Automated hotel attendant using speech recognition
First Claim
1. An automated hotel attendant system for coordinating room-to-room calling over a telephone switching system that supports a plurality of telephone extensions, comprising:
- a hotel guest database system for receiving and storing characters that spell the names of hotel guests and for assigning each guest an associated telephone extension;
a speech recognition system having a lexicon for storing word-phoneme data that associates a set of words with the phonemes that form those words;
a lexicon training system coupled to said database system for adding hotel guest names and corresponding phoneme data to said lexicon, said training system including a phoneticizer operable to convert the characters that spell a given hotel guests name into a plurality of pronunciations, where the phoneticizer employs letter-only decision trees and phoneme-mixed decision trees to produce said plurality of pronunciations; and
a call routing system coupled to said recognition system for converting a spoken name of a hotel guest entered from one of said telephone extensions into a predefined hotel guest name and using said hotel guest name to retrieve an assigned telephone extension from said database system;
said call routing system being interfaced with said telephone switching system to cause said telephone switching system to call said telephone extension in response to the entry of said spoken name from one of said telephone extensions.
2 Assignments
0 Petitions
Accused Products
Abstract
An automated hotel attendant is provided for coordinating room-to-room calling over a telephone switching system that supports multiple telephone extensions. A hotel registration system receives and stores the spelled names of hotel guests as well as assigns each guest an associated telephone extension. A lexicon training system is connected to the hotel registration system for generating pronunciations for each spelled name by converting the characters that spell those names into word-phoneme data. This word-phoneme data is in turn stored in a lexicon that is used by a speech recognition system. In particular, a phoneticizer in conjunction with a Hidden Markov Model (HMM) based model trainer serves as the basis for the lexicon training system, such that one or several HMM models associated with each guest name are stored in the lexicon. An automated attendant is coupled to the speech recognition system for converting a spoken name of a hotel guest entered from one of the telephone extensions into a predefined hotel guest name that can be used to retrieve an assigned telephone extension from the hotel registration system. Next, the automated attendant causes the telephone switching system to call the requested telephone extension in response to the entry of the spoken name from one of the telephone extensions.
70 Citations
12 Claims
-
1. An automated hotel attendant system for coordinating room-to-room calling over a telephone switching system that supports a plurality of telephone extensions, comprising:
-
a hotel guest database system for receiving and storing characters that spell the names of hotel guests and for assigning each guest an associated telephone extension;
a speech recognition system having a lexicon for storing word-phoneme data that associates a set of words with the phonemes that form those words;
a lexicon training system coupled to said database system for adding hotel guest names and corresponding phoneme data to said lexicon, said training system including a phoneticizer operable to convert the characters that spell a given hotel guests name into a plurality of pronunciations, where the phoneticizer employs letter-only decision trees and phoneme-mixed decision trees to produce said plurality of pronunciations; and
a call routing system coupled to said recognition system for converting a spoken name of a hotel guest entered from one of said telephone extensions into a predefined hotel guest name and using said hotel guest name to retrieve an assigned telephone extension from said database system;
said call routing system being interfaced with said telephone switching system to cause said telephone switching system to call said telephone extension in response to the entry of said spoken name from one of said telephone extensions. - View Dependent Claims (2, 3, 4, 5, 12)
a model trainer coupled to said phoneticizer for building an Hidden Markov Model (HMM) for each of said hotel guest names from said corresponding plurality of pronunciations and for storing said HMM in said lexicon for use by said speech recognition system.
-
-
3. The automated hotel attendant system of claim 2 further comprising a pronunciation extractor coupled between said phoneticizer and said model trainer for generating at least one aligned pronunciations to serve as input to said model trainer, said pronunciation extractor receiving a spoken pronunciation of a hotel guest'"'"'s name and being operable to select at least one of said plurality of pronunciations from said phoneticizer based on the spoken pronunciation of said hotel guest'"'"'s name.
-
4. The automated hotel attendant system of claim 1 wherein said hotel guest database system stores characters that spell the name of a first hotel service that is assigned a first telephone extension such that said automated hotel attendant system provides call routing to said first telephone extension in response to the entry of a spoken name of said first hotel service from one of said telephone extensions.
-
5. The automated hotel attendant system of claim 1 further comprising a spelled name recognizer coupled to said call routing system and said telephone switching system for coordinating room-to-room calling when said speech recognizer fails to recognize said spoken name of the hotel guest.
-
12. The automated attendant system of claim 1 wherein the letter-only decision trees include nodes representing questions about a given letter and neighboring letters in the given hotel guest name and the phoneme-mixed decision trees include nodes representing questions about a phoneme and neighboring phonemes in the given hotel guest name.
-
6. An automated attendant system for coordinating extension-to-extension calling over a telephone switching system that supports a plurality of telephone extensions, comprising:
-
a telephone database system for receiving and storing characters that spell an identifier for a telephone system user and for assigning each telephone system user an associated telephone extension;
a speech recognition system having a lexicon for storing word-phoneme data that associates a set of words with the phonemes that form those words;
a lexicon training system coupled to said telephone database system for adding said identifiers and corresponding phoneme data to said lexicon, said training system including a phoneticizer operable to convert the characters that spell a given identifier into a plurality of pronunciatons, where the phoneticizer employs letter-only decision trees and phoneme-mixed decision trees to produce said plurality of pronunciations, the letter-only decision trees having nodes representing questions about a given letter and neighboring letters in the given identifier and the phoneme-mixed decision trees having nodes representing questions about a phoneme and neighboring phonemes in the given identifier; and
a call routing system coupled to said recognition system for converting a spoken identifier of a system user entered from one of said telephone extensions into a predefined system user identifier and using said system user identifier to retrieve an assigned telephone extension from said telephone database system;
said call routing system being interfaced with said telephone switching system to cause said telephone switching system to call said telephone extension in response to the entry of said spoken identifier from one of said telephone extensions. - View Dependent Claims (7, 8, 9, 10)
a model trainer coupled to said phoneticizer for building an Hidden Markov Model (HMM) for each of said identifiers from said corresponding plurality of pronunciations and for storing said HMM in said lexicon for use by said speech recognition system.
-
-
8. The automated attendant system of claim 7 further comprising a pronunciation extractor coupled between said phoneticizer and said model trainer for generating at least one aligned pronunciations to serve as input to said model trainer, said pronunciation extractor receiving a spoken identifier of a system user and being operable to select at least one of said plurality of pronunciations from said phoneticizer based on the spoken pronunciation of said identifier.
-
9. The automated attendant system of claim 6 wherein said identifier is further defined as at least one of a name and a location associated with the assigned telephone extension.
-
10. The automated attendant system of claim 6 further comprising a spelled name recognizer coupled to said call routing system and said telephone switching system for coordinating room-to-room calling when said speech recognizer fails to recognize said spoken identifier of the system user.
-
11. An automated attendant system for providing a synthesized pronunciation of a hotel guests name over a telephone switching system that supports a plurality of telephone extensions, comprising:
-
a hotel guest database system for receiving and storing characters that spell the names of hotel guests and for storing information associated with each guest;
a speech synthesizer system having a lexicon for storing word-phoneme data that associates a set of words with the phonemes that form those words;
a lexicon training system coupled to said database system for adding hotel guest names and corresponding phoneme data to said lexicon, said training system including a phoneticizer operable to convert the characters that spell a given hotel guest name into a plurality of pronunciations, where the phoneticizer employs letter-only decision trees and phoneme-mixed decision trees to produce said plurality of pronunciations, the letter only decision trees having nodes representing questions about a given letter and neighboring letters in the given hotel guest name and the phoneme-mixed decision trees having nodes representing questions about a phoneme and neighboring phonemes in the given hotel guest name; and
a virtual attendant coupled to said hotel guest database system and said speech synthesizer system for converting a spelled name of a hotel guest entered by a system user of said hotel guest database system into a synthesized pronunciation of said spelled name using the word-phoneme data;
said virtual attendant being interfaced with said telephone switching system to provide said synthesized pronunciation to said system user over one of said telephone extensions.
-
Specification