|
 eTIPS
Did you know you could use voice commands on Windows® Desktop?
You can use voice commands to control many actions on a Windows Desktop; commands such as:
- Open Windows Explorer
- Close Window
- Open My Computer
- Maximize window
- Minimize window
- Switch to "Document/Application" name
Click the play button on the following eTIPS video image for a brief demonstration about how to use Voice Commands on Windows Desktop.
_______________back to top_______________
Did you know what is involved with a Dragon Medical 360 | Network Edition 1.0 SP5 Upgrade?
Dragon Medical 360 | Network Edition (DM360 Network Edition) 1.0 SP5 is released and available for DM360 Network Edition 1.0 SP3 customers wanting to upgrade. Refer to the DM360 Network Edition 1.0 SP5 release note (L-3656) from iSupport solution 14274, "Dragon Medical 360 | Network Edition or Enterprise Edition What's New and Can I Download the Current Version" for a summary of new features, fixes, known issues, and a historical listing of build versions. DM360 Network Edition 1.0 SP5 is not supported with Nuance cloud-based products. DM360 Network Edition 2.0 is required for the Nuance cloud-based solution.
The following chart defines product identifiers to assist in understanding DM360 Network Edition version control.
The following table outlines supported upgrade paths and product requirements for the DM360 Network Edition 1.0 SP5 release.
Pre-requisites
The items below should be verified prior to performing the DM360 Network Edition 1.0 SP5 upgrade:
- .NET Framework 4.0 installed on all DM360 Network Edition servers
- DM360 Network Edition 1.0 SP3 version currently installed
- NMS/NMC version 2.0.0.535
- PO and Speech Node version 1.30.0004 or higher
- Dragon SDK version 10.50.654.263
- DM360 Network Edition 1.0 SP5 software installation media downloaded from iSupport solution 14274 and extracted to DM360 Network Edition servers or a shared network location
Summarized Upgrade Procedure
- Within the Nuance Management Console (NMC), confirm the Profile Optimization Tasks have completed and the Profile Optimizer Status is Idle.
- Stop the DM360 Network Edition services.
- Backup the DM360 Network Edition SQL databases (NuanceMC and NuancePODB).
- Upgrade NMS server software (uninstall not required).
- Log in to the new NMC and create an Organization Token for Profile Optimizer server install.
- Uninstall the legacy Dragon SDK software on all Speech Nodes.
- Install the new SP5 Dragon SDK Client software.
- Upgrade the Profile Optimizer and Speech Node software (uninstall not required).
- Reboot servers.
Detailed Upgrade Procedure
Refer to iSupport Solution 16199, "DM360 Network Edition 1.0 SP5 Upgrade Procedure," for a detailed procedure on how to upgrade an on-premise DM360 Network Edition 1.0 SP3 system to SP5. The steps are at the end of the article.
For additional information about Dragon Medical 360 | Network Edition 1.0 SP5 Upgrade, login to iSupport and search for solution 14274, "Dragon Medical 360 | Network Edition or Enterprise Edition What's New and Can I Download the Current Version." If you are not an iSupport subscriber, click iSupport Timesaver for information about how you may get started with iSupport.
_______________back to top_______________
Did you know the Best Practice recommendations for Dragon Medical 360 | Network Edition 2.0 BestMatch V speech/acoustic models?
New to the Dragon Medical 360 | Network Edition 2.0 (DM360 Network Edition 2.0) product and its version 12.5 client software are improved speech and acoustic models. These models are known as BestMatch IV and BestMatch V.
BestMatch V
The BestMatch V model provides the highest recognition accuracy to date. By default, the DM360 Network Edition 2.0 client automatically selects the BestMatch V model during new user profile creation when the following requirements are met at the client workstation:
- Two (2) or more processors
- Four GB of RAM memory, or higher
Note: A 64-bit operating system is recommended.
There is both a speech model as well as acoustic models for the BestMatch V model operation.
BestMatch V Speech Model
The BestMatch V speech model is known as a "triple-pass" model because this model performs two (2) simultaneous recognition passes, and if the recognition accuracy from the dual pass does not meet certain criteria, an additional third (single) recognition pass is made. This "triple-pass" action occurs each time a dictator speaks an utterance. The BestMatch V speech model operates from RAM memory. DM360 Network Edition 2.0 provides several BestMatch V speech models:
BestMatch V Multiple Acoustic Models
BestMatch V assigns two (2) acoustic models to each dictation source for a User Profile. Further, both acoustic models are loaded into RAM memory and interact with the recognition process. Using multiple acoustic models instead of one (1) model improves recognition and dictation accuracy.
A User Profile with two (2) BestMatch V acoustic models uses more computer resources, including RAM memory, than a User Profile with one (1) acoustic model.
BestMatch IV Model
BestMatch IV models are improved over the BestMatch III models from DM360 Network Edition 1.0, version 10.5 clients, and also use two (2) simultaneous recognition passes like the BestMatch V model. However, no third recognition pass is made. Further, only one (1) single acoustic model is used per dictation source as compared with two (2) acoustic models used with the BestMatch V model. Finally, while the acoustic model operates from RAM memory, the language model operates from the hard drive.
Therefore, it is imperative to use appropriately-resourced workstations that facilitate the assignment of BestMatch V models during User Profile creation. Conversely, if a BestMatch V user profile is opened on an under-resourced client workstation, the BestMatch V model will revert to BestMatch IV operation, using single recognition pass and single acoustic model operation. In this case, Dragon displays the following warning message:
Your computer has a single core processor. You have opened a User Profile created on a computer with a multi-core processor. Dragon NaturallySpeaking will work normally on this computer but you may see some change in performance and accuracy.
Refer to the table below for a summary of Model attributes:
Method 1
During DM360 Network Edition 2.0 user profile creation on the "Review your choices" window, click the Advanced button to view the Speech model selected by Dragon.
Method 2
After the user profile has been created; check the user profile properties by highlighting the user profile and clicking the Properties button on the "Open User Profile" window.
Method 3
Use the dragon.log file content.
During profile creation Dragon determines the best model based on workstation resources:
S2: Info: GetBestModelAndTopic:, MHz 2533 MB 3512 cores 2 TopicType <General Medical> SourceType <HEADSET_MIC>, Lang 0, AcousticType <*> Age 3
S2: Info: using SRSF 4613 instead of MHz 2533
S2: Possible pair is <US English | Medical BestMatch IV>-<US English | Medium | General Medical><us_english_fmed_bm4>-<us_english_medium_medical> (accuracy = 25)
S2: Possible pair is <US English | Medical BestMatch V>-<US English | Medium | General Medical><us_english_fmed_bm5>-<us_english_medium_medical> (accuracy = 27)
S2: Info: best pair is <US English | Medical BestMatch V> <US English | Medium | General Medical>
When opening a BestMatch V profile:
LOG (Perf): Selecting user based on US English | Medical BestMatch V (ID=10711)
S2: Info: base model is <US English | Medical BestMatch V>
Important Best Practice Recommendations
- If a DM360 Network Edition 2.0 upgrade is performed where legacy DM360 Network Edition 1.0 profiles are being upgraded, each DM360 Network Edition 2.0 Speech Node must be have two (2) or more processors and at least 2 GB of RAM memory. Otherwise, the following conditions can occur:
- The Profile upgrade process may fail due to high CPU utilization.
- Profiles upgraded on single-processor Speech Nodes, and then opened on multi-processor workstations, present warning messages about reduced recognition accuracy. This is the result of a lesser speech model (BestMatch IV instead of BestMatch V) being assigned during the upgrade process when performed on single-processor Speech Nodes.
- Customers having providers initially create their user profiles during training need to make sure training workstations meet the requirements for BestMatch V model assignment as previously outlined at the beginning of this article. Commonly, training occurs on under-resourced workstations resulting in BestMatch IV instead of BestMatch V assignments. This means that regardless of whether production workstations meet BestMatch V model requirements, once the provider's user profile has already been assigned to BestMatch IV, BestMatch V recognition accuracy improvements will not be realized for these providers created with BestMatch IV models.
back to top
| 
What's New
iSupport is a Proven Timesaver
Contacting Support
