• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (C) 2008-2009 SVOX AG, Baslerstr. 30, 8048 Zuerich, Switzerland
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  *     http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 /**
17  * @file picopam.h
18  *
19  * Phonetic to Acoustic Mapping PU - Header file
20  *
21  * Copyright (C) 2008-2009 SVOX AG, Baslerstr. 30, 8048 Zuerich, Switzerland
22  * All rights reserved.
23  *
24  * History:
25  * - 2009-04-20 -- initial version
26  *
27  */
28 /**
29  * @addtogroup picopam
30 
31  * <b> Phonetic to acoustic mapping module </b>\n
32  *
33  * This module is responsible for mapping the phonetic domain features generated from text analysis
34  * into parametric representations suitable for signal generation.  As such it is the interface
35  * between text analysis and signal generation
36  *
37  * Most the processing of PAM is logically splittable as follows
38  * - building a suitable symbolic feature vector set for the sentence
39  * - Feeding Decision Trees with the symbolic seqence vector set
40  * - Colecting the parametric output of the Decision trees into suitable items to be sent to following PU's
41  *
42  * To perform the decision tree feeding and output collection the PU uses an internal buffer. This
43  * buffer is used several times with different meanings.
44  * - While building the symbolic feature vector set for the sentence this data structure stores syllable relevant data.
45  *   The corresponding phonetic data is stored outside as a single string of phonetic id's for all the sentence.
46  * - While feeding the decision trees the data structure is used to represent data for phonemes of the syllable
47  * - Addiotional data strucures are mantained to temporarily store items not pertaining to the PAM processing, for later resynchronization
48  *
49  * A quite detailed description of PAM processing is in the Know How section of the repository.
50  */
51 
52 #ifndef PICOPAM_H_
53 #define PICOPAM_H_
54 
55 #include "picodata.h"
56 #include "picokdt.h"
57 #include "picokpdf.h"
58 #include "picoktab.h"
59 #include "picokdbg.h"
60 
61 #ifdef __cplusplus
62 extern "C" {
63 #endif
64 #if 0
65 }
66 #endif
67 
68 /*------------------------------------------------------------------
69 Service routines
70 ------------------------------------------------------------------*/
71 picodata_ProcessingUnit picopam_newPamUnit(
72     picoos_MemoryManager mm,    picoos_Common common,
73     picodata_CharBuffer cbIn,   picodata_CharBuffer cbOut,
74     picorsrc_Voice voice);
75 
76 #ifdef __cplusplus
77 }
78 #endif
79 #endif /*PICOPAM_H_*/
80