Class: Diarize::Audio

Inherits:
Object
  • Object
show all
Includes:
ToRdf
Defined in:
lib/diarize/audio.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(uri) ⇒ Audio

Returns a new instance of Audio.



31
32
33
34
35
36
37
38
39
40
41
42
 
# File 'lib/diarize/audio.rb', line 31

def initialize(uri)
  @uri = uri
  if uri.scheme == 'file'
    # Local file
    @path = uri.path
  else
    # Remote file, we get it locally
    @path = '/tmp/' + URI.escape(uri.to_s, Regexp.new("[^#{URI::PATTERN::UNRESERVED}]"))
    Kernel.system("wget #{uri} -O #{@path}")
  end
  @file = File.new @path
end

Instance Attribute Details

#file ⇒ Object (readonly)

Returns the value of attribute file.



29
30
31
 
# File 'lib/diarize/audio.rb', line 29

def file
  @file
end

#path ⇒ Object (readonly)

Returns the value of attribute path.



29
30
31
 
# File 'lib/diarize/audio.rb', line 29

def path
  @path
end

Instance Method Details

#analyze!(train_speaker_models = true) ⇒ Object 



44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
 
# File 'lib/diarize/audio.rb', line 44

def analyze!(train_speaker_models = true)
  parameter = fr.lium.spkDiarization.parameter.Parameter.new
  parameter.show = show
  # 12 MFCC + Energy
  # 1: static coefficients are present in the file
  # 1: energy coefficient is present in the file
  # 0: delta coefficients are not present in the file
  # 0: delta energy coefficient is not present in the file
  # 0: delta delta coefficients are not present in the file
  # 0: delta delta energy coefficient is not present in the file
  # 13: total size of a feature vector in the mfcc file
  # 0:0:0: no feature normalization
  parameter.parameterInputFeature.setFeaturesDescription('audio2sphinx,1:1:0:0:0:0,13,0:0:0:0')
  parameter.parameterDiarization.cEClustering = true # We use CE clustering by default
  parameter.parameterInputFeature.setFeatureMask(@path)
  @clusters = ester2(parameter)
  @segments = Segmentation.from_clusters(self, @clusters)
  train_speaker_gmms if train_speaker_models
end

#base_uri ⇒ Object 



109
110
111
112
113
114
 
# File 'lib/diarize/audio.rb', line 109

def base_uri
  # Remove the fragment if there is one
  base = uri.clone
  base.fragment = nil
  base
end

#clean! ⇒ Object 



64
65
66
67
 
# File 'lib/diarize/audio.rb', line 64

def clean!
  return if @uri.scheme == 'file' # Don't delete local file if initialised from local URI
  File.delete(@path)
end

#duration_by_speaker(speaker) ⇒ Object 



83
84
85
86
87
88
89
 
# File 'lib/diarize/audio.rb', line 83

def duration_by_speaker(speaker)
  return unless speaker
  segments = segments_by_speaker(speaker)
  duration = 0.0
  segments.each { |segment| duration += segment.duration }
  duration
end

#namespaces ⇒ Object 



97
98
99
 
# File 'lib/diarize/audio.rb', line 97

def namespaces
  super.merge 'ws' => 'http://wsarchive.prototype0.net/ontology/', 'mo' => 'http://purl.org/ontology/mo/'
end

#rdf_mapping ⇒ Object 



124
125
126
 
# File 'lib/diarize/audio.rb', line 124

def rdf_mapping
  { 'ws:segment' => segments }
end

#segments ⇒ Object

Raises:

  • (Exception) 


69
70
71
72
 
# File 'lib/diarize/audio.rb', line 69

def segments
  raise Exception.new('You need to run analyze! before being able to access the analysis results') unless @segments
  @segments
end

#segments_by_speaker(speaker) ⇒ Object 



79
80
81
 
# File 'lib/diarize/audio.rb', line 79

def segments_by_speaker(speaker)
  segments.select { |segment| segment.speaker == speaker }
end

#show ⇒ Object 



128
129
130
131
 
# File 'lib/diarize/audio.rb', line 128

def show
  # The LIUM show name will be the file name, without extension or directory
  File.expand_path(@path).split('/')[-1].split('.')[0]
end

#speakers ⇒ Object 



74
75
76
77
 
# File 'lib/diarize/audio.rb', line 74

def speakers
  return @speakers if @speakers
  @speakers = segments.map { |segment| segment.speaker }.uniq
end

#top_speakers ⇒ Object 



91
92
93
 
# File 'lib/diarize/audio.rb', line 91

def top_speakers
  speakers.sort {|s1, s2| duration_by_speaker(s1) <=> duration_by_speaker(s2)}.reverse
end

#type_uri ⇒ Object 



116
117
118
 
# File 'lib/diarize/audio.rb', line 116

def type_uri
  @type_uri || 'mo:AudioFile'
end

#type_uri=(type_uri) ⇒ Object 



120
121
122
 
# File 'lib/diarize/audio.rb', line 120

def type_uri=(type_uri)
  @type_uri = type_uri
end

#uri ⇒ Object 



101
102
103
 
# File 'lib/diarize/audio.rb', line 101

def uri
  @uri
end

#uri=(uri) ⇒ Object 



105
106
107
 
# File 'lib/diarize/audio.rb', line 105

def uri=(uri)
  @uri = uri
end