def test_vad_is_triggered(mock_class):
context = SpeechContext()
detector = VoiceActivityDetector(sample_rate=16000,
frame_width=10,
vad_rise_delay=0,
vad_fall_delay=0)
frame = np.zeros(160, np.int16)
detector(context, frame)
assert context.is_speech
detector.close()
def test_vad_rise_delay():
context = SpeechContext()
detector = VoiceActivityDetector(frame_width=10, vad_rise_delay=30)
for i in range(3):
frame = voice_frame()
detector(context, frame)
if i < 2:
assert not context.is_speech
else:
assert context.is_speech
detector.close()
def test_vad_is_triggered():
context = SpeechContext()
detector = VoiceActivityDetector(frame_width=10)
frame = silence_frame()
detector(context, frame)
assert not context.is_speech
frame = voice_frame()
detector(context, frame)
assert context.is_speech
detector.close()
def test_vad_rise_delay(mock_class):
context = SpeechContext()
detector = VoiceActivityDetector(sample_rate=16000,
frame_width=10,
vad_rise_delay=30,
vad_fall_delay=0)
for i in range(3):
frame = np.zeros(160, np.int16)
detector(context, frame)
if i < 2:
assert not context.is_speech
else:
assert context.is_speech
detector.close()
def test_vad_fall_untriggered():
context = SpeechContext()
detector = VoiceActivityDetector(frame_width=10, vad_fall_delay=20)
voice = voice_frame()
silence = silence_frame()
detector(context, voice)
assert context.is_speech
for i in range(10):
detector(context, silence)
assert context.is_speech
detector(context, silence)
assert not context.is_speech
detector.close()
def test_vad_fall_untriggered():
context = SpeechContext()
detector = VoiceActivityDetector(sample_rate=16000,
frame_width=10,
vad_rise_delay=0,
vad_fall_delay=20)
with patch("webrtcvad.Vad.is_speech", return_value=True):
frame = np.zeros(160, np.int16)
detector(context, frame)
assert context.is_speech
with patch("webrtcvad.Vad.is_speech", return_value=False):
frame = np.zeros(160, np.int16)
detector(context, frame)
assert context.is_speech
with patch("webrtcvad.Vad.is_speech", return_value=False):
frame = np.zeros(160, np.int16)
detector(context, frame)
assert not context.is_speech
detector.close()