Speech recognition system and method for operating same

• US 20030139922A1
• Filed: 12/12/2002
• Published: 07/24/2003
• Est. Priority Date: 12/12/2001
• Status: Active Grant

• Current Assignee: Siemens AG
• Original Assignee: Siemens AG
• Inventors: Schneider, Tobias, Hoffmann, Gerhard
• Granted Patent: US 7,243,070 B2
• Time in Patent Office: Days
• US Class Current: 704/201
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/187 Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: H04M 1/271 controlled by voice recognition
• CPC Class Codes: H04M 1/7243 with interactive means for internal management of messages
• CPC Class Codes: H04M 2250/74 with voice recognition means voice activated dialling H04M1/271; voice recognition algorithms G10L15/00
• CPC Class Codes: H04W 4/14 Short messaging services, e.g. short message services [SMS] or unstructured supplementary service data [USSD]
• CPC Class Codes: H04W 4/18 Information format or content conversion, e.g. adaptation by the network of the transmitted or received information for the purpose of wireless delivery to users or terminals

Abstract

A speech recognition system and method are provided for the speech-controlled inputting of short messages into a telecommunications terminal, in particular a mobile phone or cordless phone, having a speech recognizer module which operates independently of the speaker, a text/phoneme converter module and/or an auxiliary speech recognition module which operates in a speaker-dependent fashion and has the purpose of converting the text inputs or text transfers into a phonetic script which is adapted to the speech recognition module, and a vocabulary control module for supplementing the vocabulary or for replacing elements of the vocabulary by words or phrases which have been input or transferred.

Claims

1. A speech recognition system for speech-controlled inputting of short messages into a telecommunications terminal, comprising:

a speech recognition module operating in speaker-independent fashion and in which a specific vocabulary for creating short messages is at least one of stored and processed;

a text input interface for at least one of directly inputting text and transferring text from a short message;

at least one of a text/phoneme converter module which is connected to the text input interface, and an auxiliary speech recognition module which operates in speaker-dependent fashion and converts the text into a phonetic script which is adapted to the speech recognition module; and

a vocabulary control module for at least one of supplementing the vocabulary and replacing elements of the vocabulary by the text which has been one of input and transferred.

2. A speech recognition system as claimed in claim 1, wherein at least one of the speech recognition module and the text/phoneme converter module is configured based on a neural network.

3. A speech recognition system as claimed in claim 1, wherein the text input interface has an alpha-numeric keypad integrated into the terminal with one of a multiple assignment of keys and an appropriately configured touchscreen.

4. A speech recognition system as claimed in claim 1, wherein a hidden Markov model is implemented in the speech recognition module.

5. A speech recognition system as claimed in claim 1, wherein a substitution control algorithm for replacing elements of the vocabulary by new text in accordance with a predetermined substitution strategy, as a function of at least one of time and frequency of their occurrence at the text input interface, is implemented in the vocabulary control module.

6. A speech recognition system as claimed in claim 1, further comprising parts for implementing a menu control for user-specific logic linking of phrases to in each case at least one of one voice instruction and a metaphor.

7. A speech recognition system as claimed in claim 1, further comprising parts for storing at least one of one voice instruction and a metaphor as a representation of phrases in the vocabulary of the speech recognizer.

8. A speech recognition system as claimed in claim 5, wherein the menu controller logically links in each case a plurality of phrases to a voice instruction, displays available phrases in reaction to the voice input of the voice instruction, and selects a currently desired phrase via an additional input operation.

9. A speech recognition system as claimed in claim 1, further comprising:

a message vocabulary buffer for temporarily storing the vocabulary of short messages which have been at least one of received and transmitted; and

a vocabulary comparator unit, connected to both the message vocabulary buffer and the speech recognition module, for comparing the stored vocabulary items and for outputting to the vocabulary control module words which have newly occurred in the short messages.

10. A speech recognition system as claimed in claim 9, further comprising:

a statistical evaluation unit for determining statistical characteristic variables of the newly occurred words; and

a characteristic variable comparator unit for comparing the acquired statistical characteristic variables with correspondingly characteristic variables of words stored in the speech recognizer, and for outputting a decision signal, representing a comparison result, to an input of the vocabulary control module for initiating one of supplementation and substitution of the vocabulary of the speech recognizer in response to the comparison result.

11. A speech recognition system as claimed in claim 1, wherein the speech recognizer has a vocabulary and phrase memory with a plurality of memory areas, with at least a first memory area containing a non-variant basic vocabulary.

12. A speech recognition system as claimed in claim 11, wherein a second memory area of the vocabulary and phrase memory holds a predetermined maximum number of phrases and optionally assigned voice instructions.

13. A speech recognition system as claimed in claim 12, wherein words which are transferred from the vocabulary of short messages which have been at least one of received and transmitted are stored, without influencing the memory content of the second memory area, in one of the first memory area and a third memory area of the vocabulary memory.

14. A speech recognition system as claimed in claim 1, wherein the text input interface is an external interface to another speech recognition system.

15. A speech recognition system as claimed in claim 14, wherein the another speech recognition system is one of a PC and a mobile computer.

16. A speech recognition method for speech-controlled inputting of short messages into a telecommunications terminal, the method comprising the steps of:

creating a basic vocabulary as a first step of a training phase for generating a speech recognizer vocabulary for use in creating short messages; and

performing, as a second step of the training phase, at least one of supplementing the basic vocabulary and replacing elements of the basic vocabulary via one of;

text inputting and subsequent text/phoneme conversion; and

combined text and voice inputting of new text.

17. A speech recognition method as claimed in claim 16, wherein the first step is carried out via reprogramming.

18. A speech recognition method as claimed in claim 16, wherein the first step is carried out together with the second step without previous programming.

19. A speech recognition method as claimed in claim 16, wherein the text inputting includes manual inputting of new words by a user.

20. A speech recognition method as claimed in claim 19, wherein the text inputting includes automatically supplying words which have been detected as being, with respect to the basic vocabulary, new words in short messages which have been at least one of transmitted and received.

21. A speech recognition method as claimed in claim 20, wherein the basic vocabulary and automatically supplied new words are stored in a first memory area of a vocabulary and phrase memory and manually inputted new words are stored in a second, separate memory area of the vocabulary and phrase area, such that the automatically supplied new words can replace elements of the basic vocabulary, but manually inputted words cannot.

22. A speech recognition method as claimed in claim 21, wherein, in the first step, the elements of the basic vocabulary are stored, in each case with a specification of a statistical characteristic variable which represents their significance for short messages, the new words which are detected are subjected to statistical evaluation in order to acquire a statistical characteristic variable which represents their significance for short messages, and elements of the basic vocabulary are substituted by newly supplied words as a function of a result of a comparison of their respective statistical characteristic variables.

23. A speech recognition method as claimed in claim 16, wherein manually inputted new words and optionally elements of the basic vocabulary are stored with an assignment of one of voice instructions and metaphors, and are displaced, together with further associated words, for selection within a scope of a menu prompting system during operation of the speech recognition system after the respective one assigned voice instruction and metaphor has been input.