time Microsoft filed they patent, had they wanted to do any searching:
Publ.
Date.  Number      Title
-----  ---------   ---------------------------------------------------------
1993   5,199,077   Wordspotting for voice editing and indexing
2004   6,829,582   Controlled access to audio signals based on objectionable
                       audio content detected via sound recognition
2000   6,166,780   Automated language filter
1999   5,870,708   Method of and apparatus for scanning for and replacing
                       words on video cassettes
1994   5,369,440   System and method for automatically controlling the audio
                       output of a television
So let's look at this prior art, which neither Microsoft nor the PTO did, to see how crappy Microsoft's patent is.
The first patent, dating back to 1993 (eleven years before Microsoft filed) is a foundational patent on pattern recognition systems (in this case, Hidden Markov Models) to learn to recognize and delete segments of speech, which includes questionable segments of speech that are being broadcasted:
     _________________________________________________________________
   United States Patent       5,199,077
   Wordspotting for voice editing and indexing
                                  Abstract
   A  technique  for  wordspotting based on hidden Markov models (HMM's).
   The  technique allows a speaker to specify keywords dynamically and to
   train  the  associated  HMM's  via  a  single repetition of a keyword.
   Non-keyword  speech is modeled using an HMM trained from a prerecorded
   sample   of   continuous  speech.  The  wordspotter  is  intended  for
   interactive  applications,  such  as  the  editing  of  voice  mail or
   mixed-media  documents,  and  for  keyword  indexing in single-speaker
   audio or video recordings.
   From the SUMMARY OF THE INVENTION:
   An  important application for the wordspotting system of the invention
   is   not   only  indexing  of  recorded  speech,  but  especially  for
   interactive  voice  editing  of  recorded  speech, such as voice mail,
   dictation, or audio documentation. Wordspotting can be employed by the
   user  to  enable  editing operations by locating specific words in the
   recorded speech for deletion, substitution, or insertion. It will also
   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   enable  efficient  and  automatic  means  of  indexing into long audio
   documents.  The system, while restricted to a single speaker, or pairs
   of speakers is not restricted in vocabulary size.
     _________________________________________________________________
One application of this patent is to use the Hidden Markov Models on recorded speech THAT IS BROADCASTED, using the detections to delete specific words, i.e., Microsoft's patent. Almost complete anticipation, especially when obviously combined with the following patent:
     _________________________________________________________________
   United States Patent      5,369,440
   System  and method for automatically controlling the audio output of a
   television
                                  Abstract
   A  system  and  method  for automatically controlling the audio output
   from  a  television  so  as  to  avoid listening to undesired material
   includes  a  microphone  for  converting  the  audio  output  from the
   television  into  audio signals, an audio amplifier for amplifying the
   audio  signals  from  the microphone, a waveform pattern comparator, a
   waveform  digitizer  and  recorder,  a  speech  recognition  unit,  an
   external  device controller, a computer and an input/output device. In
   setting  up  the system for subsequent use, waveform patterns of audio
   signals  corresponding to material being outputted from the television
   which the listener considers undesirable are digitized by the waveform
   digitizer  and  recorder  and  then  stored  in  the  waveform pattern
   comparator. In addition, digital signals corresponding to key words in
   undesired   material   are  entered  into  the  computer  through  the
   input/output  device  and  then  transferred  from the computer to the
   speech  recognition  unit.  Once  the  system has been loaded with the
   undesired  material, waveform pattern comparator continually digitizes
   and compares the audio signal output from the audio amplifier with the
   stored  data.  At  the same time, the speech recognition unit compares
   the  audio signal output with the stored key words. When the undesired
   material  is  detected  an  indicator  signal  is sent to the external
   device  controller  which outputs a control signal which is applied to
   the  television  to  make  some  adjustment to either the sound or the
   channel to avoid listening to the undesired material.
     _________________________________________________________________
So two patents, published in 1993 and 1994, TEN YEARS BEFORE MICROSOFT FILED THEIR PIECE OF CRAP, completely anticipate Microsoft's supposed invention.  Just to have fun, I found three more patents, also so much on point to be powerful prior art.
     _________________________________________________________________
   United States Patent      6,166,780
   Automated language filter
                                  Abstract
   A  method and apparatus for analyzing the closed captioned aspect of a
   video signal for specific undesirable words or phrases and then muting
   the  audio  portion  of those words or phrases while not affecting the
   video  portion  therein  while  simultaneously  modifying  the  closed
   captioned signal in order to display only acceptable words or phrases.
     _________________________________________________________________
     _________________________________________________________________
   United States Patent      6,829,582
   Controlled  access  to  audio  signals  based  on  objectionable audio
   content detected via sound recognition
                                  Abstract
   An   apparatus,   program  product,  and  method  restrict  access  to
   objectionable  audio  content  in an audio or audio/video transmission
   using  sound  recognition.  Sound  recognition  may  be performed, for
   example,  to  detect  and  control  access to objectionable non-spoken
   audio  content,  e.g.,  by  detecting  violent sounds such as screams,
   explosions,  gun shots, sirens, punches, kicks and/or other non-spoken
   content  such  as sexually-suggestive sounds. In addition, occurrences
   of  objectionable  audio content detected in an audio transmission may
   be  tracked so that access to the audio transmission may be controlled
   responsive   to   the   identification   of  multiple  occurrences  of
   objectionable audio content. Furthermore, access control over detected
   objectionable  audio  content  in  an audio transmission may result in
   inhibition   of   access  to  a  program  associated  with  the  audio
   transmission.
     _________________________________________________________________
     _________________________________________________________________
   United States Patent      5,870,708
   Method  of and apparatus for scanning for and replacing words on video
   cassettes
                                  Abstract
   A  scanning apparatus for scanning a video recording for objectionable
   content  is  disclosed  that includes an audio processor for analyzing
   the  recording  and  recognizing  the  objectionable  audio, and audio
   control  gate  for  manipulating the objectionable audio. The scanning
   apparatus  also  includes  an  amplifier that amplifies the signal and
   converters  to  convert the signal from analog to digital format prior
   to analysis and digital to analog format after analysis. The method of
   scanning  the  video  recording for objectionable content includes the
   steps  of  analyzing the recording and manipulating the recording. The
   method  first  includes  the steps of amplifying the recording signal,
   separating  the  audio  portion  of  the  recording from the composite
   recording,  digitizing  the portions of the recording, and storing the
   composite  portion  of  the  recording  in  a  loop  while analysis is
   performed on the audio portion of the recording.
     _________________________________________________________________
So five patents I found in 15 minutes completely invalidate Microsoft's crappily issued patent.  And this doesn't rely on even more patents, and even much more non-patent prior art that is available, had anyone bother to do any searching (apparently a firable offense at Microsoft).
15 minutes neither Microsoft nor the PTO spent doing a similar search. Because to big companies like Microsoft, and big liars like Jon Dudas, patent quality is still a big freaking joke. So when con artists like IBM lie about the need for public peer review of patents, while remaining silent about the incompetence of PTO management with regards to prior art
handling, it is just con artists cheating.  And when liars like Jon Dudas chandleristically whine that the PTO needs more prior art, it is a big freaking lie because the PTO still is unable to make use of all of the patent prior art it has complete access to, let alone non-patent prior
art.  All of these people are liars, especially since it takes longer for them to write their liars about prior art that it does for me to find the prior art.
                                  ==========
    Automatic censorship of audio data for broadcast
    Microsoft
    U.S. Patent 7,437,290
    1. A  method  for  automatically censoring audio data, comprising the
    steps of:
    (a)  automatically  processing the audio data to detect any undesired
    speech  that  may  be  included  therein,  by comparison to undesired
    speech  data,  by performing the following steps;  comparing words in
    the  audio data  against  words  comprising  the undesired speech, to
    identify   potential  matches;  dynamically   varying   a probability
    threshold  dependent upon  at  least  one criterion; and based upon a
    probability of  a  potential  match  and  the  probability threshold,
    determining whether  any  undesired  speech  is included in the audio
    data;
    (b) for each occurrence  of undesired speech  that  is  automatically
    detected, altering the undesired speech detected in  the audio  data,
    producing  censored  audio  data  in  which  the undesired  speech is
    substantially no longer perceivable by a listening audience; and
    (c)  dynamically  adjusting  the  probability threshold  based upon a
    frequency with  which  undesired  speech  by  a  specific  speaker is
    detected in  the  audio data, so that as the occurrences of undesired
    speech that  are  detected  increase,  the  probability  threshold is
    reduced                     
Tuesday, November 11, 2008
Microsoft gets awarded yet another excrement patent
Case in point - one of the revolting pieces of fecal patents just issued recently to Microsoft.  It demonstrates yet again the contempt of large companies do to any prior art searching, the love of large companies to flood the PTO with crap applications to choke the system while whining
about others who do the same [now known as Chandlerism], and once again, it demonstrates the PTO'S COMPLETE AND TOTAL MISMANAGEMENT OF PRIOR ART SEARCHING AND RESOURCES.  When Jon Dudas told Congress that most, if not all, measures of quality at the PTO are their highest in 25 years, he deliberately LIED to Congress.  Whoever is paying Dudas to wreck the PTO is getting their monies worth.
The patent.  The USPTO belched/oozed/barfed out U.S. patent 7,437,290 to Microsoft:
    Automatic censorship of audio data for broadcast
    U.S. Patent 7,437,290
    MICROSOFT  (filed October 2004, issued October 2008)
Despicable claim language aside, this patent is a method for detecting undesired speech in broadcasts, censoring (deleting, bleeping) the undeisred speech and updating probability tables for the presence of the recognized speech.  COMPLETELY UNINNOVATIVE TECHNOLOGY.  The claim is attached below.
So how much tremendous amounts of prior art, with Bill Gates' billions to sustain decent prior art searching (NOT - a lesson well learned by its stepchild Intellectual Ventures) - how much prior art is associated with this patent?  NEXT TO FREAKING NOTHING.  The patent cites two prior patents, and one non-patent prior art published too late:
    7,139,031    Automated language filter for TV receiver
    6,337,947    Method and apparatus for customized editing of
                     video and/or audio signals
    Seide et al, "Vocabulary-Independent Search in Spontaneous Speech",
                 IEEE International Conference on ASSP, May 2004
Such pathetic prior art submissions should a priori be automatic proof of intent to deceive the Patent Office.  Just how pathetic is this patent? Once again, let me do a 15-minute-Jon-Dudas-High-Quality-Patent-Bust. Thebust relies on five patents that were published by the time the examinershould have been doing ANY searching, four of which were published at the
Subscribe to:
Comments (Atom)
 
