![]() |
KnowBrainer Speech Recognition | ![]() |
Topic Title: Acoustic Optimization & Language Model Optimization ? Topic Summary: Created On: 05/22/2020 03:58 AM Status: Post and Reply |
|
![]() |
![]() |
- swayzak | - 05/22/2020 03:58 AM |
![]() |
![]() |
- Lunis Orcutt | - 05/22/2020 10:46 PM |
![]() |
![]() |
- dicdoc | - 09/14/2020 04:52 PM |
![]() |
![]() |
- Lunis Orcutt | - 09/14/2020 08:40 PM |
![]() |
![]() |
- R. Wilke | - 05/23/2020 05:25 AM |
![]() |
![]() |
- Mav | - 05/25/2020 02:56 AM |
![]() |
|
Hello
Does running these make a noticeable difference to how well DMPE performs ?
I'm using the latest UK version of DMPE 4.3.
Thanks ------------------------- DMPE v4.2 (UK) |
|
|
|
![]() |
|
It depends on who you ask. Nuance's heart was likely in the right place but you know what they say about good deeds
------------------------- Forum Mission Statement |
|
|
|
![]() |
|
WHat do you mean by "rules" ? ------------------------- Memory: 16GB Microphone Brand and Model: LFH 4000 & Flexy Mike Operating System: win 10, latest update; Surface Book 2, Processor (CPU) Type & Speed: I7 8650 Soundcard Brand and Model: none Speech Software: Dragon Medical Practice Edition, french version: 4.2 Using KnowBrainer?:Yes |
|
|
|
![]() |
|
Select any word in the Vocabulary Editor followed by clicking the Properties button. Check out options ranging from Spoken Form to Printed Form to preceded before/after, capitalization context 1 and context 2. ------------------------- Forum Mission Statement |
|
|
|
![]() |
|
Anecdotal notice won't help, but quantified and reproducible testing on a representative amount of data has proven that it does improve accuracy: in the range of a few percent, at best, however. For the average user this may not be as critical, but for a professional using it as a tool to get his work (texting) done, it does have an impact. While the effect is typically more prevalent during the early stage of a user profile, it will diminish over time.
-------------------------
|
|
|
|
![]() |
|
As a side note: DMPE 4.3 (at least the German version, but I think UK version will be the same) follows the same great idea Nuance had with Dragon 15.5 in moving the speaker profile from %PROGRAMDATA% to %LOCALAPPDATA%.
If you're wondering why there is no profile at C:\ProgramData\Nuance\NaturallySpeaking15\Users\<YourSpeakerName> then probably because Dragon moved it to C:\Users\<YourWindowsUser>\AppData\Local\Nuance\NS15\Users, so you have to backup this folder now.
Apart from that, I also think Rüdiger's assessment is true. ACO can improve accuracy but other factors (like clear speech and good audio device) usually outweigh these gains. |
|
|
FuseTalk Standard Edition v4.0 - © 1999-2021 FuseTalk™ Inc. All rights reserved.