WO1999014689A1 - System and method for creating a language grammar - Google Patents
System and method for creating a language grammar Download PDFInfo
- Publication number
- WO1999014689A1 WO1999014689A1 PCT/US1998/019432 US9819432W WO9914689A1 WO 1999014689 A1 WO1999014689 A1 WO 1999014689A1 US 9819432 W US9819432 W US 9819432W WO 9914689 A1 WO9914689 A1 WO 9914689A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- cell
- utterance
- contents
- valid
- main table
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/253—Grammatical analysis; Style critique
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
Definitions
- This invention relates generally to computerized natural language systems. More particularly, it relates to a computer system and method for creating a grammar and automatically generating a corpus of valid utterances in the grammar, as well as a file containing the set of rules that define the grammar. It also relates to including token information in such a corpus to represent the meaning of each utterance to a speech-enabled system using the grammar.
- the computer needs an automatic speech recognition system to detect the spoken words and convert them into some form of computer- readable data, such as simple text.
- the computer needs some way to analyze the computer-readable data and determine what those words, as they were used, meant.
- ASR- 1500 manufactured by Lernout & Hauspie
- Watson 2.0 manufactured by AT&T
- Nuance 5.0 by Nuance are just a few examples of effective continuous speech, speaker independent speech recognizers.
- a speech recognizer is "continuous” if it does not require the speaker to pause between words.
- a recognizer is "speaker independent” if it does not need to have heard the speaker's voice before in order to understand the speaker's words. For example, some speech recognizers must first learn the user's voice before it can understand the words. Learning the user's voice usually means having the recognizer record and memorize the user's voice as s/he recites a list of key phonetic words or sounds.
- Recognizers such as the ASR-1500 and Watson 2.0 listed above, typically require advance notice of the grammar they will be asked to listen for. This advance notice comes in the form of a vendor-specific ASR grammar file that describes the rules and contents of the grammar. However, there is no easy way to create this data file. The manufacturers of the speech recognizers listed above will provide the user with the format in which the data file must be written, but it is up to the user to actually write the data file.
- annotated corpus for the grammar.
- the annotated corpus is essentially a listing of valid utterances (statements) in the grammar, as well as a token data value that represents the meaning for each valid utterance.
- Such a corpus could contain millions of entries, and what is needed is a simplified method for generating this annotated corpus.
- the present invention satisfies both of these needs.
- a general purpose of the present invention is to provide a toolkit for generating a context-free grammar.
- Another object of the present invention is to provide a toolkit for simplifying the task of generating a context-free grammar and an annotated corpus of utterances in the grammar.
- a further object of the present invention is to provide a system and method for creating a corpus supported by a grammar that includes token data representing the meaning of each item in the corpus.
- the present invention provides interactive voice recognition system developers with a simple and intuitive way to generate grammar corpora, or the list of words and phrases (utterances) that make up a language.
- a table comprised of columns and rows of cells is initially used to define the grammar.
- a valid utterance begins with a cell in the leftmost column, ends in a column to the right, and must include the contents of one cell from each column in between.
- This left-to-right progression traces a path through the table, where the path itself is the valid utterance. Since it is possible for valid utterances with completely different meanings to begin with the same cell in the leftmost column, heavy lines within columns are used to form barriers which the path cannot cross when the next cell in the path is considered. Thus, the path of an utterance having one meaning cannot continue with a cell in the path of an utterance having a completely different meaning.
- the path of a valid utterance ends when the last word in the utterance is reached.
- the next non-blank cell in the same row as the last word of the utterance will contain token data. This token data will represent, to the speech- enabled application being developed, the meaning of the valid utterance.
- Cells can also contain other types of data. Some cells may contain references to other tables. Other cells may contain variables, which are phrases (such as numbers, dates, times and dollar amounts) that return a separate data value to the speech-enabled application. Cells may contain directives, or commands, which are understood by the speech recognizer. Cells may be identified as optional, where utterances are valid with or without the contents of the optional cell.
- the grammar When the developer is finished entering data in the table, the grammar has been defined.
- the table is then traversed to generate an enumerated list (corpus) of valid utterances in the grammar, where each utterance in the corpus also includes a token.
- the table is also processed a second time to generate a grammar file which will serve as input to the automatic speech recognition
- the corpus and grammar file are then used, in the preferred embodiment, in the interactive voice recognition system (IVR) , which is the system that is actually conversing with the speaker.
- IVR interactive voice recognition system
- Valid utterances heard by the speech recognizer are reported to the IVR, which then passes the valid utterances to the runtime interpreter (subject of copending application named above) , which compares the valid utterances detected by the speech recognizer with the corpus of valid utterances in the grammar. When a match is found, the associated token is retrieved and returned to the IVR.
- the IVR will understand, once given the token, what the detected valid utterance meant and can react accordingly.
- Figure 1 shows an overview of an embedded natural language understanding system.
- Figure 2 shows the grammar development toolkit main screen.
- Figure 3 shows a tree representation of the grammar table in figure 2.
- Figure 4 shows the various cell dialogs.
- Figure 5 is a table showing the variable types supported in the preferred embodiment.
- Figure 6 is a flow diagram showing the use of the grammar development toolkit.
- Figure 7 shows the Annotated Corpus Dialog.
- Figures 8a and 8b show sample formats for the annotated ASR corpus file and vendor-specific ASR grammar file, respectively.
- Figure 9 is a flow diagram showing the operation of the IVR as it accesses the runtime interpreter.
- Figures lOa-d are flow diagrams showing the steps taken by the toolkit to extract grammar rules from a table .
- ASR corpus file data file containing a listing of valid utterances in a grammar, as well as token data for each valid utterance which represents the meaning of the valid utterance to the interactive voice recognition system (IVR 130) .
- ASR automatic speech recognition
- cells - discrete elements within the table (the table is made up of rows and columns of cells) . In the example rule given with the definition of "rule' below, each of "I want", “I need” and “food” would be placed in a cell.
- the cells containing "I want” and “I need” are vertically adjacent to one another (same column) . Vertically adjacent cells are generally OR'd together. The cell containing "food”, however, would occur in the column to the right of the "I want” and “I need” column, indicating the fact that "food” must follow either "I want” or "I need” and as such, the cell containing "food” will be AND'd to follow the cells containing "I want” and "I need”.
- constrained grammar - a grammar that does not include each and every possible statement in the speaker' s language; limits the range of acceptable statements. corpus - a large list. grammar - the entire language that is to be understood.
- Grammars can be expressed using a set of rules, or by listing each and every statement that is allowed within the grammar.
- grammar development toolkit (104) software used to create a grammar and the set of rules representing the grammar.
- natural language understanding - identifying the meaning behind spoken statements that are spoken in a normal manner.
- phrase the "building blocks" of the grammar, a phrase is a word, group of words, or variable that occupies an entire cell within the table.
- rules these define the logic of the grammar.
- An example rule is: ("I want”
- runtime interpreter (124) software that searches through the annotated corpus (122) whenever a valid utterance is heard, and returns a token representing the meaning of the valid utterance.
- runtime interpreter application program interface (RIAPI) set of software functions that serve as the interface through which the interactive voice response system (130) uses the runtime interpreter.
- speech recognizer (116) - combination of hardware and software that is capable of detecting and identifying spoken words.
- speech recognizer compiler (114) - software included with a speech recognizer (116) that accepts, as input, a vendor-specific ASR grammar file (112) and processes the file (112) for use in a speech recognizer (116) during runtime.
- table two dimensional grid used to represent a grammar.
- token - each valid utterance in the table is followed by a cell that contains a token, where the token is a unique data value (created by the developer when s/he develops the grammar) that will represent the meaning of that valid utterance to the interactive voice response system (130) .
- utterance - a statement. utterance, spoken - an utterance that was said aloud. The spoken utterance might also be a valid utterance, if the spoken utterance follows the rules of the grammar. utterance, valid - an utterance that is found within the grammar. A valid utterance follows the rules which define the grammar. variable - "place holder" used in the corpus
- the runtime interpreter (124) is able to identify this variable in the corpus, and performs additional processing during runtime to interpret the number.
- vendor-specific ASR grammar file (112) - a data file that contains the set of rules representing a grammar, and is written in a format that will be recognized by the speech recognizer compiler (114).
- FIG. 1 The basic elements comprising: an interactive voice response system (130), or IVR; the grammar development toolkit (104); a compiler (114) and speech recognizer (116), which are part of an automatic speech recognition (ASR) system (118) ; an annotated ASR corpus file (122); a vendor-specific ASR grammar file (112); the runtime interpreter (124); and the custom processor interface (126), or CP; and the runtime interpreter application program interface (128), or RIAPI.
- the grammar development toolkit (104) is the focus of this application and is discussed in detail further below. The other elements listed above are discussed in detail in the copending application cited above. However, a general overview of the speech-enabled system will be helpful for a full understanding of the operation and purpose of the toolkit (104) .
- the first step in implementing a natural language system is creating the set of rules that govern the valid utterances in the grammar.
- a grammar for the reply to the question: "what do you want for lunch?" might be represented as:
- the grammar is generated by a developer using the grammar development toolkit (104).
- the toolkit (104) is a novel spreadsheet-oriented software package that provides the developer of a natural language application with a simplified way of generating a grammar.
- the toolkit (104) resides on a computer that contains a central processing unit (102) , application specific software (106), memory files (108) and input device such as keyboard (110) .
- the toolkit (104) When the developer has completed the grammar using the toolkit (104) , two outputs are generated by the toolkit (104) for use in the natural language system.
- the first such output is a vendor-specific ASR grammar file (112), which is saved in a format that will be recognizable by the speech recognizer (116) .
- Speech recognizer (116) is a continuous speech, speaker independent speech recognizer.
- Commercially available speech recognizers (116) include the ASR-1500, manufactured by Lernout & Hauspie; Watson 2.0, manufactured by AT&T; and Nuance 5.0, by Nuance.
- the preferred embodiment of the toolkit (104) is able to generate grammar files for any of these recognizers.
- the vendor-specific ASR grammar file (112) contains information regarding the words and phrases that the speech recognizer (116) will be required to recognize, written in a form that is compatible with the recognizer.
- the file is also optimized to take advantage of peculiarities relating to the chosen speech recognizer (116) . For example, experience with the L&H recognizers has shown that L&H grammars perform well if the grammar avoids having multiple rules with the same beginning
- the speech recognizer will need to compile the vendor-specific ASR grammar file (112) using a compiler tool (114) supplied by the recognizer vendor.
- the preferred embodiment of the toolkit (104) knows, when the grammar is first generated, which speech recognizer (116) will be used and is able to format the vendor-specific ASR grammar file (112) accordingly.
- the second output from the toolkit (104) is an annotated ASR corpus (122), which is actually a pair of flat files.
- the first of the pair is a corpus file, and contains a listing of all possible logical sentences or phrases in the grammar (with the exception of variables, discussed below), the compartments (groups of tables) in which they appear, and a value representing the class of the utterance (sentence) heard.
- the second is an answers file that maps each utterance class with a token, or data value that represents the meaning of the utterance.
- the recognizer (116) identifies the words and phrases it hears and notifies the IVR (130) when a valid utterance has been heard.
- the IVR (130) is the system which needs the speech understanding capabilities, and includes the necessary external connections and hardware to function (for example, a banking IVR - 130 might include a connection to the bank database, a keypad for entering data, a visual display for displaying information, a dispenser for dispensing money, and a speaker for speaking back to the user) .
- This valid utterance is passed, in a computer-readable form such as text, to the IVR (130) which then notifies the runtime interpreter (124) of the utterance that was heard.
- the runtime interpreter (124) consults the annotated ASR corpus (122) and returns an appropriate token to the IVR (130) for the valid sentence heard by the recognizer (116) .
- This token represents the meaning of the utterance that was heard by the recognizer (116), and the IVR (130) is then able to properly respond to the utterance.
- the CP (126) and RIAPI (128) serve as software interfaces through which the IVR (130) may access the runtime interpreter (124) . It is the IVR (130) that ultimately uses the speech capabilities to interact with the speaker during runtime.
- the toolkit is developed using "Visual Basic” (trademark Microsoft Corp.) operating on a “Pentium”-based (trademark Intel Corp.) computer system.
- the toolkit is designed to run on a "Windows NT” (trademark Microsoft Corporation) .
- Windows NT trademark Microsoft Corporation
- the toolkit is a software tool which enables a developer to visualize and create optimized ASR grammars for specified speech recognizers that use a BNF- based language model for recognizing speech.
- One novel aspect of the toolkit is the spreadsheet format used to visualize the grammar being created.
- Figure 2 shows a typical main toolkit screen.
- the toolkit main screen displays a table (206) that is within a project.
- a project is defined as the set of compartments and tables which form the basis of a particular speech application. Compartments contain the one (or more) table (s) related to a particular task within a project.
- a table essentially contains the tree for valid utterances. Thus, a typical project would be “banking”. The "banking" project would contain compartments for the various tasks related to banking such as "account inquiry” and "loan inquiry”. Tables within the "account inquiry” compartment would then define the valid utterances which pertain to the topic of an account inquiry.
- a table (206) is a matrix where the developer enters data which will be used to generate the grammar, and is composed of rows and columns of cells.
- Each cell can contain data of one of the following types: Terminal symbols, Non-terminal symbols, Variables and ASR Directives.
- Terminal symbols are any ASCII string (with the exception of special characters that have specific meaning to either the runtime interpreter 124 or the speech recognizer 116) , and they are the basic words and phrases of the language written in text form. Special characters in the preferred embodiment include '!', V*' and x #' for the L&H recognizer, and parentheses, brackets, braces and carets in general.
- Non-terminal symbols serve as a cross-reference to other tables.
- "LOAN" in row 1, col. 3 and row 7, col. 3 of main table 206 is a non-terminal which references the sub-table 205a.
- Non-terminal symbols will be replaced with the referenced table.
- "LOAN” will be replaced with the contents of sub-table 205a.
- Variables are a type of Non-terminal symbol that, in addition to referencing another table, return a value to the controlling software (i.e., IVR) during runtime.
- row 4 in main table 206 is a cell containing the variable "YEARS" which references sub-table 205b.
- the table referenced by a variable defines the grammar for the variable and will be processed as any other referenced table when the vendor-specific ASR grammar (112) is created.
- variables are not completely enumerated when the corpus file (122) is generated. This is because there would be too many possibilities to list if efficiency is to be maintained. For example, the valid utterance "My favorite number from 1 to a thousand is xxx" results in a thousand different possibilities.
- utterances that include variables are written to the corpus file (122) with special characters "holding the place” of where variables will occur in the utterance.
- our example utterance would have a single corpus file (122) entry: "My favorite number from 1 to a thousand is [INTEGER1]".
- the runtime interpreter (124) searches the corpus file (122) for a detected utterance that includes variables, only the non- variable portion of the detected utterance must match.
- the variable portion will be interpreted separately using algorithms located in the runtime interpreter (124) (in the example, the algorithm for INTEGER1) and resulting values will be stored in system memory for the IVR (130) to retrieve.
- the variable types which are supported with such algorithms are shown in figure 5.
- variables are set off from the regular text with square brackets ( * [ ⁇ and ⁇ ]').
- brackets * [ ⁇ and ⁇ ]'.
- the characters between the brackets identify the variable type and the algorithm which will be processed at runtime when the variable is heard.
- ASR Directives are special commands meant for the speech recognizer (116) . These cells will appear in the grammar file (112), but not in the corpus file (122). For example, ASR Directives are used in the grammar file (112) to inform the recognizer that there is a variable in the utterance and another table should be referenced. Cells can also be Marked, Underlined, or characterized as Optional. These features will be explained during the discussion of table traversal below.
- All valid utterances in the grammar represented by the table can be obtained by traversing the table from left to right.
- a valid utterance begins when one of the phrases in the first (leftmost) column is heard. When this first phrase is heard, the range of row numbers for the phrases that are allowed to follow is determined by going up from the first phrase until a barrier line is encountered that also appears in the next column to the right, and going down until a barrier line is encountered which also appears in the next column to the right. Barrier lines are created by underlining cells. These barrier lines form a barrier in the table which is not to be crossed during the enumeration traversal below. The next phrase must be found (if the sentence is to be valid) in the first column to the right of the current column that contains entries within this range of rows. This process is continued until the next non-blank column is a marked column. Each compartment has a main table, and the rightmost column of this main table is the marked column which contains the token to be returned to the IVR (130) for each valid utterance.
- the range of row values for the next valid phrase is 1 (got to top of column without encountering a barrier line) to 6 (can't cross barrier line under row 7 because it also appears in column 2) .
- the next column that contains entries within this range of rows (skipping any blank ones) is column 2, so the next valid phrase must be found in column 2 between rows 1 and 6. If the phrase "I want" was heard next, it would be valid because this phrase exists in row 2 (between 1 and 6) of column 2. Once "I want” is heard, the process of determining the next valid range begins again, but now starts from the cell containing "I want". Going up and down as before, the next valid range of rows also goes from row 1 to 6, but the next column with entries in this range is column
- a non-terminal cell simply identifies another table which should appear in place of the non-terminal cell.
- the referenced table is traversed in the same manner, and any of the utterances from this "sub-table" may be placed in the non- terminal cell within the current utterance of the main traversal. Functionally, it is as if the referenced table replaced the non-terminal cell in the current table.
- Cells may also be characterized as Optional. Each utterance that includes an optional cell will be valid with or without the contents of that cell.
- the optional cells in the utterance result in multiple corpus items (utterances) to account for the various possibilities.
- ASR Directives are used in the vendor-specific grammar file (112) to inform the speech recognizer (116) that a particular cell is optional.
- traversal discussed above demonstrates how a table can be used to check the validity of a given utterance.
- Another type of traversal is actually performed by the toolkit (104) software. This second type of traversal is used to enumerate all possible valid utterances from the table (with a few exceptions, such as variables), or in other words, the enumeration traversal seeks to "see what's at the end of every valid path through the table" and record the valid utterance formed by each path.
- the software when the enumeration traversal takes place, the software logically travels completely down one path until it ends (in a
- an enumeration traversal might start with the valid utterance "I think I need a mortgage with a twenty year term" (where ⁇ twenty' would be a YEAR variable) .
- the next valid utterance after "backing up” from the end of the first valid utterance until a branch, would be "I think I need a mortgage that has a twenty year term”.
- the branch occurred in column 4, where "with a” and “that has a” both would have resulted in valid utterances.
- This is merely the type of enumeration traversal performed in the preferred embodiment of the invention, and it is understood that various methods of traversal would also result in a full enumeration.
- the list of valid utterances in the corpus (122) need not be in any particular order, so any enumeration traversal method would work.
- a tool (202) is displayed across the top of the toolkit main screen in figure 2. This tool allows the user to manage compartments and projects. Under the "File' menu, the user can create, open or close projects, NOT FURNISHED UPON FILING
- Figure 4a shows the Non- terminal specifications dialog (400) .
- a Non-terminal cell type references another table.
- This dialog (400) simply prompts the developer for the name of the referenced table.
- Figure 4b shows the ASR Directives Dialog (410) , which provides the user with a listbox of the available ASR Directives for the current speech recognizer (116) . If the selected ASR Directive requires additional argument data, the developer can provide this additional argument data in the dialog.
- Figure 4c shows the Variable Specification Dialog (420), which is displayed when the developer designates a variable cell.
- the dialog (420) prompts the developer for the name of the variable, the table that is referenced by the variable, and the variable type.
- a list of variable types supported by the preferred embodiment, as well as the value type to be returned by the variable, is found in figure 5.
- Figure 6 shows the steps a developer will take when using the toolkit (104) to create a grammar.
- the developer begins in step 600 by opening the toolkit (104) program. In a typical windowing environment, this will involve selecting an icon. While the preferred embodiment of the toolkit (104) is as an OLE server for inclusion in another vendor-developed toolkit, the toolkit could also be made in a standalone version.
- the developer will need to open a new project to work on in step 602.
- the developer then creates a compartment in step 604, and defines the number of columns and rows for the compartment's main table in step 606. This can be done using the sizing buttons (204b) .
- the next step (608) is to fill in the cells that are to contain terminal data.
- terminal data is simply text data that defines the non-variable phrases ("I'd like”, "I want”, etc.) which are valid in the grammar.
- step 610 the developer fills in the variable cells (if any) of the table.
- the variables dialog shown in figure 4c is used to define the variable, which includes the variable name and a reference to the algorithm which will be performed when the variable is processed (the actual algorithm is located within the runtime interpreter 124) .
- step 612 the developer fills in the cells (if any) which are to contain ASR directives.
- the dialog in figure 4b is used to select the ASR directive from a list of supported ASR directives for the current speech recognizer (116) (the current speech recognizer is chosen from a list in the "File' menu of the tool - 202) .
- step 614 the developer identifies which cells (if any) are to be optional by using the right mouse button and the "Make Optional" option of the menu shown in figure 2b.
- step 616 the developer identifies and fills in the column in the compartment main table that contains the token data.
- the rightmost column of the compartment main table is marked by default as having token data.
- step 618 the developer identifies the barriers used during the traversal discussed above. In the preferred embodiment, this is done by underlining certain cells using the menu shown in figure 2b.
- step 620 the developer fills in the nonterminal cells. In the preferred embodiment, this is done using the dialog shown in figure 4a. The developer will also need to create a new table to contain the subgrammar, which can be done using the "table' menu. This table will also need the terminal, non-terminal, variable, optional and ASR directive cell steps discussed above if the referenced table requires any of those cell types. However, in the preferred embodiment, referenced tables do not have marked columns and do not return tokens.
- step 622 the compartment is complete. If another compartment is needed for the project, the developer proceeds to step 604 and creates the new compartment. If no other compartments are necessary, the developer can proceed to either of two steps, 624 and 626.
- the grammar file (112) is generated for the ASR (114, 116). In the preferred embodiment, this process is initiated using the "File' menu option in the compartment window (203) . This generation is done by performing an analysis of the table, and results in a file (112) stored in memory.
- This file (112) contains the grammar rules embodied in the tables and is in a form that can be compiled and used by the speech recognizer (116) .
- the analysis performed during generation of the grammar file (112) seeks to generate a set of BNF rules which are followed by valid utterances in the grammar.
- the generation of the BNF rules begins with a table that has no non-terminals. All non-terminals are replaced with the tables that they reference, resulting in a complete, expanded main table.
- HandleRegion itself receives a table, or portion of a table, as input.
- any non-terminals originally found in the tables given to HandleRegion are replaced with the sub- tables that they reference before HandleRegion is called.
- Tables are represented by a two-dimensional array, and defined by their corners (i.e. rowl,coll - row4,col4).
- HandleRegion assembles the grammar rules defined by the table by recursively dividing the table into smaller and smaller tables ("regions") until "simple" regions are formed.
- a simple region is a region of cells in which there are no heavy lines (barriers) .
- Table la depicts a simple region.
- HandleRegion is begun (step 1002), a check is made to determine whether the region to be handled is a simple one. If a rule is to be extracted from table la, then this check succeeds.
- the region is a simple one, vertically adjacent cells are grouped using logical "OR' (step 1006) . In table la, four groups would be formed, one for each column. The groups would be, (AIEIIIM), (BIFUIN), (C
- step 1012 the region to be handled is checked to see if it is a "prefixed" region.
- a prefixed region has no barrier lines in the first "n' columns (these are simple columns) , and thus, table lb is a prefixed region because it has no barrier lines in its first 1 column.
- step 1014 the prefixed region is divided into two sub-regions. The first sub-region is the leftmost n columns (contains A, E, I and M) , and the second sub-region is the rest of the prefixed region.
- step 1016 the rule for the prefixed region is defined as the rule for the first region AND the rule for the second region.
- HandleRegion is then recursively called once to find the rule for the first region, and again to find the rule for the second region.
- this rule is returned and HandleRegion ends in step 1072.
- step 1012 If, in step 1012, the region is not a prefixed region, it is checked (step 1020) to see if it is a
- postfixed region is the reverse of a prefixed region, and has no barrier lines in the last "n' columns. If the region is a postfix region, it is divided in step 1022 into two sub-regions, one for the rightmost n columns, and another for the rest. Step 1024 defines the rule for the region as the logical AND of the two sub-regions, much like step 1016 for the prefixed region. Again, the rule is returned (step 1026) and HandleRegion ends in step 1072.
- step 1020 If, in step 1020, the region is not a postfixed region, it is checked (in step 1028) to see if there are any "overlapping" barriers in the region. Overlapping barriers occur in columns which have more than one barrier, and table lc contains an overlapped column
- step 1030 the process starts with the top, left cell in the region to be handled. The contents of this cell ( "A' ) are added to the current simple region being defined.
- step 1032 the process attempts to go down the current column until a barrier line is encountered.
- Cells that the process enters are added to the current simple region.
- table lc the process can go down until cell "I', and each cell that it enters is added to the current simple region (so far, "A' , "E' and “I' are in the current simple region) .
- a "ceiling” and “ground” are noted.
- the “ceiling” is the first barrier or border in the current column above the current cell (border above "A' )
- the "ground” is the first barrier or border in the current column that is below the current cell (barrier under "I' in row 3) .
- step 1034 the process attempts to go right until either the "ceiling" or “ground” of the next cell to the right is different.
- the process is at cell “I' . It then attempts to go right, to cell V .
- the "ground” under “J' is the same (barrier still under row 3), but the "ceiling” is different.
- the first barrier or border above “J' is NOT the border above "A' (which was the case for "I' ) , but the barrier under "F' .
- step 1042 the overlapping process checks to see if all cells have been assigned to a simple region. If not, the process goes to step 1044, which chooses the next cell to be assigned to a simple region. This selection is done from left to right across the current row, and when a row is done, the first cell of the next row is chosen. In our example, the first simple region began with the top, left cell ( "A' ) . The next cell to the right that isn't already part of a simple region is "B' , so "B' is the next starting point.
- each path is converted to a rule by discarding the BEGIN and END, substituting logical "AND' for the "->” and replacing the region numbers with the simple region rules generated in step 1046.
- These rules, one for each path, are then combined using logical "OR' to form the one rule for the overlapping region. This one rule is returned in step 1056, and the HandleRegion process ends in step 1072.
- step 1058 the HandleRegion process moves to step 1058, where the longest barrier line in the region to be handled is determined. If there is a tie, one is arbitrarily chosen. In step 1060, this line is checked to see if it extends from one edge of the region to the other. If it does not, the region to be handled is split vertically into two horizontally adjacent sub-regions. This split is done at an arbitrary edge of the longest barrier line. In the preferred embodiment, the left edge is used, and if the left edge of the longest line is the left border of the region, the right edge is used.
- the barrier under "A' and "B' is arbitrarily chosen over the one under "0' and "P' . Furthermore, the split occurs on the right edge of this barrier, forming two sub-regions. The first sub-region includes columns 1 and 2, while the second sub-region contains columns 3 and 4.
- the HandleRegion process is called once for each sub-region, and the returned rules for the sub- regions are combined using a logical "AND' .
- step 1066 simply divides the region into two sub-regions based on the longest line. HandleRegion is then called once for each sub-region, and these regions are combined, in step 1068, using a logical "OR' . Step 1070 returns whatever rule was created (either step 1064 or 1068), and 1072 ends the HandleRegion process .
- the grammar rules represented by the tables may be modified to accommodate peculiarities of the chosen speech recognizer (116) and enhance performance. For example, combining multiple rules that begin or end with the same words into a single rule increases efficiency for some recognizers. If two rules begin with the same words (i.e. AB and AC), then the two rules can be combined into a single rule with an additional logical "OR' (i.e. A(B
- step 626 which can occur before or after step 624, the annotated ASR corpus file is generated.
- this is initiated with a "generate corpus" in the "File' menu of the compartment window (203) .
- individual corpora are first generated and stored in memory for each compartment, and the individual corpora are merged by selecting this option in the tool (202) "File' menu.
- the Annotated Corpus Dialog shown in figure 7 appears and the developer selects which corpora will be merged as well as the name of the newly merged corpus. Merging essentially appends the listings of the various corpora into a single listing.
- the resulting corpus file (122) is stored in memory and will be used by the interpreter (124) during runtime to process utterances that are heard by the speech recognizer (116) .
- a compartment corpus When a compartment corpus is generated, the enumeration traversal described above is performed for the tables within the compartment.
- valid utterances within a compartment begin and end in the compartment main table, but the valid utterance "paths" may pass through several tables referenced by non-terminals. Each valid path through the table is written (in text form) to the compartment corpus. Each valid path will also include token data (which serves as a token class identifier) found at the end of the valid path in the compartment's main table (the marked column).
- token data which serves as a token class identifier
- the resulting corpus (122) is actually a pair of files in the preferred embodiment, and the toolkit (104) generates both of these files when the various compartment corpora are merged into a single corpus.
- the preferred format for these files is as follows: corpus file:
- the first number indicates the class of the utterance.
- the second number denotes a member of this class.
- the first number is the utterance class, and the next line contains the token to be returned.
- utterances of class "1' will return the token "years' to the IVR (130), and utterances of class "2' will return the token "days' .
- the resulting corpus file (122) is ready for use at runtime by the runtime interpreter (124).
- the developer has completed development with the grammar development toolkit (104) .
- the toolkit is terminated' in step 628 by either selecting the "Exit' option from the File menu, or by clicking on the "x' in the upper right-hand corner of the toolkit window.
Abstract
Description
Claims
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AT98946100T ATE222008T1 (en) | 1997-09-17 | 1998-09-17 | SYSTEM AND METHOD FOR CREATING A LANGUAGE GRAMMAR |
CA002304251A CA2304251C (en) | 1997-09-17 | 1998-09-17 | System and method for creating a language grammar |
EP98946100A EP1016001B1 (en) | 1997-09-17 | 1998-09-17 | System and method for creating a language grammar |
DE69807091T DE69807091T2 (en) | 1997-09-17 | 1998-09-17 | SYSTEM AND METHOD FOR CREATING LANGUAGE GRAMMAR |
JP2000512155A JP3964134B2 (en) | 1997-09-17 | 1998-09-17 | Method for creating language grammar |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US08/932,937 | 1997-09-17 | ||
US08/932,937 US5995918A (en) | 1997-09-17 | 1997-09-17 | System and method for creating a language grammar using a spreadsheet or table interface |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1999014689A1 true WO1999014689A1 (en) | 1999-03-25 |
Family
ID=25463176
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/US1998/019432 WO1999014689A1 (en) | 1997-09-17 | 1998-09-17 | System and method for creating a language grammar |
Country Status (7)
Country | Link |
---|---|
US (1) | US5995918A (en) |
EP (1) | EP1016001B1 (en) |
JP (2) | JP3964134B2 (en) |
AT (1) | ATE222008T1 (en) |
CA (1) | CA2304251C (en) |
DE (1) | DE69807091T2 (en) |
WO (1) | WO1999014689A1 (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2000065474A1 (en) * | 1999-04-23 | 2000-11-02 | Nuance Communications | Method and apparatus for creating and editing grammars for speech recognition |
EP1546920A1 (en) * | 2002-09-06 | 2005-06-29 | Telstra Corporation Limited | A development system for a dialog system |
US7917363B2 (en) | 2003-02-11 | 2011-03-29 | Telstra Corporation Limited | System for predicting speech recognition accuracy and development for a dialog system |
US8296129B2 (en) | 2003-04-29 | 2012-10-23 | Telstra Corporation Limited | System and process for grammatical inference |
Families Citing this family (108)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6292767B1 (en) * | 1995-07-18 | 2001-09-18 | Nuance Communications | Method and system for building and running natural language understanding systems |
US6094635A (en) * | 1997-09-17 | 2000-07-25 | Unisys Corporation | System and method for speech enabled application |
US5995918A (en) * | 1997-09-17 | 1999-11-30 | Unisys Corporation | System and method for creating a language grammar using a spreadsheet or table interface |
US6499013B1 (en) * | 1998-09-09 | 2002-12-24 | One Voice Technologies, Inc. | Interactive user interface using speech recognition and natural language processing |
US6314402B1 (en) * | 1999-04-23 | 2001-11-06 | Nuance Communications | Method and apparatus for creating modifiable and combinable speech objects for acquiring information from a speaker in an interactive voice response system |
US6275792B1 (en) * | 1999-05-05 | 2001-08-14 | International Business Machines Corp. | Method and system for generating a minimal set of test phrases for testing a natural commands grammar |
US20020032564A1 (en) * | 2000-04-19 | 2002-03-14 | Farzad Ehsani | Phrase-based dialogue modeling with particular application to creating a recognition grammar for a voice-controlled user interface |
JP2003505778A (en) | 1999-05-28 | 2003-02-12 | セーダ インコーポレイテッド | Phrase-based dialogue modeling with specific use in creating recognition grammars for voice control user interfaces |
US6301554B1 (en) * | 1999-09-23 | 2001-10-09 | Wordstream, Inc. | Language translation using a constrained grammar in the form of structured sentences formed according to pre-defined grammar templates |
US7392185B2 (en) | 1999-11-12 | 2008-06-24 | Phoenix Solutions, Inc. | Speech based learning/training system using semantic decoding |
US9076448B2 (en) * | 1999-11-12 | 2015-07-07 | Nuance Communications, Inc. | Distributed real time speech recognition system |
US7050977B1 (en) | 1999-11-12 | 2006-05-23 | Phoenix Solutions, Inc. | Speech-enabled server for internet website and method |
US7725307B2 (en) | 1999-11-12 | 2010-05-25 | Phoenix Solutions, Inc. | Query engine for processing voice based queries including semantic decoding |
US6684183B1 (en) | 1999-12-06 | 2004-01-27 | Comverse Ltd. | Generic natural language service creation environment |
US8645137B2 (en) | 2000-03-16 | 2014-02-04 | Apple Inc. | Fast, language-independent method for user authentication by voice |
AU2001263138A1 (en) * | 2000-05-15 | 2001-11-26 | Infoactiv, Inc. | Automated voice-based dialogue with a voice mail system by imitation of the human voice |
US7024350B2 (en) * | 2000-07-20 | 2006-04-04 | Microsoft Corporation | Compact easily parseable binary format for a context-free grammer |
US7024348B1 (en) | 2000-09-28 | 2006-04-04 | Unisys Corporation | Dialogue flow interpreter development tool |
US6937986B2 (en) * | 2000-12-28 | 2005-08-30 | Comverse, Inc. | Automatic dynamic speech recognition vocabulary based on external sources of information |
US20020156616A1 (en) * | 2001-02-16 | 2002-10-24 | Russell Dale W. | Interactive tool for semi-automatic creation of a natural language grammar from a domain model |
US6985862B2 (en) * | 2001-03-22 | 2006-01-10 | Tellme Networks, Inc. | Histogram grammar weighting and error corrective training of grammar weights |
US7146308B2 (en) * | 2001-04-05 | 2006-12-05 | Dekang Lin | Discovery of inference rules from text |
US7257529B2 (en) * | 2001-09-10 | 2007-08-14 | Intel Corporation | Apparatus and method for an automated grammar file expansion tool |
JP3542578B2 (en) * | 2001-11-22 | 2004-07-14 | キヤノン株式会社 | Speech recognition apparatus and method, and program |
US7080004B2 (en) * | 2001-12-05 | 2006-07-18 | Microsoft Corporation | Grammar authoring system |
US7324942B1 (en) | 2002-01-29 | 2008-01-29 | Microstrategy, Incorporated | System and method for interactive voice services using markup language with N-best filter element |
US20030144846A1 (en) * | 2002-01-31 | 2003-07-31 | Denenberg Lawrence A. | Method and system for modifying the behavior of an application based upon the application's grammar |
US7054813B2 (en) * | 2002-03-01 | 2006-05-30 | International Business Machines Corporation | Automatic generation of efficient grammar for heading selection |
FR2842923B1 (en) * | 2002-07-26 | 2004-09-24 | Thales Sa | GENERATOR OF A CORPUS SPECIFIC TO A DOMAIN |
US6907397B2 (en) * | 2002-09-16 | 2005-06-14 | Matsushita Electric Industrial Co., Ltd. | System and method of media file access and retrieval using speech recognition |
US7783475B2 (en) * | 2003-01-31 | 2010-08-24 | Comverse, Inc. | Menu-based, speech actuated system with speak-ahead capability |
US20040190774A1 (en) * | 2003-03-28 | 2004-09-30 | Baker Daniel P. | Method for classifying and accessing writing composition examples |
US7996223B2 (en) * | 2003-10-01 | 2011-08-09 | Dictaphone Corporation | System and method for post processing speech recognition output |
EP1673763A1 (en) * | 2003-10-10 | 2006-06-28 | Philips Intellectual Property & Standards GmbH | A dialog control for dialog systems |
US7739117B2 (en) * | 2004-09-20 | 2010-06-15 | International Business Machines Corporation | Method and system for voice-enabled autofill |
US7426688B2 (en) | 2004-11-02 | 2008-09-16 | Hewlett-Packard Development Company, L.P. | Embedded spreadsheet commands |
WO2006127504A2 (en) * | 2005-05-20 | 2006-11-30 | Sony Computer Entertainment Inc. | Optimisation of a grammar for speech recognition |
US8677377B2 (en) | 2005-09-08 | 2014-03-18 | Apple Inc. | Method and apparatus for building an intelligent automated assistant |
US8607147B2 (en) * | 2005-12-09 | 2013-12-10 | International Business Machines Corporation | System and methods for previewing alternative compositions and arrangements when composing a strictly-structured flow diagram |
US7835911B2 (en) * | 2005-12-30 | 2010-11-16 | Nuance Communications, Inc. | Method and system for automatically building natural language understanding models |
US8229733B2 (en) * | 2006-02-09 | 2012-07-24 | John Harney | Method and apparatus for linguistic independent parsing in a natural language systems |
US7725308B2 (en) * | 2006-06-07 | 2010-05-25 | Motorola, Inc. | Interactive tool for semi-automatic generation of a natural language grammar from a device descriptor |
US9318108B2 (en) | 2010-01-18 | 2016-04-19 | Apple Inc. | Intelligent automated assistant |
US8149999B1 (en) * | 2006-12-22 | 2012-04-03 | Tellme Networks, Inc. | Generating reference variations |
US7954047B2 (en) * | 2007-08-06 | 2011-05-31 | Apple Inc. | Cutting and copying discontiguous selections of cells |
JP2009229529A (en) * | 2008-03-19 | 2009-10-08 | Toshiba Corp | Speech recognition device and speech recognition method |
US8996376B2 (en) | 2008-04-05 | 2015-03-31 | Apple Inc. | Intelligent text-to-speech conversion |
US8086444B2 (en) * | 2008-05-21 | 2011-12-27 | Resolvity, Inc. | Method and system for grammar relaxation |
US10241644B2 (en) | 2011-06-03 | 2019-03-26 | Apple Inc. | Actionable reminder entries |
US10241752B2 (en) | 2011-09-30 | 2019-03-26 | Apple Inc. | Interface for a virtual digital assistant |
US9431006B2 (en) | 2009-07-02 | 2016-08-30 | Apple Inc. | Methods and apparatuses for automatic speech recognition |
KR101597289B1 (en) * | 2009-07-31 | 2016-03-08 | 삼성전자주식회사 | Apparatus for recognizing speech according to dynamic picture and method thereof |
US8515734B2 (en) * | 2010-02-08 | 2013-08-20 | Adacel Systems, Inc. | Integrated language model, related systems and methods |
JP5718084B2 (en) * | 2010-02-16 | 2015-05-13 | 岐阜サービス株式会社 | Grammar creation support program for speech recognition |
US8682667B2 (en) | 2010-02-25 | 2014-03-25 | Apple Inc. | User profiling for selecting user specific voice input processing information |
US8560318B2 (en) | 2010-05-14 | 2013-10-15 | Sony Computer Entertainment Inc. | Methods and system for evaluating potential confusion within grammar structure for set of statements to be used in speech recognition during computing event |
US9262612B2 (en) | 2011-03-21 | 2016-02-16 | Apple Inc. | Device access using voice authentication |
US9280610B2 (en) | 2012-05-14 | 2016-03-08 | Apple Inc. | Crowd sourcing information to fulfill user requests |
US9721563B2 (en) | 2012-06-08 | 2017-08-01 | Apple Inc. | Name recognition system |
US9547647B2 (en) | 2012-09-19 | 2017-01-17 | Apple Inc. | Voice-based media searching |
US9330090B2 (en) * | 2013-01-29 | 2016-05-03 | Microsoft Technology Licensing, Llc. | Translating natural language descriptions to programs in a domain-specific language for spreadsheets |
US9582608B2 (en) | 2013-06-07 | 2017-02-28 | Apple Inc. | Unified ranking with entropy-weighted information for phrase-based semantic auto-completion |
WO2014197336A1 (en) | 2013-06-07 | 2014-12-11 | Apple Inc. | System and method for detecting errors in interactions with a voice-based digital assistant |
WO2014197334A2 (en) | 2013-06-07 | 2014-12-11 | Apple Inc. | System and method for user-specified pronunciation of words for speech synthesis and recognition |
WO2014197335A1 (en) | 2013-06-08 | 2014-12-11 | Apple Inc. | Interpreting and acting upon commands that involve sharing information with remote devices |
EP3008641A1 (en) | 2013-06-09 | 2016-04-20 | Apple Inc. | Device, method, and graphical user interface for enabling conversation persistence across two or more instances of a digital assistant |
US10176167B2 (en) | 2013-06-09 | 2019-01-08 | Apple Inc. | System and method for inferring user intent from speech inputs |
US9502031B2 (en) * | 2014-05-27 | 2016-11-22 | Apple Inc. | Method for supporting dynamic grammars in WFST-based ASR |
US9430463B2 (en) | 2014-05-30 | 2016-08-30 | Apple Inc. | Exemplar-based natural language processing |
US9338493B2 (en) | 2014-06-30 | 2016-05-10 | Apple Inc. | Intelligent automated assistant for TV user interactions |
US9582498B2 (en) | 2014-09-12 | 2017-02-28 | Microsoft Technology Licensing, Llc | Actions on digital document elements from voice |
US9668121B2 (en) | 2014-09-30 | 2017-05-30 | Apple Inc. | Social reminders |
US10176157B2 (en) | 2015-01-03 | 2019-01-08 | International Business Machines Corporation | Detect annotation error by segmenting unannotated document segments into smallest partition |
US10747498B2 (en) | 2015-09-08 | 2020-08-18 | Apple Inc. | Zero latency digital assistant |
US10671428B2 (en) | 2015-09-08 | 2020-06-02 | Apple Inc. | Distributed personal assistant |
US11010550B2 (en) | 2015-09-29 | 2021-05-18 | Apple Inc. | Unified language modeling framework for word prediction, auto-completion and auto-correction |
US10366158B2 (en) | 2015-09-29 | 2019-07-30 | Apple Inc. | Efficient word encoding for recurrent neural network language models |
US11587559B2 (en) | 2015-09-30 | 2023-02-21 | Apple Inc. | Intelligent device identification |
US10691473B2 (en) | 2015-11-06 | 2020-06-23 | Apple Inc. | Intelligent automated assistant in a messaging environment |
US10049668B2 (en) | 2015-12-02 | 2018-08-14 | Apple Inc. | Applying neural network language models to weighted finite state transducers for automatic speech recognition |
US10223066B2 (en) | 2015-12-23 | 2019-03-05 | Apple Inc. | Proactive assistance based on dialog communication between devices |
US10446143B2 (en) | 2016-03-14 | 2019-10-15 | Apple Inc. | Identification of voice inputs providing credentials |
US9934775B2 (en) | 2016-05-26 | 2018-04-03 | Apple Inc. | Unit-selection text-to-speech synthesis based on predicted concatenation parameters |
US9972304B2 (en) | 2016-06-03 | 2018-05-15 | Apple Inc. | Privacy preserving distributed evaluation framework for embedded personalized systems |
US10249300B2 (en) | 2016-06-06 | 2019-04-02 | Apple Inc. | Intelligent list reading |
US10049663B2 (en) | 2016-06-08 | 2018-08-14 | Apple, Inc. | Intelligent automated assistant for media exploration |
DK179309B1 (en) | 2016-06-09 | 2018-04-23 | Apple Inc | Intelligent automated assistant in a home environment |
US10067938B2 (en) | 2016-06-10 | 2018-09-04 | Apple Inc. | Multilingual word prediction |
US10509862B2 (en) | 2016-06-10 | 2019-12-17 | Apple Inc. | Dynamic phrase expansion of language input |
US10490187B2 (en) | 2016-06-10 | 2019-11-26 | Apple Inc. | Digital assistant providing automated status report |
US10586535B2 (en) | 2016-06-10 | 2020-03-10 | Apple Inc. | Intelligent digital assistant in a multi-tasking environment |
US10192552B2 (en) | 2016-06-10 | 2019-01-29 | Apple Inc. | Digital assistant providing whispered speech |
DK179343B1 (en) | 2016-06-11 | 2018-05-14 | Apple Inc | Intelligent task discovery |
DK179415B1 (en) | 2016-06-11 | 2018-06-14 | Apple Inc | Intelligent device arbitration and control |
DK201670540A1 (en) | 2016-06-11 | 2018-01-08 | Apple Inc | Application integration with a digital assistant |
DK179049B1 (en) | 2016-06-11 | 2017-09-18 | Apple Inc | Data driven natural language event detection and classification |
US9875235B1 (en) | 2016-10-05 | 2018-01-23 | Microsoft Technology Licensing, Llc | Process flow diagramming based on natural language processing |
WO2018085760A1 (en) | 2016-11-04 | 2018-05-11 | Semantic Machines, Inc. | Data collection for a new conversational dialogue system |
US10593346B2 (en) | 2016-12-22 | 2020-03-17 | Apple Inc. | Rank-reduced token representation for automatic speech recognition |
US10713288B2 (en) | 2017-02-08 | 2020-07-14 | Semantic Machines, Inc. | Natural language content generator |
US10586530B2 (en) | 2017-02-23 | 2020-03-10 | Semantic Machines, Inc. | Expandable dialogue system |
US11069340B2 (en) | 2017-02-23 | 2021-07-20 | Microsoft Technology Licensing, Llc | Flexible and expandable dialogue system |
US10762892B2 (en) | 2017-02-23 | 2020-09-01 | Semantic Machines, Inc. | Rapid deployment of dialogue system |
DK179745B1 (en) | 2017-05-12 | 2019-05-01 | Apple Inc. | SYNCHRONIZATION AND TASK DELEGATION OF A DIGITAL ASSISTANT |
DK201770431A1 (en) | 2017-05-15 | 2018-12-20 | Apple Inc. | Optimizing dialogue policy decisions for digital assistants using implicit feedback |
US11132499B2 (en) | 2017-08-28 | 2021-09-28 | Microsoft Technology Licensing, Llc | Robust expandable dialogue system |
US10496382B2 (en) * | 2018-02-22 | 2019-12-03 | Midea Group Co., Ltd. | Machine generation of context-free grammar for intent deduction |
US11837004B1 (en) * | 2023-02-24 | 2023-12-05 | Oracle Financial Services Software Limited | Searchable table extraction |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0681284A2 (en) * | 1994-04-29 | 1995-11-08 | Sun Microsystems, Inc. | Speech interpreter with a unified grammar compiler |
US5544298A (en) * | 1993-01-22 | 1996-08-06 | Data Management Corp. | Code generation and data access system |
WO1996026484A2 (en) * | 1995-02-22 | 1996-08-29 | Egilsson Agust S | Graphical environment for managing and developing applications |
Family Cites Families (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US33597A (en) * | 1861-10-29 | Improvement in the manufacture of springs from rubber and steel | ||
US4156868A (en) * | 1977-05-05 | 1979-05-29 | Bell Telephone Laboratories, Incorporated | Syntactic word recognizer |
USRE33597E (en) | 1982-10-15 | 1991-05-28 | Hidden Markov model speech recognition arrangement | |
US4914704A (en) * | 1984-10-30 | 1990-04-03 | International Business Machines Corporation | Text editor for speech input |
JPH05197389A (en) * | 1991-08-13 | 1993-08-06 | Toshiba Corp | Voice recognition device |
DE69232407T2 (en) * | 1991-11-18 | 2002-09-12 | Toshiba Kawasaki Kk | Speech dialogue system to facilitate computer-human interaction |
EP0930566A3 (en) * | 1992-07-06 | 2006-07-05 | Microsoft Corporation | Method and system for composing objects |
US5390279A (en) * | 1992-12-31 | 1995-02-14 | Apple Computer, Inc. | Partitioning speech rules by context for speech recognition |
US5652897A (en) * | 1993-05-24 | 1997-07-29 | Unisys Corporation | Robust language processor for segmenting and parsing-language containing multiple instructions |
US5623609A (en) * | 1993-06-14 | 1997-04-22 | Hal Trust, L.L.C. | Computer system and computer-implemented process for phonology-based automatic speech recognition |
US5799276A (en) * | 1995-11-07 | 1998-08-25 | Accent Incorporated | Knowledge-based speech recognition system and methods having frame length computed based upon estimated pitch period of vocalic intervals |
US5850629A (en) * | 1996-09-09 | 1998-12-15 | Matsushita Electric Industrial Co., Ltd. | User interface controller for text-to-speech synthesizer |
US5995918A (en) * | 1997-09-17 | 1999-11-30 | Unisys Corporation | System and method for creating a language grammar using a spreadsheet or table interface |
-
1997
- 1997-09-17 US US08/932,937 patent/US5995918A/en not_active Expired - Lifetime
-
1998
- 1998-09-17 WO PCT/US1998/019432 patent/WO1999014689A1/en active IP Right Grant
- 1998-09-17 JP JP2000512155A patent/JP3964134B2/en not_active Expired - Fee Related
- 1998-09-17 DE DE69807091T patent/DE69807091T2/en not_active Expired - Lifetime
- 1998-09-17 CA CA002304251A patent/CA2304251C/en not_active Expired - Fee Related
- 1998-09-17 EP EP98946100A patent/EP1016001B1/en not_active Expired - Lifetime
- 1998-09-17 AT AT98946100T patent/ATE222008T1/en not_active IP Right Cessation
-
2004
- 2004-04-21 JP JP2004125503A patent/JP2004258674A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5544298A (en) * | 1993-01-22 | 1996-08-06 | Data Management Corp. | Code generation and data access system |
EP0681284A2 (en) * | 1994-04-29 | 1995-11-08 | Sun Microsystems, Inc. | Speech interpreter with a unified grammar compiler |
WO1996026484A2 (en) * | 1995-02-22 | 1996-08-29 | Egilsson Agust S | Graphical environment for managing and developing applications |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2000065474A1 (en) * | 1999-04-23 | 2000-11-02 | Nuance Communications | Method and apparatus for creating and editing grammars for speech recognition |
US6434523B1 (en) | 1999-04-23 | 2002-08-13 | Nuance Communications | Creating and editing grammars for speech recognition graphically |
EP1546920A1 (en) * | 2002-09-06 | 2005-06-29 | Telstra Corporation Limited | A development system for a dialog system |
EP1546920A4 (en) * | 2002-09-06 | 2010-04-14 | Telstra Corp Ltd | A development system for a dialog system |
US8046227B2 (en) | 2002-09-06 | 2011-10-25 | Telestra Corporation Limited | Development system for a dialog system |
US7917363B2 (en) | 2003-02-11 | 2011-03-29 | Telstra Corporation Limited | System for predicting speech recognition accuracy and development for a dialog system |
US8296129B2 (en) | 2003-04-29 | 2012-10-23 | Telstra Corporation Limited | System and process for grammatical inference |
Also Published As
Publication number | Publication date |
---|---|
JP2003517647A (en) | 2003-05-27 |
JP3964134B2 (en) | 2007-08-22 |
US5995918A (en) | 1999-11-30 |
CA2304251C (en) | 2005-04-26 |
EP1016001B1 (en) | 2002-08-07 |
DE69807091T2 (en) | 2003-05-08 |
DE69807091D1 (en) | 2002-09-12 |
JP2004258674A (en) | 2004-09-16 |
ATE222008T1 (en) | 2002-08-15 |
CA2304251A1 (en) | 1999-03-25 |
EP1016001A1 (en) | 2000-07-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP1016001B1 (en) | System and method for creating a language grammar | |
US6094635A (en) | System and method for speech enabled application | |
US7024348B1 (en) | Dialogue flow interpreter development tool | |
EP1091303B1 (en) | Method and system for providing alternatives for text derived from stochastic input sources | |
US6173266B1 (en) | System and method for developing interactive speech applications | |
AU690830B2 (en) | Speech recognition system and method with automatic syntax generation | |
US5970448A (en) | Historical database storing relationships of successively spoken words | |
US6035275A (en) | Method and apparatus for executing a human-machine dialogue in the form of two-sided speech as based on a modular dialogue structure | |
EP1049072A2 (en) | Graphical user interface and method for modyfying pronunciations in text-to-speech and speech recognition systems | |
US11538466B2 (en) | Development of voice and other interaction applications | |
JP2007122747A (en) | Dialogue flow interpreter | |
US20210056951A1 (en) | Development of Voice and Other Interaction Applications | |
EP4018436A1 (en) | Development of voice and other interaction applications | |
Di Fabbrizio et al. | AT&t help desk. | |
US7031907B1 (en) | Tool for constructing voice recognition grammars | |
KR19990047859A (en) | Natural Language Conversation System for Book Libraries Database Search | |
JP3357213B2 (en) | Information processing device | |
Alshawi et al. | Variant transduction: A method for rapid development of interactive spoken interfaces | |
JP2023181684A (en) | Reception flow creation supporting device and reception flow creation supporting method | |
Turunen et al. | Speech application design and development | |
JPH01114973A (en) | Device for supporting document formation/calibration |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): CA CN IL JP SG |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
ENP | Entry into the national phase |
Ref document number: 2304251 Country of ref document: CA Ref country code: CA Ref document number: 2304251 Kind code of ref document: A Format of ref document f/p: F |
|
ENP | Entry into the national phase |
Ref country code: JP Ref document number: 2000 512155 Kind code of ref document: A Format of ref document f/p: F |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1998946100 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1998946100 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1998946100 Country of ref document: EP |