Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition
Reexamination Certificate
1999-04-20
2001-12-18
Smits, Talivaldis I. (Department: 2641)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Recognition
C704S251000, C704S270000
Reexamination Certificate
active
06332120
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to content verification techniques in the broadcast industry, and in particular to methods and apparatus for monitoring airplay (i.e., radio and television broadcasts) using a speech recognition (SR) (also referred to herein as “voice recognition” or “VR”) system.
Today's media environment provides an almost limitless set of outlets to reach consumers. The content is much as it has been for decades, i.e., advertising, programs, music, and news. What has changed dramatically is that the geographical stratification (e.g., national, regional, local) has been blurred. Particularly true of news, alerts, and commentaries, this blurring means that information is disseminated more quickly than ever. The rapid expansion of media outlets, including radio, television, cable, and the Internet, are proving to be a boon in terms of reaching consumers with information. This has many useful implications, including cost competition to reach an ever-fractionalized marketplace.
All of this calls for an effective and cost-efficient means for monitoring broadcasts that differs from existing systems. Existing systems rely on either (i) passive pattern matching, particularly in the case of music, or (ii) the use of an embedded signal. In the latter instance, a unique signal is encoded into transmitted information (“content”) such that the signal can be monitored to verify play. For example, a radio advertisement (“commercial”) can include an inaudible signal that is monitored by a special receiver. The receiver detects the inaudible signal to verify that the commercial has been played at the proper time. Each of the existing systems noted above has relative advantages and disadvantages. A common disadvantage of both, however, is that they are not able to intercept spontaneous comments (e.g., impromptu statements made by newscasters), live news feeds, or other such content.
It would be advantageous to provide a method and apparatus for improving the tracking of spoken information that is broadcast to a population of receivers, e.g., over the airwaves, on cable, over the Internet, or the like. It would be further advantageous to provide a cost-effective means for tracking broadcast channels for specific words, phrases, and contexts of these.
The present invention provides a method and apparatus having the aforementioned and other advantages. In particular, the present invention enables specific words and phrases to be automatically detected such that the information can serve numerous purposes. For example, novel services can be offered by providing automated tracking of advertising, newscasts, and programs. In this manner, a cost-effective system is provided for, e.g., tracking public relations efforts, localizing news stories, as well as enabling simple verification of airplay of contracted materials. The invention uses voice recognition technology to achieve these and other objectives.
SUMMARY OF THE INVENTION
In accordance with the invention, a method is provided for monitoring broadcast audio for information of interest. A vocabulary of keywords of interest is provided. Received audio is processed using a voice recognition system to detect when one of the keywords appears in the audio. A report is generated based on the detection of a keyword.
In an illustrated embodiment, successive portions of the audio are stored for processing during the processing step. Portions of the stored audio adjacent detected keywords are analyzed to ascertain the context in which the keywords are used. Portions of audio before and after the occurrence of a keyword can be stored for the context analysis. The context can be used to determine an action taken in response to the detection of the keyword. For example, the action can comprise an alert when the context comprises a crisis (e.g., an emergency or an unexpected problem that needs to be quickly resolved).
A portion of audio adjacent a detected keyword can be stored for inclusion with the report as an audio clip. In this manner, a person receiving the report can also hear the detected keyword in context within a segment of received audio. Preferably, portions of audio before and after the occurrence of the detected keyword will be stored for inclusion in the audio clip.
In accordance with the invention, the type of audio program received can also be detected. For example, the program may be a talk show, news broadcast, editorial, commercial advertisement, or other audio segment. Information identifying the audio program type can then be provided in the report. Moreover, the amount of audio stored for context analysis can be based on the type of audio program received.
In another embodiment, broadcast audio is received and input to a voice recognition system. The voice recognition system converts speech portions of the received audio to text. The text is analyzed using a computer to locate at least one predetermined keyword therein. A report is then generated in response to the analyzing step upon the occurrence of the at least one keyword.
In such an embodiment, successive portions of the text can be stored for processing during the analyzing step. Portions of the stored text adjacent the at least one keyword are analyzed to ascertain a context in which the keyword was used. Preferably, portions of text both before and after the occurrence of the keyword are stored for the context analysis.
Apparatus in accordance with the invention includes a computer processor. A memory associated with the computer processor stores a vocabulary of keywords of interest. An audio receiver provides received audio to a voice recognition system that is also associated with the computer processor. The voice recognition system detects when one of the keywords appears in audio received by the audio receiver. A report generator associated with the computer processor is responsive to the detection of a keyword for generating a report. The report is indicative of the detected keyword and, when implemented, the context in which the keyword was used.
The report generator can be provided locally or at a central processing station remote from the computer processor, audio receiver and voice recognition system. In the event that the report generator is located at a remote (e.g., central) location, the computer processor communicates with the report generator via a communications network.
The computer processor can be provided with the capability of analyzing the context in which detected keywords are used. In such cases, the report generator can be responsive to the context analysis for generating the report.
In an alternate embodiment, the voice recognition system converts the audio to text. The keyword detection is then provided by comparing words in the text to words in the stored keyword vocabulary.
REFERENCES:
patent: 5051924 (1991-09-01), Bergeron et al.
patent: 5168548 (1992-12-01), Kaufman et al.
patent: 5231670 (1993-07-01), Goldhor et al.
patent: 5481296 (1996-01-01), Cragun et al.
patent: 5561457 (1996-10-01), Cragun et al.
patent: 5612729 (1997-03-01), Ellis et al.
patent: 5664227 (1997-09-01), Mauldin et al.
patent: 5859662 (1999-01-01), Cragun et al.
patent: 5970448 (1999-10-01), Goldhor et al.
patent: 6011854 (2000-01-01), Van Ryzin
patent: 6061056 (2000-05-01), Menard et al.
patent: 6125347 (2000-09-01), Cote et al.
patent: 10028068 (1998-01-01), None
Lipsitz Barry R.
Smits Talivaldis I.
Solana Technology Development Corporation
LandOfFree
Broadcast speech recognition system for keyword monitoring does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Broadcast speech recognition system for keyword monitoring, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Broadcast speech recognition system for keyword monitoring will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2575139