Voice recognition not working

[Phillip Gallagher 101]

No matter what I do I cannot get through the voice training, it either pauses or gets stuck.  Then in the program the copilot does not respond or hear my voice at all.  The problem is not my microphone.  It may be some issue with Windows 10 since Voxatc doesn’t work either...with Voxatc I get the error message “speech training error” “speech training is not configured properly....

 

anybody know now how to fix this? 

[FS++ 710]

5 hours ago, Phillip Gallagher said:

No matter what I do I cannot get through the voice training, it either pauses or gets stuck.  Then in the program the copilot does not respond or hear my voice at all.  The problem is not my microphone.  It may be some issue with Windows 10 since Voxatc doesn’t work either...with Voxatc I get the error message “speech training error” “speech training is not configured properly....

 

anybody know now how to fix this? 

The "speech training error" you're getting with VoxAtc is very likely caused by the changes Microsoft made in Windows 10 recently. It no longer accepts files with text that spans longer than 15 lines. We've had to trim our files as well.

As for Fo not hearing you, go to <About> tab in MCE user interface and check what devices are listed under "Sound In" and "Sound Out".

If incorrect, run the "assign sound device" tool.

Start->All Apps->Multi Crew Experience->Assign sound device. Select the correct input and output, click OK and let the toll make changes automatically if required.

If using a USB headset, plug it straight to the back of PC and not via a USB hub.

If it's a conventional headset with 3 mm plugs (jacks), try connecting it to the back of the PC directly to the sound device.

You don't have to do that, but when the built-in audio device is from Realtek (most PCs these days use that), it would be best to un-install Realtek supplied HD audio drivers and let Windows 10 use its own digitally certified drivers. They tend to work better.

 

 

[Phillip Gallagher 101]

Thanks for trying to help, Gerald.  I gave up on it I guess.  I got the microphone to work but My verbal commands etc. are misunderstood 75% of the time.  For example I’ll say “battery on” and a panel might open up or some other random action.  I’m guessing it’s a speech training recognition problem with Windows 10.  There are lots of solutions to this online but none have worked.  No biggie, if this is my biggest problem then things sre pretty good. 😉

Edited September 15, 2018 by Phillip Gallagher
Spelling

[FS++ 710]

1 hour ago, Phillip Gallagher said:

Thanks for trying to help, Gerald.  I gave up on it I guess.  I got the microphone to work but My verbal commands etc. are misunderstood 75% of the time.  For example I’ll say “battery on” and a panel might open up or some other random action.  I’m guessing it’s a speech training recognition problem with Windows 10.  There are lots of solutions to this online but none have worked.  No biggie, if this is my biggest problem then things sre pretty good. 😉

MCE doesn't use the "Dictation" part of the speech engine which will never be 100% reliable.

It uses the more reliable "command and control", and very light on CPU feature of the speech engine. Should expect 85%= accuracy, even with tens of thousands of commands available at any given time.

I believe you need a new speech profile.

The process involves discarding the existing speech profile, creating a new one which knows nothing about your voice, then training it from scratch using our tool.

Before you begin, if you happen to have a Realtek audio device to which your expensive headset is plugged, you may want to let Windows use its own device drivers.

Go to Windows Control Panel, ad-remove programs.

Locate Realtek high definition audio and un-install it.

Restart computer and let Windows re-detect the device and assign it its own digitally signed drivers.

After that, calibrate your microphone via Windows and don't mess with mike settings (things like microphone gain).

Now, need to create a new speech profile.

Go to Windows Control Panel and run <Speech Recognition> applet

On the left pane, select <Advanced speech Options"

Click <New> button and create a brand new speech profile. Give it a name that relates to the current headset, just in case you need to create another one later for a different headset.

The new speech profile should be listed. Select it (tick the box) to make it active.

You now have a new speech profile assigned to your user login account. Just need to train it.

Start->All Apps->Multi Crew Experience->Speech training.

Start with "ATC phraseology" topic as it helps the speech engine work out how you pronounce NATO alphabet and digits.

No need to perform more than one session per topic.

[Phillip Gallagher 101]

Okay we are making progress but why, during the training do I not even have time to finish the sentence and it already switches to the next screen with a new set of phrases to read back?  Either that or it pauses even after I've already read it.  If I've already read the sentence, and the screen doesn't automatically advance in the training, do I read it again?

 

Thanks again,

 

Phil  

[FS++ 710]

9 hours ago, Phillip Gallagher said:

Okay we are making progress but why, during the training do I not even have time to finish the sentence and it already switches to the next screen with a new set of phrases to read back?  Either that or it pauses even after I've already read it.  If I've already read the sentence, and the screen doesn't automatically advance in the training, do I read it again?

 

Thanks again,

 

Phil  

Poor microphone calibration. Speech trainer detecting noise as speech. And obviously will end-up with a useless speech profile if it goes on like this.

 

[Phillip Gallagher 101]

I'm using a HyperX Cloud headphone with microphone attachment.  When I shake the earmuffs I get a "green-bar" registering response on the computer on the testing screen (peter talks to his computer.....) However, if I hold the whole headset still and just talk into the microphone it barely registers.  

[FS++ 710]

2 hours ago, Phillip Gallagher said:

I'm using a HyperX Cloud headphone with microphone attachment.  When I shake the earmuffs I get a "green-bar" registering response on the computer on the testing screen (peter talks to his computer.....) However, if I hold the whole headset still and just talk into the microphone it barely registers.  

Well, sometimes you have to consider getting a new headset.

The good news, any USB wired headset in the US$ 20 range on Amazon.com will do the job. Do not fall for the expensive wireless or even wired propositions.

All it takes is a trip on the wire to damage the microphone (tear the fragile connection). Wireless ones aren't ideal either. They have issues with battery life, interference from other radio based devices etc..

Speech reco in Windows 7 or higher works very well, even for non-native English speakers, at least for the command and control side which MCE relies on.

I don't think the issue is with accent or speech itself. More like an unreliable audio input to me.

I've already gone through 4 headset in 10 years. And most of the time it's a case of microphone connection being de-soldered after tripping on the wire. Never lost the audio out on headphones on any.

 

 

 

[Phillip Gallagher 101]

The $20.00 headset did the trick AND made me fell like a complete word not allowed!  Oh well.  Live and learn!  Thanks!

[FS++ 710]

57 minutes ago, Phillip Gallagher said:

The $20.00 headset did the trick AND made me fell like a complete word not allowed!  Oh well.  Live and learn!  Thanks!

You're welcome.

Can happen to anyone. 

It's hard to imagine a headset you paid a lot of money for, would let you down in this subtle manner. Neither fully working, nor completely broken.

Now keep in mind a speech profile is tied to...

- A user voice

- A specific headset

- Specific set of ambient noise conditions (current room prevailing background noise)

Change any of the above, and technically, would need to create a new speech profile and train it from scratch.

Hope you can finally start enjoying voice recognition for Prepar3D/FSX/FS9 (also available for X-Plane 11 & 10) and recommend it to others, who sometimes for unfortunate reasons like yours when going through the short Demo, end-up with the misguided impression it's not workable.

All critical actions are secured against accidental triggers (FO will ask for confirmation, and keeping silent means No, thanks disregard last instruction)). Therefore, no need to be too careful with what you say, nor worry about the impact on your flight.

We run MCE (and the speech engine it hosts inside "mce.exe") intentionally outside the simulator process for better stability of the latter and to be able to recover from the occasional MCE or speech engine malfunction, not to mention prevent speech engine eating valuable VAS in FSX.

 

 

 

 

[mpw8679 428]

On 9/21/2018 at 4:03 PM, FS++ said:

You're welcome.

Can happen to anyone. 

It's hard to imagine a headset you paid a lot of money for, would let you down in this subtle manner. Neither fully working, nor completely broken.

Now keep in mind a speech profile is tied to...

- A user voice

- A specific headset

- Specific set of ambient noise conditions (current room prevailing background noise)

Change any of the above, and technically, would need to create a new speech profile and train it from scratch.

Hope you can finally start enjoying voice recognition for Prepar3D/FSX/FS9 (also available for X-Plane 11 & 10) and recommend it to others, who sometimes for unfortunate reasons like yours when going through the short Demo, end-up with the misguided impression it's not workable.

All critical actions are secured against accidental triggers (FO will ask for confirmation, and keeping silent means No, thanks disregard last instruction)). Therefore, no need to be too careful with what you say, nor worry about the impact on your flight.

We run MCE (and the speech engine it hosts inside "mce.exe") intentionally outside the simulator process for better stability of the latter and to be able to recover from the occasional MCE or speech engine malfunction, not to mention prevent speech engine eating valuable VAS in FSX.

 

 

 

 

Hi,  I am having same issues as the original poster. I have tried all the troubleshooting tips as listed.  Corona in Windows understands me loud and clear.  I am at a loss.  This is after a new pc build.  Original pc build worked fine.  Issues with both xp and P3d versions.  Thank u,

[FS++ 710]

21 minutes ago, mpw8679 said:

Hi,  I am having same issues as the original poster. I have tried all the troubleshooting tips as listed.  Corona in Windows understands me loud and clear.  I am at a loss.  This is after a new pc build.  Original pc build worked fine.  Issues with both xp and P3d versions.  Thank u,

Shutdown Cortana.

USB headset or conventional one?

 

[mpw8679 428]

Just now, FS++ said:

Shutdown Cortana.

USB headset or conventional one?

 

USB plugged directly into mobo.  I have tried two different headsets.  I will shut down Cortona.  

[FS++ 710]

1 minute ago, mpw8679 said:

USB plugged directly into mobo.  I have tried two different headsets.  I will shut down Cortona.  

To make sure MCE is set to listen on the correct device, run the "assign sound device tool"

Windows menu Start->All Apps->Multi Crew Experience->Assign sound device.

Do the same for the MCE X-Plane edition

[mpw8679 428]

29 minutes ago, FS++ said:

To make sure MCE is set to listen on the correct device, run the "assign sound device tool"

Windows menu Start->All Apps->Multi Crew Experience->Assign sound device.

Do the same for the MCE X-Plane edition

Same result.  Speech training does not register or pauses