Speech screening

• US 20080294439A1
• Filed: 05/16/2008
• Published: 11/27/2008
• Est. Priority Date: 05/18/2007
• Status: Active Grant

• Current Assignee: Arlington Technologies, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Aurix Ltd. (Avaya Incorporated)
• Inventors: Kirby, James M.
• Granted Patent: US 8,121,845 B2
• Time in Patent Office: Days
• US Class Current: 704/251
• CPC Class Codes: G06F 16/685 using automatically derived transcript of audio data, e.g. lyrics speech recognition G10L15/00
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: H04L 51/04 Real-time or near real-time messaging, e.g. instant messaging [IM]
• CPC Class Codes: H04L 51/212 using filtering or selective blocking
• CPC Class Codes: H04M 2201/40 using speech recognition speech recognition per se G10L15/00
• CPC Class Codes: H04M 3/2281 Call monitoring, e.g. for law enforcement purposes; Call tracing; Detection or prevention of malicious calls
• CPC Class Codes: H04M 7/006 Networks other than PSTN/ISDN providing telephone service, e.g. Voice over Internet Protocol (VoIP) , including next generation networks with a packet-switched transport layer H04L65/00 takes precedence; aspects not specific to the type of network H04M3/00; special services in those networks H04M3/42

Abstract

This invention relates to screening of spoken audio data so as to detect threat words or phrases. The method is particularly useful for protecting children or vulnerable adults from unsuitable content and/or suspicious or threatening contact with others via a communication medium. The method is applicable to screening speech transmitted over a computer network such as the internet and provides screening of access to stored content, e.g. audio or multimedia data files, as well as real time speech such as live broadcasts or communication via voice over IP or similar communication protocols. The method allows an administrator, e.g. a parent, to identify groups of threat words or phrases to be monitored, to set user access levels and to determine appropriate responses when threat words or phrases are detected.

Claims

1. A method of providing access control for a user device comprising arranging a speech recogniser to process audio data a user attempts to access to detect any of a set of pre-defined threat words or phrases and, on detection of one or more said threat words or phrases, initiate a threat detection response.

2. A method as claimed in claim 1 wherein the audio data is part of an audio-video data file.

3. A method as claimed in claim 1 wherein the audio data may be stored locally on a memory in the user device.

4. A method as claimed in claim 1 wherein the user device is adapted to access audio data stored remotely.

5. A method as claimed in claim 1 wherein the set of pre-defined threat words or phrases comprise at least one of profane or obscene language, blasphemous language or language indicating violence or unacceptable attitudes.

6. A method as claimed in claim 1 wherein the threat words or phrases are categorised.

7. A method as claimed in claim 1 wherein the threat detection response comprises at least one of blocking access to some or all of the audio data, logging the attempt to access the audio data or communicating the attempt to access the audio data to a system administrator.

8. A method as claimed in claim 1 wherein the threat response depends on the threat word or phrase detected or category of the threat word or phrase.

9. A method as claimed in claim 1 wherein the threat response depends on the number of detected threat words in the audio data.

10. A method as claimed in claim 1 wherein the set or sets of pre-defined threat words checked for varies according to a user access level.

11. A method as claimed in claim 1 wherein the threat detection response varies according to a user access level.

12. A method as claimed in claim 1 wherein the method comprises the step of processing the whole of the audio to detect any threat words or phrases prior playing the audio to the user.

13. A method as claimed in claim 1 wherein segments the audio data are processed by the speech recogniser and, if acceptable, played to the user whilst the next segment of the audio data is processed.

14. A method as claimed in claim 1 wherein the processing of the audio data is carried out in real time but with a delay applied to the audio prior to playing to allow an appropriate threat response prior to an occurrence of a threat word.

15. A method as claimed in claim 1 further comprising the step of screening for the existence of any threat words or phrases in any textual data a user attempts to access and/or any metadata relating to data a user attempts to access.

26. A program for a computer adapted to perform the method of monitoring audio data transmitted over a computer network as claimed in claim 1.

28. A method as claimed in as claimed in claim 1 wherein the speech recogniser processes the audio data against various audio models to determine the likelihood of a segment of spoken audio corresponding to a particular phoneme.

29. A method as claimed in claim 28 wherein the threat words and phrases are converted to phoneme sequences and the search looks to identify any likely matches to that phoneme sequence.

16. An access control apparatus for a user device capable of accessing audio data comprising a speech recogniser adapted to process any audio data a user attempts to access so as to detect the presence of any pre-defined threat words or phrases and activate a threat detection response on detection of one or more said threat words.

17. A method of monitoring audio data transmitted over a computer network comprising arranging a speech recogniser to process said audio data to detect any of a set of pre-defined threat words or phrases and, on detection of one or more said threat words or phrases, initiate a threat detection response.

18. A method as claimed in claim 17 wherein the speech recogniser is adapted to process any audio data transmitted to the user device and also any audio data transmitted from the user device.

19. A method as claimed in claim 17 wherein the set of pre-defined threat words or phrases comprise at least one of profane or obscene language, blasphemous language, language indicating violence or unacceptable attitudes or grooming words or phrases.

20. A method as claimed in claim 17 wherein the threat words or phrases are categorised.

21. A method as claimed in claim 17 wherein the threat detection response comprises at least one of terminating the connection, removing or masking the threat word or phrase from the audio output from the device, logging the attempt to access the audio data, recording the spoken audio or communicating the attempt to access the audio data to a system administrator or specified agency.

22. A method as claimed in claim 17 wherein the threat response depends on the threat word or phrase detected or category of the threat word or phrase.

23. A method as claimed in claim 17 wherein the set or sets of pre-defined threat words checked for varies according to a user access level.

24. A method as claimed in claim 17 wherein the threat detection response varies according to a user access level.

25. A method as claimed in claim 17 wherein the communication is Voice-over-IP (VoIP) audio communication.

27. The use of speech recognition to monitor and regulate exposure of a user to offensive or inappropriate language transmitted over a network.