pocketsphinx-ruby 0.3.0 → 0.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 1be9c28022d45172119190075e2c4f340f129022
-  data.tar.gz: f00d504978967e201a795874fcdee1038cb96d03
+  metadata.gz: 1299af5883e11346cca23dff74ba7015670384d0
+  data.tar.gz: 504770d54160fcf2bbcfc80e9910dd4283063e9a
 SHA512:
-  metadata.gz: 3407c5ce61d32c3576ce1fa6d16bc3249e0f74fc990b7b5dc8005c545e108cdca776e52e985354636a96ad626db2de90f71c5dd499ab29171b7b25a314cd5c26
-  data.tar.gz: b8be240f48b2962dd417c19589be7dba6a5b5460c9046deb66972507fe4655718884ad60d86ab8838c3428c946913e2e38ba5524bc1403f320ad01ee43a8e7f2
+  metadata.gz: 6dfaec5b4603b38344646fa77e41d941d70ac0455fb3932eecb8949f7e463d2f08c7f64d033df14067dd5e51e227f7b2aa180aa25eeec365e6603b55b03ea2fa
+  data.tar.gz: 0fa93e6bd7b2a37b3a0775c0ce045cb9666a13d97a0ef740798732268eea250e59d0cde41c3a1da7ce37afc0072c85917d31665defe7c9e46dec16be0dafc1f1

data/.travis.yml

@@ -1,11 +1,10 @@
 language: ruby
 rvm:
+  - 2.4.1
+  - 2.3.4
   - 2.2.2
   - 2.1.6
-  - 2.0.0
-  - 1.9.3
-  - rbx-2.2.9
-  - jruby-1.7.19
+  - jruby-9.1.9.0
 before_install:
   - sudo apt-get update -qq
   - sudo apt-get install -y swig

data/CHANGELOG.md

@@ -1,5 +1,10 @@
 # Changelog
 
+**v0.3.1 - 16/07/17**
+
+* SphinxBase API change [removes `err_set_debug_level`](https://github.com/cmusphinx/sphinxbase/commit/69c473ca648b8e2f8e453f27a405107a245bbcdd).
+
+
 **v0.3.0 - 17/04/15**
 
 * Add Words support to Decoder

data/README.md

@@ -224,3 +224,8 @@ An error like this probably means that you have an old version of the Pocketsphi
 3. Commit your changes (`git commit -am 'Add some feature'`)
 4. Push to the branch (`git push origin my-new-feature`)
 5. Create a new Pull Request
+
+### Projects Using pocketsphinx-ruby
+
+* [Isabella](https://github.com/chrisvfritz/isabella) - A voice-computing assistant built in Ruby.
+* [sphinxtrain-ruby](https://github.com/watsonbox/sphinxtrain-ruby) - A Toolkit for training/adapting CMU Sphinx acoustic models.

data/lib/pocketsphinx/api/pocketsphinx.rb

@@ -6,12 +6,16 @@ module Pocketsphinx
 
       typedef :pointer, :decoder
       typedef :pointer, :configuration
+      typedef :pointer, :logmath
 
       # Allows expect(API::Pocketsphinx).to receive(:ps_init) in JRuby specs
       def self.ps_init(*args)
         ps_init_private(*args)
       end
 
+      attach_function :ps_get_logmath, [:decoder], :logmath
+      attach_function :logmath_exp, [:logmath, :int], :double
+
       attach_function :ps_init_private, :ps_init, [:configuration], :decoder
       attach_function :ps_reinit, [:decoder, :configuration], :int
       attach_function :ps_default_search_args, [:pointer], :void
@@ -22,6 +26,7 @@ module Pocketsphinx
       attach_function :ps_end_utt, [:decoder], :int
       attach_function :ps_get_in_speech, [:decoder], :uint8
       attach_function :ps_get_hyp, [:decoder, :pointer], :string
+      attach_function :ps_get_prob, [:decoder], :int32
       attach_function :ps_set_jsgf_string, [:decoder, :string, :string], :int
       attach_function :ps_unset_search, [:decoder, :string], :int
       attach_function :ps_get_search, [:decoder], :string

data/lib/pocketsphinx/api/sphinxbase.rb

@@ -19,7 +19,6 @@ module Pocketsphinx
       attach_function :cmd_ln_set_int_r, [:pointer, :string, :int], :void
       attach_function :cmd_ln_str_r, [:pointer, :string], :string
       attach_function :cmd_ln_set_str_r, [:pointer, :string, :string], :void
-      attach_function :err_set_debug_level, [:int], :int
       attach_function :err_set_logfile, [:string], :int
       attach_function :err_set_logfp, [:pointer], :void
     end

data/lib/pocketsphinx/configuration/grammar.rb

@@ -12,7 +12,6 @@ module Pocketsphinx
       # Since JSGF strings are not supported in Pocketsphinx configuration (only files),
       # we use the post_init_decoder hook to configure the JSGF
       def post_init_decoder(decoder)
-        decoder.unset_search
         decoder.set_jsgf_string(grammar.raw)
         decoder.set_search
       end

data/lib/pocketsphinx/decoder.rb

@@ -6,9 +6,12 @@ module Pocketsphinx
 
     class Hypothesis < SimpleDelegator
       attr_accessor :path_score
+      attr_accessor :posterior_prob
 
-      def initialize(string, path_score)
+      def initialize(string, path_score, posterior_prob = nil)
         @path_score = path_score
+        @posterior_prob = posterior_prob
+
         super(string)
       end
     end
@@ -116,10 +119,12 @@ module Pocketsphinx
       mp_path_score = FFI::MemoryPointer.new(:int32, 1)
 
       hypothesis = ps_api.ps_get_hyp(ps_decoder, mp_path_score)
+      posterior_prob = ps_api.ps_get_prob(ps_decoder)
 
       hypothesis.nil? ? nil : Hypothesis.new(
         hypothesis,
-        mp_path_score.get_int32(0)
+        log_prob_to_linear(mp_path_score.get_int32(0)),
+        log_prob_to_linear(posterior_prob)
       )
     end
 
@@ -206,5 +211,11 @@ module Pocketsphinx
         configuration.post_init_decoder(self)
       end
     end
+
+    # Convert logarithmic probability to linear floating point
+    def log_prob_to_linear(log_prob)
+      logmath = ps_api.ps_get_logmath(ps_decoder)
+      ps_api.logmath_exp(logmath, log_prob)
+    end
   end
 end

data/lib/pocketsphinx/speech_recognizer.rb

@@ -108,9 +108,10 @@ module Pocketsphinx
     def recognize_continuous(max_samples, buffer)
       process_audio(buffer, max_samples).tap do
         if hypothesis = decoder.hypothesis
+          decoder.end_utterance
+
           yield hypothesis
 
-          decoder.end_utterance
           decoder.start_utterance
         end
       end
@@ -126,10 +127,12 @@ module Pocketsphinx
         end
 
         decoder.end_utterance
-        hypothesis = decoder.hypothesis
-        decoder.start_utterance
 
-        yield hypothesis if hypothesis
+        if hypothesis = decoder.hypothesis
+          yield hypothesis
+        end
+
+        decoder.start_utterance
       end
 
       process_audio(buffer, max_samples)

data/lib/pocketsphinx/version.rb

@@ -1,3 +1,3 @@
 module Pocketsphinx
-  VERSION = "0.3.0"
+  VERSION = "0.4.0"
 end

data/pocketsphinx-ruby.gemspec

@@ -22,6 +22,6 @@ Gem::Specification.new do |spec|
 
   spec.add_development_dependency "bundler", "~> 1.6"
   spec.add_development_dependency "rake"
-  spec.add_development_dependency "rspec", "~> 3.1.0"
+  spec.add_development_dependency "rspec", "~> 3.6.0"
   spec.add_development_dependency "coveralls"
 end

data/spec/configuration_spec.rb

@@ -65,7 +65,7 @@ describe Pocketsphinx::Configuration do
 
   describe '#setting_names' do
     it 'contains the names of all possible system settings' do
-      expect(subject.setting_names.count).to eq(114)
+      expect(subject.setting_names.count).to eq(113)
     end
   end
 
@@ -74,9 +74,9 @@ describe Pocketsphinx::Configuration do
       expect(subject.details 'vad_threshold').to eq({
         name: "vad_threshold",
         type: :float,
-        default: 2.0,
+        default: 3.0,
         required: false,
-        value: 2.0,
+        value: 3.0,
         info: "Threshold for decision between noise and silence frames. Log-ratio between signal level and noise level."
       })
     end
@@ -84,7 +84,7 @@ describe Pocketsphinx::Configuration do
     it 'gives details for all settings when no name is specified' do
       details = subject.details
 
-      expect(details.count).to eq(114)
+      expect(details.count).to eq(113)
       expect(details.first).to eq({
         name: "agc",
         type: :string,

data/spec/decoder_spec.rb

@@ -143,7 +143,7 @@ describe Pocketsphinx::Decoder do
   describe '#hypothesis' do
     it 'calls libpocketsphinx' do
       expect(ps_api)
-        .to receive(:ps_get_hyp) do |ps_decoder, mp_path_score|
+        .to receive(:ps_get_hyp).ordered do |ps_decoder, mp_path_score|
           expect(ps_decoder).to eq(subject.ps_decoder)
           expect(mp_path_score).to be_a(FFI::MemoryPointer)
 
@@ -152,10 +152,17 @@ describe Pocketsphinx::Decoder do
           "Hypothesis"
         end
 
+      expect(ps_api).to receive(:ps_get_prob).with(subject.ps_decoder).ordered.and_return(1)
+      expect(ps_api).to receive(:ps_get_logmath).with(subject.ps_decoder).ordered.and_return(:logmath)
+      expect(ps_api).to receive(:logmath_exp).with(:logmath, 20).ordered.and_return(0.5)
+      expect(ps_api).to receive(:ps_get_logmath).with(subject.ps_decoder).ordered.and_return(:logmath)
+      expect(ps_api).to receive(:logmath_exp).with(:logmath, 1).ordered.and_return(0.4)
+
       hypothesis = subject.hypothesis
 
       expect(hypothesis).to eq("Hypothesis")
-      expect(hypothesis.path_score).to eq(20)
+      expect(hypothesis.path_score).to eq(0.5)
+      expect(hypothesis.posterior_prob).to eq(0.4)
     end
   end
 

data/spec/integration/decoder_spec.rb

@@ -11,14 +11,17 @@ describe Pocketsphinx::Decoder do
   end
 
   it 'reads cmninit configuration values from default acoustic model feat.params' do
-    expect(configuration.details('cmninit')[:default]).to eq("8.0")
-    expect(configuration.details('cmninit')[:value]).to eq("40,3,-1")
+    expect(configuration.details('cmninit')[:default]).to eq("40,3,-1")
+    expect(configuration.details('cmninit')[:value]).to eq("41.00,-5.29,-0.12,5.09,2.48,-4.07,-1.37,-1.78,-5.08,-2.05,-6.45,-1.42,1.17")
   end
 
   describe '#decode' do
     it 'correctly decodes the speech in goforward.raw' do
       subject.decode File.open('spec/assets/audio/goforward.raw', 'rb')
       expect(subject.hypothesis).to eq("go forward ten meters")
+
+      expect(subject.hypothesis.path_score).to eq(0.49327822492667395)
+      expect(subject.hypothesis.posterior_prob).to eq(0.037719430745321664)
     end
 
     # FIXME: This test illustrates a current issue discussed in:
@@ -40,9 +43,9 @@ describe Pocketsphinx::Decoder do
     it 'reports words with start/end frame values' do
       subject.decode File.open('spec/assets/audio/goforward.raw', 'rb')
 
-      expect(subject.words.map(&:word)).to eq(["<s>", "go", "forward", "ten", "meters", "</s>"])
-      expect(subject.words.map(&:start_frame)).to eq([0, 46, 64, 117, 153, 212])
-      expect(subject.words.map(&:end_frame)).to eq([45, 63, 116, 152, 211, 260])
+      expect(subject.words.map(&:word)).to eq(["<s>", "<sil>", "go", "forward", "ten", "meters", "</s>"])
+      expect(subject.words.map(&:start_frame)).to eq([0, 25, 46, 64, 117, 153, 212])
+      expect(subject.words.map(&:end_frame)).to eq([24, 45, 63, 116, 152, 211, 260])
     end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pocketsphinx-ruby
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 0.4.0
 platform: ruby
 authors:
 - Howard Wilson
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-04-17 00:00:00.000000000 Z
+date: 2017-07-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -58,14 +58,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 3.1.0
+        version: 3.6.0
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: 3.1.0
+        version: 3.6.0
 - !ruby/object:Gem::Dependency
   name: coveralls
   requirement: !ruby/object:Gem::Requirement
@@ -157,7 +157,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.2.2
+rubygems_version: 2.5.1
 signing_key: 
 specification_version: 4
 summary: Ruby speech recognition with Pocketsphinx