def testMaxAudioWithBaselineShift(self):
low_base = b"".join(["\x10\x00\x01\x00"] * 100)
higher_base = b"".join(["\x01\x00\x00\x01"] * 100)
source = MockSource()
for i in range(100):
source.stream.inject(low_base)
source.stream.inject(higher_base)
recognizer = ResponsiveRecognizer(None)
sec_per_buffer = float(source.CHUNK) / (source.SAMPLE_RATE *
source.SAMPLE_WIDTH)
test_seconds = 30.0
while test_seconds > 0:
test_seconds -= sec_per_buffer
data = source.stream.read(source.CHUNK)
energy = recognizer.calc_energy(data, source.SAMPLE_WIDTH)
recognizer.adjust_threshold(energy, sec_per_buffer)
higher_base_energy = audioop.rms(higher_base, source.SAMPLE_WIDTH)
# after recalibration (because of max audio length) new threshold
# should be >= 1.5 * higher_base_energy
delta_below_threshold = (
recognizer.energy_threshold - higher_base_energy)
min_delta = higher_base_energy * .5
assert abs(delta_below_threshold - min_delta) < 1
def testMaxAudioWithBaselineShift(self):
low_base = b"".join(["\x10\x00\x01\x00"] * 100)
higher_base = b"".join(["\x01\x00\x00\x01"] * 100)
source = MockSource()
for i in xrange(100):
source.stream.inject(low_base)
source.stream.inject(higher_base)
recognizer = ResponsiveRecognizer(None)
sec_per_buffer = float(
source.CHUNK) / (source.SAMPLE_RATE * source.SAMPLE_WIDTH)
test_seconds = 30.0
while test_seconds > 0:
test_seconds -= sec_per_buffer
data = source.stream.read(source.CHUNK)
energy = recognizer.calc_energy(data, source.SAMPLE_WIDTH)
recognizer.adjust_threshold(energy, sec_per_buffer)
higher_base_energy = audioop.rms(higher_base, source.SAMPLE_WIDTH)
# after recalibration (because of max audio length) new threshold
# should be >= 1.5 * higher_base_energy
delta_below_threshold = (recognizer.energy_threshold -
higher_base_energy)
min_delta = higher_base_energy * .5
assert abs(delta_below_threshold - min_delta) < 1
