How is it possible - ASR

Posted: Nov 16th, 2016 - 9:09 pm

How is it possible that ASR just gets stupider?? Omg, it is now hyphenating "posterior-lateral." This morning it was typing "taut," which was right, and now it's typing "taught." !!! This morning it was getting things right that it is now getting wrong! And it's typing sound-alikes that are even worse, like "morcellized" instead of "morselized," and "multi-link" instead of "multi-length" stent. It got these right before, now it's getting them all wrong. HOW is this even possible?

Post A Reply Reply By Email Options

Use Rich Text Editor with Full Options CLOSE

Edit/Delete Message

Report this message

X CLOSE

Complete Discussion Below: (

marks the location of current message within thread)

How is it possible - ASR (Views: 781, 2016-11-16, 9:09 pm)
- "Train ASR," they said - MT (Views: 685, 2016-11-16, 9:30 pm)
  - It's been really bad recently. 'corner artery disease." sm - Weary (Views: 594, 2016-11-16, 11:29 pm)
    - Corner artery disease - Made a macro (Views: 711, 2016-11-17, 5:23 pm)
      - Great idea! I'm going to do that, too. Today it was "ice" for "eyes." nm - Weary (Views: 582, 2016-11-17, 6:15 pm)
        Agree great idea - sm (Views: 603, 2016-11-17, 7:31 pm)
    - Sabotaged? - Chillywinter (Views: 662, 2016-11-17, 9:38 pm)
  - trained asr/MM technology suffering - typngal (Views: 661, 2016-11-17, 1:23 am)
    - ASR cannot catch its mistakes, it is not a person, - sitting ideally by watching, sm (Views: 769, 2016-11-18, 11:03 am)
      - voice recognition per ex boss - typngal (Views: 571, 2016-11-19, 11:34 pm)
        every MT has nothing to do with VR mistakes, - sm (Views: 823, 2016-11-22, 1:39 pm)
        This is interesting reading about VR and the way it learns - sm (Views: 449, 2016-11-22, 2:02 pm)
- Worse - Jayelle (Views: 724, 2016-11-17, 7:01 am)
- because it's not a person, it's a software package - nm (Views: 583, 2016-11-17, 7:59 am)
  - The software can be trained but has to be trained by the - person using it, sm (Views: 572, 2016-11-17, 12:49 pm)
    - Also, the term voice recognition means it recognizes one's voice - sm (Views: 556, 2016-11-17, 12:54 pm)
- How is it possible - SadMT (Views: 592, 2016-11-17, 8:19 am)
  - Update - noname (Views: 572, 2016-11-17, 9:24 am)
- I trained the original ASR - sm (Views: 621, 2016-11-17, 4:11 pm)
  - you cannot train a system for a specific doctor's voice, - the best you can do, sm (Views: 515, 2016-11-18, 10:46 am)
    - Those of us there in the beginning - sm (Views: 602, 2016-11-18, 11:13 am)
      - It is not unlearning. - sm (Views: 638, 2016-11-18, 11:37 am)
        It actually does go back through - LDSMT (Views: 516, 2016-11-18, 2:46 pm)
        This is also what my supervisor told me - When it first came out (Views: 610, 2016-11-18, 3:38 pm)
        Well, there are thousands of documents, time consuming - sm (Views: 824, 2016-11-22, 1:43 pm)