The message states the problem. There's no listening device. That means there's no microphone. Add a microphone to your computer and you'll be up and running.

As for the dictionary - ensure you have the local installed for the desired language.

DJ. He said that speech reco works on the machine, so clearly there is a microphone.

Quote:

The actual Microsoft Speech recognition program is working on the laptop as I can open and close programs, dictate text on Notepad etc.

And a big WELCOME BACK Steve! You have been missed.

Alan

Hi DJ. It's been a while. I hope you're keeping well.

So the problem at hand. Just to confirm and as stated in my first post, there is definitely a microphone. Had there not have been one, I wouldn't have been able to do speech training, set microphone levels or use MS speech recognition control to open/close apps, dictate text, which I mentioned confirmed M/S speech recognition does work and is set up.

I also did mention that I do actually have the correct desired language installed.

Any other thoughts?

Steve,

Please download the plugin: https://synthiam.com/redirect/legacy?table=plugin&id=133

And check the input options to see if a device is available.

Hey Alan.

Just posted at the same time, lol. I hope you're well buddy, and thanks for the warm welcome back.

I'll check out the microphone settings/ properties and see what is what.

Thanks.

@ptp.

Cool, I will do. Thanks mate.

beep.beep.boop.boop.

good to see you back.steve G

Doing super good good to see you back as well!

I should have expanded my answer about no input device. Meaning also verify the microphone is configured as an input device in the audio settings for windows. Some programs allow you to choose an audio input device - while others use the default. ARC does the latter.

Built in windows 8 corona has its own audio input device setting.

Hello @Steve G! Great to have you back!

@DJ.

Hey buddy. Really pleased to hear you're doing good. I have been checking in from time to time to see what's been going on. Really pleased to see lots of new users.

Thanks for expanding on your answer. Alan did mention about verifying about the mic being configured as an input device, so will deffo check on that. Thanks. I've put an order in for a v4/2 dev kit and an ioTiny which are on their way to me now as we speak. Look forward to playing with them. Love the new Robot Program btw. Very cool and informative.

@Jeremie.

Thanks for the warm welcome back. I've missed you guys and pleased to be back. I'll try out the advice given tomorrow and post my results.

@nomad.

Boot beep bop bleep. Good to see you're still around buddy.

Thanks for the robot program feedback! It's a huge initiative but seems to be slowly catching on. I'm really enjoying participating in the episodes. Professor e is great!

Hi Steve! Good to see you back and building. I hope everything is doing better with you and your family then was last time we chatted. I know you had some dark days back then.

Hopefully the suggestions you've been given by the guys will set you in the right direction. I bet you'll find your microphone issues Within the recording settings section of windows like was mentioned. Good luck.

Steve G, Glad to see your back on the forum, I have always enjoyed your work. Steve S

Hi Dave.

Good to hear from you. Yeah it was a tough time back then, but I'm doing pretty well now. Settled in to my new house by the coast now and very happy here. I figured now would be a good time to blow the dust off my tools and get busy with robots again, lol.

Thanks Steve. It's nice to be back and good to see you're still around as I've always enjoyed your work too.

So back to my problems with S/R.

At the moment I have three recording devices... "Microphone", "Line in" and "Stereo Mix" on my laptops sound menu. I have disabled line in and stereo mix and only have microphone enabled. Still no luck. So here's what I've tried so far for UK settings...

IN WINDOWS: Set language to Microsoft Speech Recognizer 8.0 for Windows (English - UK) Set up microphone, Configured microphone, Trained computer to better understand you (both courses), Under "Advanced" selected "Use this audio input device". Opened and tested Microsoft Speech Recognition which successfully opened/closed programs and dictated text to notepad and Word.

IN ARC: Opened new ARC project and added S/R control. Opened config menu, reduce confidence to 50%, setup microphone, selected en-GB, saved settings. When config menu closes, I still get the debug message...

Quote:

Set Dictionary Error loading phrases (custom): The Language for the grammar does not match the language of the speech recognizer.

Quote:

Speech Recognition Culture: en -US Loaded speech recognition successfully

One other thing I tried was to go to the Add Control... Audio and add the Microphone control and Sound servo (PC Mic) control to the new project. Unlike the S/R control, both of these controls are indeed picking up audio input from the laptops microphone, as seen on the controls spectrum analyser's and audio level readout seen in the photo below...

So I believe that from what I've outlined above from the testing I did this morning, I have established that this is not a microphone issue, but rather a language issue (although I'm not sure why the S/R control is not picking up any audio). I'm out of ideas now of what to try next. Any thoughts guys?

Thanks.

ADDITIONAL:

I just tried adding the Pandorabot control as this uses speech recognition as well. Using text input worked and Pandorabot responded. However using speech also did not work with no visual display on this controls spectrum analyser showing. I don't know if this means anything, but i have noticed that both the spectrum analysers on the speech recognition and Pandorabot controls are white (like they are greyed out, so to speak) where they used to be black.

Like a said, I'm not sure if this means anything, or if this was a simple cosmetic change from a black background to a white one done in an update. Figured i would cover all the bases... just in case.

Okay, a bit of an update to my previous post.

In an aim to dial in the problems I am having, I have just tried the "Bing Search Test" project in ARC and Bing speech recognition plugin, and had success with that. I verbally asked a few questions to the Pandorabot in the project, speech recognition was successfully picked up, and answers were returned. I also noticed that the debug window on this project says that the speech recognition culture is set to en-US.

So just need to figure out whats going on using the ARC speech recognition control and my laptop/copy of Windows 10.

@SteveG

Windows 10 has a new Speech API ("code name" One Core) and introduced the UWP (universal windows platform) a new framework to develop universal applications only available via Microsoft Store.

ARC is a desktop app and uses the legacy Speech API (SAPI).

Please check your SAPI settings:

open windows run:
"%windir%\SysWOW64\Speech\SpeechUX\sapi.cpl"

verify and adjust the speech recognition and tts settings.

Wow, how frustrating for you. Looks like you've tried everything. Have you considered scrapping the S/R control and going full Bing? It's a much better S/R platform. However you need to always be connected to the internet and that limits you traveling with the robot away from your wifi signal.

I'm no expert but I have had my share of sound issues I've worked through. Lets look at a few things. What follows is my personal "Idiot's Guide" to my personal sound issues :

Are you using an external USB mic one plugged into the Mic jack or the laptop mic? I've found that when I use a USB mic that Window's sets it up and I need to use that one as my default recording device.

When you're in Windows recording settings window and you see the several devices you mention, you see a sound vertical sound level graph to the right of the default device. When you speak or tap the mic do you see the level jump and fall? If so that's the one you are actually using and it's working (forgive if you already know this). If you don't see that rise and fall , single click on the device to highlight it then click on it's Properties. The on this next window you will see tabs at the top. Sometime, I use the listen tab to actually hear what is being heard from the mic through headphones or the computer's speaker (depending on what I choose). You need to click on "Listen to this Device", choose the playback device then click Apply at the bottom. When you done you'll need to reverse this process to stop hearing yourself when you speak. Next look under the Level tab. Raise your Microphone and Microphone boost up (maybe all the way?) and see if that helps. Also make sure the little speaker icon is not callicles and muting your microphone.

Also try using each recording device listed with your mic and raise the levels as mentioned above . With me I've found I need to use a different device listed other than the one I thought it would be.

In the distant past of EZ Robot's people were having problems using S/R with other languages other than the US setting. I've never had to do this so I may be wrong on the process; Did you try changing your whole Windows OP through Windows control Panel / Language or just Window's S/R control? Maybe you need to set your entire Window's OP platform to the English/US and the Window's S/R control.

Lastly, try deleting the ARC S/R control and reinstalling it.

Good luck and have fun.

@ptp

Thanks for the advice. When I start "Run" and copy/paste the info you supplied in your post, I either hit "OK" which opens speech properties, or click "Browse" which gives a list of four files but not sure what to do with them.

@Dave.

Thanks for the post buddy. Yeah I make you right about the Bing setup which does work pretty well, but would like to solve the native S/R control issue so I don't have to be online. So in regards to your "things to try" list...

I am using then on-board mic at the moment. It's a good mic, (not ideal for speech rec, I know, but okay for testing), but everything I have tried that works... works really well.

Yes the green graph goes up and down very well, so mic gain setting are okay there. I have found that when I adjust mic levels and save them, it sometimes reverts back to the previous setting? Rather annoying, but do remember it doing this on an older laptop with Windows 8.1 and 10, so levels is something I check on often.

I have tried the device options I have, with different combinations of what's enabled and what's not. No joy

I have just tried setting the system to US, but has not made any difference I'm afraid. But saying that, using "Category" in "Control Panel", then "Clock, Language, Region", I noticed that the language preference for "English, United Kingdom" was available for download and not enabled. I downloaded it, set everything back to UK, opened a new EZ project, added the speech rec control, set it to en-UK, saved it and the debug message now says...

Quote:

Speech Recognition Culture: en -UK Loaded speech recognition successfully

Quote:

@ptp

Thanks for the advice. When I start "Run" and copy/paste the info you supplied in your post, I either hit "OK" which opens speech properties, or click "Browse" which gives a list of four files but not sure what to do with them.

You will want to run it and go into the speech properties and test speech reco from there. I believe what @ptp was saying is that this is a different speech engine than the one used by Cortana or some other Windows 10 apps, so needs to be set up independently.

Alan

Oh, that's what I have been doing already but assessed the Speech properties via the control panel, speech recognition, advanced settings.

I just wonder if there's a package I need to download/instal?

Taking some time out now as I've been on this for most of the day and take another crack at it tomorrow, lol. Anyway date night tonight so I'm taking my girl to watch the new Bladerunner movie.

Any more thoughts or suggestions... keep them coming guys.

Thanks.

OK, nucular option; uninstall ARC, download and reinstall the latest version. You may have something broken in the background of ARC. Or it may reconnect something that was broken between Windows and ARC. Couldn't hurt.

@Dave.

Thanks buddy, I will give that a try. It was a fresh instal when I first started with this a couple of days ago, the the recent update. But I will wipe it all and try it again.

Btw, I enjoyed Bladerunner 2049. Anyone who hasn't seen it and is a fan of the original, I recommend you give it a go.

Since you mentioned Windows Speech Recognition working, I did a little experimenting today to see if leaving it running might be causing a conflict. Doesn't seem to be the case.

I am at a bit of a loss on this one.

As is common with these kinds of things, it will probably wind up being something really simple we are all missing. Doesn't appear to be an ARC bug, but temporarily changing your settings to US or Canadian English from UK English might be worth a try.

Alan

Oh, and unrelated. Thanks for the recommendation about the movie. You are the second person I know who has seen and recommended it. I will probably try to go sometime this week. I recorded the original on my DVR from SyFi last night since it has been a long time since I have seen it. My wife remembers not liking it, so I may need to go alone which will be difficult this week since her birthday is Tuesday.

Alan

Happy Birthday Mrs. Alan!

Hey guys.

So I've just completely uninstalled ARC, restarted laptop and re-installed ARC again. Added speech rec control... still not working.

On the other hand, I have had a play with a new Acer W3 running Windows 8.1 which arrived this morning. I had to install .Net4.6.1 to download ARC, opened it up, added a speech rec control, said a couple of phrases and it was recognised and responded straight away using the on-board mic.

Not a fix, obviously, but a workaround for now. I really am stumped as to why ARCs speech recognition and my laptop are still not playing nice with each other after all I've tried... even disabling antivirus just in case. Rather frustrating. Like you said Alan, certainly not an ARC bug.

So I want to briefly focus on one thing for suggestions. Using the W3 tablet, the visual representation of speech being picked up (spectrum analyser) on the top right of the speech rec control is indeed black with the green wave line showing audio pickup. So curious to why it is white with no wave line on my laptop. This must mean something (other than it's not working, lol).

i always delete my download history in google settings. also see if there's are no files left behind when removing ARC, on other places on the pc. then restart pc and download again.

Thanks Patrick. I did check that all files were deleted, but I'll check again just in case I missed something.

you can also check for updates.

Okay, double checked that everything EZ-Robot related was deleted and removed and did an uninstall/reinstall on the laptop. Still nothing.

Below is a screen grab of a new project on the laptop and tablet, both with speech recognition and Pandorabot controls added. While saying numbers out loud, you can see the difference between both devices (laptop screen on top of screen grab, and tablet in the lower part of the grab)...

Laptop not working, tablet working perfectly.

i dont use this,but here is mine . can you try with an open excample jd project?to see if it works?

do you change the name off the download? the file can get corrupt , if you change the name.and not working good.

Well, sorry the Nuck option didn't work. It was worth a try.

I see in your screen grab that the spectrum analyser in the S/R control is now black and showing your voice as you speak. That means that that control is now hearing your voice. Are all your error messages gone now? What exactly are you trying to have the S/R control do?

Perhaps now your issue is something with what you want the control to do. Look inside the control's settings (send a screen grab if you can) and look to see if you have anything in there that would keep it from running like something in the enable phrase command box. Look at the script you (should) have loaded to see it that is written properly. Even so, if you have a bad script you should see a response in the debug window of the S/R control.

Also (sorry, I lost track) did you try changing the language setting inside this ARC S/R control to match the Windows and Window's S/R language setting? Perhaps you can change all three to the English US language and try that if you haven't tried this already.

Thanks Patrick.

There was no need to change the name as I have repeatedly been using new projects so there's nothing there to be curupt.

On another note, I've just installed ARC onto my desktop PC running Windows 10. Started a new project, added speech recognition control... and it works. So it's deffo the laptop that's throwing up these issues and something somewhere in the system that's not talking (pardon the pun) to ARC as it should. Just wish I knew what and where to look next.

maybe the speech engine from the laptop itself ?

it seems like there is a component of SAPI 5.4 that is missing (As PTP pointed out, Windows built in speech reco uses One Core, not SAPI, so it works).

Unfortunately, it is supposed to be installed by default in Windows 10 so doesn't seem to be available as a separate download. I have been searching for sample applications you could try in order to validate but not having much luck (also thinking that installing anothr application that uses SAPI 5.4 could potentially install any missing component, or at least provide a verbose error message telling you what is missing). There is a page on MSDN that describes some sample apps, but no links to download: https://msdn.microsoft.com/en-us/library/ee125085(v=vs.85).aspx I think they might be part of the Microsoft Windows SDK but I haven't been able to verify that.

Alan

Hi Dave.

You are right, it deffo was worth a try.

The screen grab in post #32 is showing my laptops screen (behind and top of the grab) which is showing white analysers, and my W3 tablet (sitting in front) showing that's working with the black analysers (lower part of the grab). Yeah I did try changing the language in the config menu a few times now, along with system language.

When I'm doing these tests, I'm just using the preset phrases already populated in the S/R control and I use the "Robot, show phrase list" phrase. I have also tried a couple of simple "print" scripts too.

Hey Alan.

Many thanks for looking in to it. Much appreciated. After all the elimination tests I've carried out, I think you and ptp maybe right about a component missing.

I wonder if it's worth backing everything up and re-installing Windows 10 on the laptop again? Not something I really want to do, but might be worth it.

It might be worth it, but I would wait for PTP to chime in again first. He knows this stuff pretty well and might have some ideas.

A note about your W3. The W3 can be upgraded to Wndows 10, but at this time, the latest "Creator's update" is not compatible with the CPU. Supposedly Acer and Microsoft are working on a solution because Intel is not interested in providing drivers.

Alan

Thanks for the info about the W3 Alan. Lol, yeah I did find that out yesterday when I tried to do the Windows 10 upgrade. Interesting to know why it didn't work though. Still, I'm quite happy to use 8.1 on it for now.

I'll take you advice and hold off on the laptop upgrade for now in case ptp comes back to me.

Thanks.

At this point I'd back up everything get copies of all your wanted installed programs, passwords and licenses and reinstall win 10. However that's a lot of work and you're going to have to reinstall all your apps and reset them up. Maybe the other "experts" have a better solution. Maybe figure out what is missing and install a different program that has the same component that's missing in your system. Just a thought. I'm now out of ideas. good luck and I'll be watching. This may help me someday. tired

EDIT: looks like I'm late to the party again. You guys are ahead of me as usual. lol.

@Dave.

Late to the party alright, but if you've bought some booze with ya, it's all good buddy, lol.

Seriously though, I am still kinda considering a full Windows re-instal. It wouldn't be the first one I've done so I know what to expect/do and have a large portable drive ready use for the backup. It's just so time consuming as you know yourself for something that may or may not fix the problem. Anyway, watch this space...

I am sure you have thought of this, but if you do the re-install, install and test ARC before any other applications to rule out some incompatibility. For instance, there is a known issue with GoPro software and virtually all other video software including ARC. Could be something relevant that is breaking Sapi.

Alan

Thanks Alan, good point. Thanks for the heads up.

@Steve G,

This is a snippet of my (control panel/SAPI) Speech settings:

I've highlighted the sound level, when you speak check if the sound level is showing up, if not check the microphone levels.

If still not working, go to windows 10 speech recognition settings, and start the setup from the begin.

recheck (SAPI) Speech settings.

if still not working please post a print screen of your SAPI speech settings.

@ptp

Thanks for getting back to me buddy.

I've had the Windows speech properties open a quite a few times during my tests to double check mic levels, and where I've been changing the language repeatedly from between English-US and English-UK. The visual sound level that you mentioned in "speech properties" does bounce up and down when I speak. I will do a screengrab of this tomorrow.

When you say "post a print screen of your SAPI speech settings" (unless you mean do a screen grab of "speech Properties" ), can you give me a quick "how to" of exactly what you mean or want to see to get to the SAPI speech settings, just so I (and any new commers) know we're on the same page of what to look for. Thanks buddy, and much appreciated.

post a pic similar to mine.

I used the windows program "snipping tool".

@ptp

Here's the screen grab of the speech properties you asked for. I said a few words when I took the grab to show the mic levels showing that the mic is indeed picking up audio...

@Steve G

Please uncheck the option "Review documents and mail to improve accuracy" Reboot the PC and check ARC.

Made no difference...

Boing!

It was not a random guess .. those three options are related to the Shared Recognizer... so my first guess was a "resource conflict" between application and system... although it's called "shared..."

btw

"Review documents and mail to improve accuracy. "

When this is checked, the speech recognition system will look at your documents and look at your mail and check the vocabulary that’s there, and phrases and words that it may not know, or it will find ones that you know and probably use frequently. And when you do that, the speech recognition system will become more accurate over time because it’ll know what you talk about and what you’re reading.

You may or may not want to do this because of privacy concerns...

1. Did you installed any Application with (Speech Capabilities: TTS/Speech Recognition)?

2. SAPI Custom Voices ?

3. Is Windows 10 Cortana working ?

@ptp.

Lol, I didn't think that was a random guess.

So, to answer your questions...

1. No, I haven't installed any applications that has/uses Speech or TTS (except ARC of course)

2. I also have not installed any custom voices

3. Windows 10 Cortana is working.

Also, thanks for the info about the ""Reviewing Documents..." option. I didn't actually know that it look's at your documents and look at your mail and check's the vocabulary. Interesting.

hi steve

i see by ptp two voices in settingd.while yours has only one,your name? chould there not the name in there from the voice you use?

@Nomad 6R, That is recognition profile, you can have multiple trained and you select one. I'm almost sure that is not the issue.

@Steve G: I think i found the issue, so when you are online ping back the thread. I'll provide more details.

@ptp.

Pinging you back. Now I'm intrigued. If you're about, I'll be online for the next hour.

Steve.

@Steve G,

1) Do you have both the UK and the US speech recognizer on your drop down ?

2. If yes:

select US recognizer, Jump to point 4)

3. if no:

Goto Region & language settings

Note: I'm doing the opposite (Adding English / GB language) you should add the ( English / US)

Then add the speech voice files (although you can add everything else like hand writing, in the US we use a different hand writing ), yes we are different

this is a result with multiple english languages (although the default is UK)

4. Reboot

5. Call ARC and test the speech recognition culture selection: en-US, en-GB. Both should work.

Let me know the results.

@ptp.

Thanks for your reply. There's quite a few posts now on this thread, but just wanted to refer back to posts #1 and #19 with a bit of a recap to answer your question.

In the speech recognizer drop down, I do have both English US and English UK. I have tried both available languages using different configurations using en-GB and en-US in ARC, non of which worked (even after reboots).

The only difference I had (post #19) was that English UK was not enabled in the Reagon and Language settings. When I enabled this, the debug message changed from...

Set Dictionary Error loading phrases (custom): The Language for the grammar does not match the language of the speech recognizer.

to...

Speech Recognition Culture: en -UK Loaded speech recognition successfully

but speech recognition still did not/doesn't work and the spectrum analysers in both the speech rec and Pandorabot controls being white in colour... not black with an active green wave line confirming that speech (or any noise) is heard.

This was the same result with multiple...

Complete removal's & re-installs of ARC, Removing and adding S/R controls, Using new empty projects, Laptop restarts, Microphone setups, Speech training etc.

@Steve G,

Believe me we are close...

I'll need to take a few screenshots to match the correct configuration (combination).

There's nothing wrong in your windows, neither with your ARC setup.

This is a software problem related to the way ARC handles the culture when calling the SAPI API.

I'll be back to my batcave in 1 hour, ping the thread when you are available.

Quote:

A note about your W3. The W3 can be upgraded to Wndows 10, but at this time, the latest "Creator's update" is not compatible with the CPU. Supposedly Acer and Microsoft are working on a solution because Intel is not interested in providing drivers.

Back to slightly off topic. It looks like MS and Acer solved the driver issue. My W3 just updated to Creator's Edition, although so far it is running like cr-p. I think it is time to start looking for a faster/better tablet or micro-PC. Having some trouble finding anything affordable that does what I need. Acer doesn't make any windows tablets now, and most of the Chinese ones on Gearbest have a single OTG USB port for charging and for USB, so I can't keep it on external charger and use any USB devices.....

Alan

I've had my w3 updated to 10 since last year. It's currently the computer running ezrobot's bimbo the clown amusement game in our arcade area. But - it's super slow. Once ARC is loaded, it's acceptable. Takes like 4 minutes to boot and load though...

Quote:

I've had my w3 updated to 10 since last year

Yeah, Windows 10 update was fine. It is the creator's edition that failed. Intel didn't produce drivers for the chipset. Microsoft and Acer collaborated to create them, so the update is working now.

But it does seem to be even slower now. Maybe just spoiled because I got a new work laptop that is super fast so everything else seems slow.

Alan