelope = np.ones_like(t)
attack_samples = int(0.1 * sample_rate)
release_samples = int(0.2 * sample_rate)
envelope[:attack_samples] = np.linspace(0, 1, attack_samples)
envelope[-release_samples:] = np.linspace(1, 0, release_samples)
audio *= envelope
# Add some noise (breath sounds)
noise = np.random.normal(0, 0.02, len(t))
audio += noise
# Normalize
audio = audio / np.max(np.abs(audio)) * 0.8
return audio, sample_rate
def create_ai_voice_sample(duration=3.0, sample_rate=16000):
"""
Create a synthetic AI-like voice sample.
More regular patterns, less natural variation.
"""
t = np.linspace(0, duration, int(sample_rate * duration))
# More regular fundamental frequency
f0 = 180 # Hz
# Create very regular harmonics (AI voices tend to be more precise)
audio = np.zeros_like(t)
for harmonic in range(1, 10):
frequency = f0 * harmonic
amplitude = 1.0 / (harmonic * 1.2)
audio += amplitude * np.sin(2 * np.pi * frequency * t)
# More regular envelope
envelope = np.ones_like(t)
attack_samples = int(0.05 * sample_rate)
release_samples = int(0.1 * sample_rate)
envelope[:attack_samples] = np.linspace(0, 1, attack_samples)
envelope[-release_samples:] = np.linspace(1, 0, release_samples)
audio *= envelope
# Less noise (AI voices are cleaner)
noise = np.random.normal(0, 0.005, len(t))
audio += noise
# Normalize
audio = audio / np.max(np.abs(audio)) * 0.8
return audio, sample_rate
def main():
output_dir = "tests/audio_samples"
os.makedirs(output_dir, exist_ok=True)
print("Creating test audio samples...")
# Create human voice samples
print("1. Creating human voice sample...")
human_audio, sr = create_human_voice_sample(duration=3.0)
human_path = os.path.join(output_dir, "human_voice_test.wav")
sf.write(human_path, human_audio, sr)
print(f" ✅ Saved: {human_path}")
# Create AI voice samples
print("2. Creating AI voice sample...")
ai_audio, sr = create_ai_voice_sample(duration=3.0)
ai_path = os.path.join(output_dir, "ai_voice_test.wav")
sf.write(ai_path, ai_audio, sr)
print(f" ✅ Saved: {ai_path}")
# Create a short sample (for edge case testing)
print("3. Creating short sample...")
short_audio, sr = create_human_voice_sample(duration=0.5)
short_path = os.path.join(output_dir, "short_sample.wav")
sf.write(short_path, short_audio, sr)
print(f" ✅ Saved: {short_path}")
print(f"\n✅ All test audio samples created in: {output_dir}")
print("\nYou can now test these with:")
print(" python verify.py tests/audio_samples/human_voice_test.wav")
print(" python verify.py tests/audio_samples/ai_voice_test.wav")
if __name__ == "__main__":
main()
|