File size: 6,881 Bytes
12bfd03
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
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
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
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
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;                                                                       ;;
;;;                Centre for Speech Technology Research                  ;;
;;;                     University of Edinburgh, UK                       ;;
;;;                       Copyright (c) 1996,1997                         ;;
;;;                        All Rights Reserved.                           ;;
;;;                                                                       ;;
;;;  Permission is hereby granted, free of charge, to use and distribute  ;;
;;;  this software and its documentation without restriction, including   ;;
;;;  without limitation the rights to use, copy, modify, merge, publish,  ;;
;;;  distribute, sublicense, and/or sell copies of this work, and to      ;;
;;;  permit persons to whom this work is furnished to do so, subject to   ;;
;;;  the following conditions:                                            ;;
;;;   1. The code must retain the above copyright notice, this list of    ;;
;;;      conditions and the following disclaimer.                         ;;
;;;   2. Any modifications must be clearly marked as such.                ;;
;;;   3. Original authors' names are not deleted.                         ;;
;;;   4. The authors' names are not used to endorse or promote products   ;;
;;;      derived from this software without specific prior written        ;;
;;;      permission.                                                      ;;
;;;                                                                       ;;
;;;  THE UNIVERSITY OF EDINBURGH AND THE CONTRIBUTORS TO THIS WORK        ;;
;;;  DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING      ;;
;;;  ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT   ;;
;;;  SHALL THE UNIVERSITY OF EDINBURGH NOR THE CONTRIBUTORS BE LIABLE     ;;
;;;  FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES    ;;
;;;  WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN   ;;
;;;  AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,          ;;
;;;  ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF       ;;
;;;  THIS SOFTWARE.                                                       ;;
;;;                                                                       ;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;
;;;   Phrase boundary prediction.
;;;   
;;;   Two methods supported, if POS is enabled we use ngrams for that
;;;   otherwise we use a CART tree
;;;
;;;   Models trained from the IBM/Lancaster Spoken English Corpus and 
;;;   Boston University's FM Radio Corpus.

;;;
;;;  Here's a very simple CART tree for predicting phrase breaks
;;;  based on punctuation only
;;;
(set! simple_phrase_cart_tree
'
((lisp_token_end_punc in ("?" "." ":"))
  ((BB))
  ((lisp_token_end_punc in ("'" "\"" "," ";"))
   ((B))
   ((n.name is 0)  ;; end of utterance
    ((BB))
    ((NB))))))

(define (token_end_punc word)
  "(token_end_punc UTT WORD)
If punctuation at end of related Token and if WORD is last word
in Token return punc, otherwise 0."
  (if (item.relation.next word "Token")
      "0"
      (item.feat word "R:Token.parent.punc")))

;;;  This is a simple CART tree used after boundaries are predicted
;;;  by the probabilistic method to get two levels of break
(set! english_phrase_type_tree
'((pbreak is NB)
  ((num_break is 1)
   ((mB))
   ((R:Token.parent.EMPH is 1)
    ((NB))
    ((n.R:Token.parent.EMPH is 1)
     ((NB))
     ((NB)))))
  ((pbreak is BB)
   ((BB))
   ((pbreak is mB)
    ((mB))
    ((name in ("." "!" "?"));; only (potentially) change Bs to BBs
     ((BB))
     ((B)))))))

(set! f2b_phrase_cart_tree
'
((gpos is punc)
 (((1 0.00238095) (3 0) (4 0.997619) B))
 (((4 0.00238095) (3 0) (1 0.997619) NB))))

;;;  For more detailed prediction of phrase breaks we use POS and
;;;  probability distribution of breaks
;;;  These models were trained using data from the Lancaster/IBM
;;;  Spoken English Corpus

(require 'pos)   ;; for part of speech map

(defvar pbreak_ngram_dir libdir
  "pbreak_ngram_dir
  The directory containing the ngram models for predicting phrase
  breaks.  By default this is the standard library directory.")

(defvar english_phr_break_params
  (list
   ;; The name and filename off the ngram with the a priori ngram model
   ;; for predicting phrase breaks in the Phrasify module.  This model should 
   ;; predict probability distributions for B and NB given some context of 
   ;; part of  speech tags.
   (list 'pos_ngram_name 'english_break_pos_ngram)
   (list 'pos_ngram_filename
	 (path-append pbreak_ngram_dir "sec.ts20.quad.ngrambin"))
   ;; The name and filename of the ngram  containing the a posteriori ngram
   ;; for predicting phrase breaks in the Phrasify module.  This module should
   ;; predict probability distributions for B and NB given previous B and
   ;; NBs.
   (list 'break_ngram_name 'english_break_ngram)
   (list 'break_ngram_filename
	 (path-append pbreak_ngram_dir "sec.B.hept.ngrambin"))
   ;; A weighting factor for breaks in the break/non-break ngram.
   (list 'gram_scale_s 0.05)
   ;; When Phrase_Method is prob_models, this tree, if set is used to 
   ;; potentially predict phrase type.  At least some prob_models only
   ;; predict B or NB, this tree may be used to change some Bs into
   ;; BBs.  If it is nil, the pbreak value predicted by prob_models
   ;; remains the same.
   (list 'phrase_type_tree english_phrase_type_tree)
   ;; A list of tags used in identifying breaks.  Typically B and NB (and
   ;; BB).  This should be the alphabet of the ngram identified in
   ;; break_ngram_name
   (list 'break_tags '(B NB))
   (list 'pos_map english_pos_map_wp39_to_wp20)
   )
  "english_phr_break_params
Parameters for English phrase break statistical model.")

(defvar phr_break_params nil
  "phr_break_params
Parameters for phrase break statistical model.  This is typcal set by
a voice selection function to the parameters for a particular model.")

;;;
;;; Declaration of some features 
;;; 

(def_feature_docstring 
  'Word.pbreak
  "Word.pbreak
  Result from statistical phrasing module, may be B or NB denoting
  phrase break or non-phrase break after the word.")

(def_feature_docstring 
  'Word.pbreak_score
  "Word.pbreak_score
  Log likelihood score from statistical phrasing module, for pbreak
  value.")

(def_feature_docstring 
  'Word.blevel
  "Word.blevel
  A crude translation of phrase break into ToBI like phrase level.
  Values may be 0,1,2,3,4.")

(define (Phrasify utt)
"(Phrasify utt)                                
Construct phrasify over Words module."
  (let ((rval (apply_method 'Phrasify_Method utt)))
    (cond
     (rval rval) ;; new style
     (t
      (Classic_Phrasify utt)))))


(provide 'phrase)