Class: Waveform

Inherits:
Object
  • Object
show all
Defined in:
lib/waveform.rb,
lib/waveform.rb

Defined Under Namespace

Classes: ArgumentError, Log, RuntimeError

Constant Summary collapse

VERSION = 
"0.0.3"
DefaultOptions = 
{
  :method => :peak,
  :width => 1800,
  :height => 280,
  :background_color => "#666666",
  :color => "#00ccff",
  :force => false
}
TransparencyMask = 
"#00ff00"
TransparencyAlternate =

in case the mask is the background color!

"#ffff00"

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(source, log = nil) ⇒ Waveform

Setup a new Waveform for the given audio file. If given anything besides a WAV file it will attempt to first convert the file to a WAV using ffmpeg.

Optionally takes an IO stream to which it will print log/benchmarking info.

See #generate for how to generate the waveform image from the given audio file.

Available conversions depend on your installation of ffmpeg.

Example:

Waveform.new("mp3s/Kickstart My Heart.mp3")
Waveform.new("mp3s/Kickstart My Heart.mp3", $stdout)

Raises:



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
 
# File 'lib/waveform.rb', line 47

def initialize(source, log=nil)
  raise ArgumentError.new("No source audio filename given, must be an existing sound file.") unless source
  raise RuntimeError.new("Source audio file '#{source}' not found.") unless File.exist?(source)
  
  @log = Log.new(log)
  @log.start!
  
  # @source is the path to the given source file, whatever it may be
  @source = source
  
  # @audio is the path to the actual audio file we will process, always wav
  if File.extname(source) == ".wav"
    @audio = File.open(source, "rb")
  else
    # This happens in initialize so you can generate multiple waveforms from
    # the same audio without decoding multiple times
    # 
    # Note that we're leaving it up to the ruby/system GC to clean up these
    # tempfiles because someone may be generating multiple waveform images
    # from a single audio source so we can't explicitly unlink the tempfile.
    @audio = to_wav(source)
  end
  
  raise RuntimeError.new("Unable to decode source \'#{@source}\' to WAV. Do you have ffmpeg installed with an appropriate decoder for your source file?") unless @audio
end

Instance Attribute Details

#source ⇒ Object (readonly)

Returns the value of attribute source.



25
26
27
 
# File 'lib/waveform.rb', line 25

def source
  @source
end

Instance Method Details

#frames(width, method = :peak) ⇒ Object

Returns a sampling of frames from the given wave file using the given method the sample size is determined by the given pixel width – we want one sample frame per horizontal pixel.

Raises:



181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
 
# File 'lib/waveform.rb', line 181

def frames(width, method = :peak)
  raise ArgumentError.new("Unknown sampling method #{method}") unless [ :peak, :rms ].include?(method)
  
  frames = []
  
  RubyAudio::Sound.open(@audio.path) do |snd|
    frames_read       = 0
    frames_per_sample = (snd.info.frames.to_f / width.to_f).to_i
    sample            = RubyAudio::Buffer.new("float", frames_per_sample, snd.info.channels)

    @log.timed("Sampling #{frames_per_sample} frames per sample: ") do
      while(frames_read = snd.read(sample)) > 0
        frames << send(method, sample, snd.info.channels)
        @log.out(".")
      end
    end
  end

  frames
end

#generate(filename, options = {}) ⇒ Object

Generate a Waveform image at the given filename with the given options.

Available options are:

:method => The method used to read sample frames, available methods
  are peak and rms. peak is probably what you're used to seeing, it uses
  the maximum amplitude per sample to generate the waveform, so the
  waveform looks more dynamic. RMS gives a more fluid waveform and
  probably more accurately reflects what you hear, but isn't as
  pronounced (typically).

  Can be :rms or :peak
  Default is :peak.

:width => The width (in pixels) of the final waveform image.
  Default is 1800.

:height => The height (in pixels) of the final waveform image.
  Default is 280.

:background_color => Hex code of the background color of the generated
  waveform image.
  Default is #666666 (gray).

:color => Hex code of the color to draw the waveform, or can pass
  :transparent to render the waveform transparent (use w/ a solid
  color background to achieve a "cutout" effect).
  Default is #00ccff (cyan-ish).

:force => Force generation of waveform, overwriting WAV or PNG file.

Example:

waveform = Waveform.new("mp3s/Kickstart My Heart.mp3")

waveform.generate("waves/Kickstart My Heart.png")
waveform.generate("waves/Kickstart My Heart.png", :method => :rms)
waveform.generate("waves/Kickstart My Heart.png", :color => "#ff00ff")

Raises:



111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
 
# File 'lib/waveform.rb', line 111

def generate(filename, options={})
  raise ArgumentError.new("No destination filename given for waveform") unless filename

  if File.exists?(filename)
    if options[:force]
      @log.out("Output file #{filename} encountered. Removing.")
      File.unlink(filename)
    else
      raise RuntimeError.new("Destination file #{filename} exists. Use --force if you want to automatically remove it.")
    end
  end    

  options = DefaultOptions.merge(options)

  # Frames gives the amplitudes for each channel, for our waveform we're
  # saying the "visual" amplitude is the average of the amplitude across all
  # the channels. This might be a little weird w/ the "peak" method if the
  # frames are very wide (i.e. the image width is very small) -- I *think*
  # the larger the frames are, the more "peaky" the waveform should get,
  # perhaps to the point of inaccurately reflecting the actual sound.
  samples = frames(options[:width], options[:method]).collect do |frame|
    frame.inject(0.0) { |sum, peak| sum + peak } / frame.size
  end
  
  @log.timed("\nDrawing...") do
    background_color = options[:background_color] == :transparent ? ChunkyPNG::Color::TRANSPARENT : options[:background_color]
    
    if options[:color] == :transparent
      color = transparent = ChunkyPNG::Color.from_hex(
        # Have to do this little bit because it's possible the color we were
        # intending to use a transparency mask *is* the background color, and
        # then we'd end up wiping out the whole image.
        options[:background_color].downcase == TransparencyMask ? TransparencyAlternate : TransparencyMask
      )
    else
      color = ChunkyPNG::Color.from_hex(options[:color])
    end

    image = ChunkyPNG::Image.new(options[:width], options[:height], background_color)
    # Calling "zero" the middle of the waveform, like there's positive and
    # negative amplitude
    zero = options[:height] / 2.0
    
    samples.each_with_index do |sample, x|
      # Half the amplitude goes above zero, half below
      amplitude = sample * options[:height].to_f / 2.0
      # If you give ChunkyPNG floats for pixel positions all sorts of things
      # go haywire.
      image.line(x, (zero - amplitude).round, x, (zero + amplitude).round, color)
    end
    
    # Simple transparency masking, it just loops over every pixel and makes
    # ones which match the transparency mask color completely clear.
    if transparent
      (0..image.width - 1).each do |x|
        (0..image.height - 1).each do |y|
          image[x, y] = ChunkyPNG::Color.rgba(0, 0, 0, 0) if image[x, y] == transparent
        end
      end
    end
    
    image.save(filename)
  end

  @log.done!("Generated waveform '#{filename}'")
end