amine@2: """
amine@2: @author: Amine SEHILI <amine.sehili@gmail.com>
amine@2: September, 2015
amine@2: """
amine@2: 
amine@10: # Trim leading and trailing silence from a record
amine@2: 
amine@331: from auditok import (
amine@331:     ADSFactory,
amine@331:     AudioEnergyValidator,
amine@331:     StreamTokenizer,
amine@331:     player_for,
amine@331:     dataset,
amine@331: )
amine@2: import pyaudio
amine@10: import sys
amine@2: 
amine@2: """
amine@2: The  tokenizer in the following example is set up to remove the silence
amine@331: that precedes the first acoustic activity or follows the last activity
amine@2: in a record. It preserves whatever it founds between the two activities.
amine@10: In other words, it removes the leading and trailing silence.
amine@2: 
amine@2: Sampling rate is 44100 sample per second, we'll use an analysis window of 100 ms
amine@2: (i.e. bloc_ksize == 4410)
amine@2: 
amine@2: Energy threshold is 50.
amine@2: 
amine@2: The tokenizer will start accumulating windows up from the moment it encounters
amine@331: the first analysis window of an energy >= 50. ALL the following windows will be
amine@10: kept regardless of their energy. At the end of the analysis, it will drop trailing
amine@2:  windows with an energy below 50.
amine@2: 
amine@2: This is an interesting example because the audio file we're analyzing contains a very
amine@331: brief noise that occurs within the leading silence. We certainly do want our tokenizer
amine@2: to stop at this point and considers whatever it comes after as a useful signal.
amine@2: To force the tokenizer to ignore that brief event we use two other parameters `init_min`
amine@2: ans `init_max_silence`. By `init_min`=3 and `init_max_silence`=1 we tell the tokenizer
amine@2: that a valid event must start with at least 3 noisy windows, between which there
amine@2: is at most 1 silent window.
amine@2: 
amine@2: Still with this configuration we can get the tokenizer detect that noise as a valid event
amine@2: (if it actually contains 3 consecutive noisy frames). To circummvent this we use an enough
amine@2: large analysis window (here of 100 ms) to ensure that the brief noise be surrounded by a much
amine@2: longer silence and hence the energy of the overall analysis window will be below 50.
amine@2: 
amine@2: When using a shorter analysis window (of 10ms for instance, block_size == 441), the brief
amine@2: noise contributes more to energy calculation which yields an energy of over 50 for the window.
amine@2: Again we can deal with this situation by using a higher energy threshold (55 for example)
amine@331: 
amine@2: """
amine@2: 
amine@10: try:
amine@331:     # record = True so that we'll be able to rewind the source.
amine@331:     asource = ADSFactory.ads(
amine@331:         filename=dataset.was_der_mensch_saet_mono_44100_lead_tail_silence,
amine@331:         record=True,
amine@331:         block_size=4410,
amine@331:     )
amine@331:     asource.open()
amine@2: 
amine@331:     original_signal = []
amine@331:     # Read the whole signal
amine@331:     while True:
amine@331:         w = asource.read()
amine@331:         if w is None:
amine@331:             break
amine@331:         original_signal.append(w)
amine@2: 
amine@331:     original_signal = b"".join(original_signal)
amine@2: 
amine@331:     # rewind source
amine@331:     asource.rewind()
amine@2: 
amine@331:     # Create a validator with an energy threshold of 50
amine@331:     validator = AudioEnergyValidator(
amine@331:         sample_width=asource.get_sample_width(), energy_threshold=50
amine@331:     )
amine@2: 
amine@331:     # Create a tokenizer with an unlimited token length and continuous silence within a token
amine@331:     # Note the DROP_TRAILING_SILENCE mode that will ensure removing trailing silence
amine@331:     trimmer = StreamTokenizer(
amine@331:         validator,
amine@331:         min_length=20,
amine@331:         max_length=99999999,
amine@331:         max_continuous_silence=9999999,
amine@331:         mode=StreamTokenizer.DROP_TRAILING_SILENCE,
amine@331:         init_min=3,
amine@331:         init_max_silence=1,
amine@331:     )
amine@2: 
amine@331:     tokens = trimmer.tokenize(asource)
amine@2: 
amine@331:     # Make sure we only have one token
amine@331:     assert len(tokens) == 1, "Should have detected one single token"
amine@2: 
amine@331:     trimmed_signal = b"".join(tokens[0][0])
amine@2: 
amine@331:     player = player_for(asource)
amine@2: 
amine@331:     print(
amine@331:         "\n ** Playing original signal (with leading and trailing silence)..."
amine@331:     )
amine@331:     player.play(original_signal)
amine@331:     print("\n ** Playing trimmed signal...")
amine@331:     player.play(trimmed_signal)
amine@2: 
amine@331:     player.stop()
amine@331:     asource.close()
amine@2: 
amine@10: except KeyboardInterrupt:
amine@10: 
amine@331:     player.stop()
amine@331:     asource.close()
amine@331:     sys.exit(0)
amine@10: 
amine@10: except Exception as e:
amine@331: 
amine@331:     sys.stderr.write(str(e) + "\n")
amine@331:     sys.exit(1)