SYSTEM AND METHOD FOR RECOGNIZING A USER VOICE COMMAND IN NOISY ENVIRONMENT

• US 20130166279A1
• Filed: 02/21/2013
• Published: 06/27/2013
• Est. Priority Date: 08/24/2010
• Status: Active Grant

• Current Assignee: Veovox SA
• Original Assignee: Veovox SA
• Inventors: CARMONA, Jorge, DINES, John, MASSON, Olivier, ARADILLA, Guillermo
• Granted Patent: US 9,318,103 B2
• Time in Patent Office: Days
• US Class Current: 704/8
• CPC Class Codes: G10L 15/02 Feature extraction for speech recognition; Selection of recognition unit
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/12 using dynamic programming techniques, e.g. dynamic time warping [DTW]
• CPC Class Codes: G10L 15/16 using artificial neural networks
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 17/02 Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction

Abstract

An automatic speech recognition system for recognizing a user voice command in noisy environment, including: matching means for matching elements retrieved from speech units forming said command with templates in a template library; characterized by processing means including a MultiLayer Perceptron for computing posterior templates (P(O^template(q))) stored as said templates in said template library; means for retrieving posterior vectors (P(O^test(q))) from said speech units, said posterior vectors being used as said elements. The present invention relates also to a method for recognizing a user voice command in noisy environments.

Claims

1. An automatic speech recognition system for recognizing a user voice command in noisy environment, comprising:

matching means for matching elements retrieved from speech units forming said command with templates in a template library library;

processing means including a MultiLayer Perceptron for computing posterior templates (P(O^template(q))) stored as said templates in said template library;

means for retrieving posterior vectors (P(O^test(q))) from said speech units, said posterior vectors being used as said elements;

calculating means for automatically selecting posterior templates stored in said template library, wherein said calculating means use a graph approach, such as the Gabriel'"'"'s approach, or the relative neighbour approach, or a linear interpolation to prepare posterior templates from training templates.

2. The system of claim 1, further comprising:

a DTW decoder for matching posterior vectors with posterior templates.

3. The system of claim 2, further comprising

a voice activity detector;

and

a dictionary.

4. The system of claim 1, wherein said MultiLayer Perceptron is multilingual.

5. The system of claim 1, comprising at least two MultiLayer Perceptrons, wherein each of said MultiLayer Perceptrons is used for a specific language.

6. The system of claim 1, wherein said template library is a pre-existing template library generated from training templates spoken by another user.

7. The system of claim 1, comprising means for creating said template library from a pronunciation dictionary.

8. The system of claim 7, wherein said means comprise KL-divergence metric.

9. The system of claim 1, comprising means for automatically adapting said template library, the adaptation including the activation/deactivation and/or addition and/or deletion and/or substitution of said posterior templates.

10. The system of claim 9, wherein said adaptation uses a feedback of said user input on a user device.

11. The system of claim 1, comprising a grammar.

12. The system of claim 1, comprising a voice activity detector means that can be selected and de-selected by said user.

13. The system of claim 12, wherein said grammar is selected by means of said voice activity detector means.

14. The system of claim 2, wherein said DTW decoder incorporates an insertion penalty, a scale factor and a filter silence.

15. The system of claim 10, wherein said user device is adapted for allowing a user to enter said voice commands, and wherein said system comprises:

pre-processing means in said user device, adapted for pre-processing said entered voice commands;

connection means for transmitting pre-processed signals to a central server in a bar, restaurant or hotel;

restaurant, bar or hotel management software for managing bar, restaurant or hotel orders entered by said user through said voice commands.

16. An automatic speech recognition method of recognizing a voice command spoken by a user in a noisy environment, said method comprising:

matching elements retrieved from speech units forming said command with templates in a template library;

determining a sequence of templates that minimizes the distance between said elements and said templates;

wherein said templates are posterior templates (P(O^template(q))) and said elements retrieved from speech units are posterior vectors (P(O^test(q)));

said posterior templates and said posterior vectors being generated with at least one MultiLayer Perceptron;

wherein it includes a step for selecting said posterior templates from training templates, said step comprising;

determination of Gabriel/relative neighbours by calculating a matrix of distances between all said training template,

visiting each training template,

marking a training template if all its neighbours are of the same class as the current training template,

deleting all marked training templates.

17. The method of claim 16, wherein a DTW decoder is used for matching posterior vectors with posterior templates.

18. The method of claim 16, further comprising:

entering said voice commands corresponding to bar, restaurant or hotel orders in a user device;

pre-processing said voice commands in said user device;

transmitting pre-processed signals to a server;

converting said pre-processed signals into text orders in said server;

displaying said text orders;

communicating said orders to software and/or systems used by said bar, restaurant or hotel.

19. The method of claim 16, further comprising:

recording continuously said voice command by means of an audio acquisitioning system,

selecting a voice activity detector means,

de-selecting said voice activity detector means,

processing said voice command a time before of selecting said voice activity detector means and a time after de-selecting said voice activity detector means.