pihsi 0.0.5 → 0.0.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 1a6adab4ac73877425b169cfba97535972b4b7cd
-  data.tar.gz: a4c59e6f444ab4847488184e916525a9f44833e2
+  metadata.gz: 59186bbdfc0d6a8209230283e39111f2dea70f67
+  data.tar.gz: 798220a80b22f1a512aa9685db6897f6145719b2
 SHA512:
-  metadata.gz: a48cfd2701d562e9cacf3b8234a7476210a112f48eb50ce7f78d5d55809b51bd2ae1b46bba2f2e06e5b22fdd1d20daa1490eb7477ad35ca89b84a31799f3de13
-  data.tar.gz: f12d78ca7636e6fd6f431e246bc5c22c5147040c05bd34817734765ef17e3f65e47b463669baacdfb2365dbc8d0708e91aadb46250606705b1a02a8c91b9aed6
+  metadata.gz: 94ffd1f8da9d88e6d230ec824e2c0e945061d4bbca532960806c6f8913e992ed37c40df0e26adcc82422b25455bc20c13312b558cd11d418efa8b441412a8a2b
+  data.tar.gz: 834d2901bdf0addffde57f43757197b19ab1f327ee315b76bc9d814407536b3d17f85fe03535b27543b87772f7d300f65018c1023755093d3dab0dc938c74c62

data/.travis.yml

@@ -5,8 +5,6 @@ rvm:
   - 1.9.3
   - rbx-2.2.7
 before_install:
-  - sudo apt-get update
-  - sudo apt-get install build-essential bison
   - wget http://downloads.sourceforge.net/project/cmusphinx/sphinxbase/0.8/sphinxbase-0.8.tar.gz
   - tar -zxf sphinxbase-0.8.tar.gz
   - cd sphinxbase-0.8
@@ -21,3 +19,4 @@ before_install:
   - make
   - sudo make install
   - cd ..
+  - sudo ldconfig

data/ext/pocket_sphinx/pocket_sphinx.c

@@ -1,33 +1,29 @@
 #include <ruby.h>
 #include <pocketsphinx.h>
 
-VALUE recognize(VALUE self, VALUE data) {
+typedef struct ps {
+	ps_decoder_t *decoder;
+} PocketSphinx;
+
+VALUE decode(VALUE self, VALUE data) {
 	char const *hyp, *uttid;
 	int rv;
 	int32 score;
 	ps_decoder_t *ps;
-	cmd_ln_t *config;
+	PocketSphinx *pocketSphinx;
+	VALUE result = rb_hash_new();
+
+	Data_Get_Struct(self, PocketSphinx, pocketSphinx);
+	ps = pocketSphinx -> decoder;
 
 	int data_length = RARRAY_LEN(data);
 	int16 *c_data = malloc(sizeof(int16) * data_length);
 
-	char *hmm = RSTRING_PTR(rb_funcall(self, rb_intern("hmm"), 0));
-	char *lm = RSTRING_PTR(rb_funcall(self, rb_intern("lm"), 0));
-	char *dict = RSTRING_PTR(rb_funcall(self, rb_intern("dict"), 0));
-
 	int i;
 	for(i = 0; i < data_length; i++) {
-		c_data[i] = NUM2SHORT(rb_ary_entry(data, i));
-	}
-
-	config = cmd_ln_init(NULL, ps_args(), TRUE, "-hmm", hmm, "-lm", lm, "-dict", dict, "-logfn", "/dev/null", NULL);
-
-	if (config == NULL) {
-		free(c_data);
-		rb_raise(rb_eStandardError, "configuration might be wrong");
+		c_data[i] = NUM2INT(rb_ary_entry(data, i)); // Rubinius does not support NUM2SHORT();
 	}
 
-	ps = ps_init(config);
 	rv = ps_start_utt(ps, "goforward");
 
 	if (rv < 0) {
@@ -50,12 +46,59 @@ VALUE recognize(VALUE self, VALUE data) {
 	if (hyp == NULL) {
 		return Qnil;
 	} else {
-		return rb_str_new2(hyp);
+		rb_hash_aset(result, rb_str_new2("score"), INT2NUM(hyp));
+		rb_hash_aset(result, rb_str_new2("hypothesis"), rb_str_new2(hyp));
+		return result;
 	}
 }
 
+static void deallocate(void *ps) {
+	PocketSphinx *pocketSphinx = ps;
+	ps_free(pocketSphinx -> decoder);
+	free(pocketSphinx);
+}
+
+static VALUE allocate(VALUE self) {
+	PocketSphinx *ps;
+	return Data_Make_Struct(self, PocketSphinx, 0, deallocate, ps);
+}
+
+VALUE initialize(VALUE self, VALUE options) {
+	int i;
+	PocketSphinx *ps;
+	VALUE option, key, value, klass;
+	char *c_key;
+	cmd_ln_t *config = cmd_ln_init(NULL, ps_args(), TRUE, "-logfn", "/dev/null", NULL);
+
+	for (i = 0; i < RARRAY_LEN(options); i++) {
+		option = rb_ary_entry(options, i);
+		key = rb_ary_entry(option, 0);
+		c_key = RSTRING_PTR(key);
+		value = rb_ary_entry(option, 1);
+		klass = rb_funcall(value, rb_intern("class"), 0);
+
+		if (klass == rb_cString) {
+			cmd_ln_set_str_r(config, c_key, RSTRING_PTR(value));
+		} else if (klass == rb_cFixnum) {
+			cmd_ln_set_int_r(config, c_key, NUM2LONG(value));
+		} else if (klass == rb_cFloat) {
+			cmd_ln_set_float_r(config, c_key, NUM2DBL(value));
+		}
+	}
+
+	if (config == NULL)
+		rb_raise(rb_eStandardError, "bad configuration");
+
+	Data_Get_Struct(self, PocketSphinx, ps);
+	ps -> decoder = ps_init(config);
+	return self;
+}
+
 void Init_pocket_sphinx() {
 	VALUE rb_mPihsi = rb_define_module("Pihsi");
 	VALUE rb_mPocketSphinx = rb_define_module_under(rb_mPihsi, "PocketSphinx");
-	rb_define_method(rb_mPocketSphinx, "recognize", recognize, 1);
+	VALUE rb_cDecoder = rb_define_class_under(rb_mPocketSphinx, "Decoder", rb_cObject);
+	rb_define_alloc_func(rb_cDecoder, allocate);
+	rb_define_method(rb_cDecoder, "initialize", initialize, 1);
+	rb_define_method(rb_cDecoder, "decode", decode, 1);
 }

data/lib/pihsi/speech_recognizer.rb

@@ -2,15 +2,20 @@ require "pocket_sphinx/pocket_sphinx"
 
 module Pihsi
   class SpeechRecognizer
-    include PocketSphinx
-    attr_reader :hmm, :lm, :dict
+    attr_reader :decoder
 
-    def initialize(hmm, lm, dict)
-      @hmm, @lm, @dict = hmm, lm, dict
+    def initialize(options = {})
+      _options = options.inject({}) do |result, (key, value)|
+        result["-#{key}"] = value unless value.nil?
+        result
+      end.to_a
+      @decoder = PocketSphinx::Decoder.new(_options)
     end
 
     def recognize(data)
-      super data.unpack('s*')
+      if result = decoder.decode(data.unpack('s*'))
+        result["hypothesis"]
+      end
     end
   end
 end

data/lib/pihsi/version.rb

@@ -1,3 +1,3 @@
 module Pihsi
-  VERSION = "0.0.5"
+  VERSION = "0.0.6"
 end

data/spec/factories/speech_recognizer.rb

@@ -3,7 +3,10 @@ FactoryGirl.define do
     hmm 'spec/fixtures/model/hmm/en_US/hub4wsj_sc_8k'
     lm 'spec/fixtures/model/lm/en/turtle.DMP'
     dict 'spec/fixtures/model/lm/en/turtle.dic'
+    pip 1.0
+    samprate 16000.0
+    maxhmmpf -1
 
-    initialize_with { new(hmm, lm, dict) }
+    initialize_with { new(attributes) }
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pihsi
 version: !ruby/object:Gem::Version
-  version: 0.0.5
+  version: 0.0.6
 platform: ruby
 authors:
 - Zhi-Qiang Lei
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-09-01 00:00:00.000000000 Z
+date: 2014-09-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler