Jason Carr

One really odd thing that I've noticed with my laptop is that it seems to want to automatically "fix" volume levels, and completely disregards any changes I make. I'm wondering if that has something to do with the issue as well.

Alright, new beta is out. This should at least tell us what volume levels (if any) are coming into LaunchBox. There should be a new box that shows up at the top of the screen that will give you an "audio level" of the microphone as you speak. This will at least confirm that LaunchBox is receiving microphone input and at what level. Please let me know what the numbers max out at as you speak. Also test with the accuracy slider back in the center. I also fixed the Options dialog to stop trying to reset everything when you save the options. So after enabling or disabling voice recognition, a restart will be required before the changes take effect. There shouldn't be any more performance issues when closing the Options dialog. Hopefully this version will at least help us troubleshoot; let me know what you guys see. Thanks!

Thanks @smallsbig. That helps. Gonna force a restart after changing the settings.

Just out of curiosity, @shinra358, try turning your volume down on your microphone. I'm curious if we're having volume leveling issues. Try and see if you can get the green bars to fill half way up as you speak. It may be that I just need to add some controls for auto-leveling the volume of the microphone.

Thanks guys. Make sure you have the right microphone selected as your default microphone in your audio settings. I'm going to have to add some settings to let you choose and test your microphone because this seems to be a common problem so we'll at least need a way to figure out what's going on. Also, @donarumo, that is very helpful information. I'm going to make it so you have to restart LaunchBox in order to enable voice recogntion since it seems to be problematic time-wise in initializing it all post-startup.

What microphone are you using @shinra358?

Awesome, sounds good @smallsbig.

@Mute64 That error was happening because of the machine not having a microphone attached. However, it should be fixed in beta 11. @polymorph77 Most likely the 30% thing is just because it's only using one CPU core for the voice recognition initialization. 14 minutes obviously still isn't acceptable so sounds like it's back to the drawing board. Thanks though.

Yes, sorry guys, working on some critical bugs with the voice recognition. Apparently it is extremely CPU intensive and is also taking forever to load on some systems. Any performance or microphone testing you guys can do with the latest beta (beta 11) would be appreciated. Please see this thread: Also, keep in mind that microphone quality apparently matters a whole lot. For example, the built-in microphone on my Dell XPS 13 is absolutely useless and won't work at all for the recognition. However, of course my recording microphones work perfectly. I'm going to test with some more mics to see how it fares.

Ah, sure. Probably because the laptop microphone quality varies as well. That makes sense.

FYI, even though this message suggests that it's writing to the folder, it's not. It's just trying to read the folder only in order to detect the applications that you have installed for the voice recognition. Still working on a resolution though.

Alright, it might be 5 minutes or so for the new update to register, but I just put out another beta guys. Let me know if this does anything to resolve the performance issues if you would. I am noticing some huge performance differences between different CPUs, which is somewhat surprising. This beta loads the voice recognition in less than 2 seconds on my i7-4770K but it takes around 8 seconds on my laptop's i5-5200U. So apparently it's very CPU intensive and requires some monster hardware. Let me know though if this latest beta makes a significant difference for you guys.

Yeah, I'm doing some more testing as well and sadly I'm finding that the built-in microphone on my Dell XPS 13 is pretty much useless for it. So microphone quality apparently matters a lot. I get great results with my recording microphones but I can't get the laptop microphone to work hardly at all.

So odd. Sigh. So it's taking 16 minutes to load on your system and 5 seconds to load on mine, using the same XML. There's some kind of underlying issue there that is going to be really hard to troubleshoot. At that rate nothing I do will be performant, so we're gonna have to figure out what on earth the problem is. I might have to come up with a debug version or something.

I have used it in the past without issues, yes, but it is third party software so unfortunately I can't guarantee it either.

Ah, sure. That is possible but I need to better document it because it is a bit complicated. Will try and get to this next week.

Thank you @ckp! The top or middle items should be safe to use, just avoid the bottom "replace all metadata and images" option.

Sadly I don't think there's a good way to do this outside of using a custom theme or removing all fanart from your games. I will add this to my list, though @itzafugazi.

Hi Rolando, these issues will not happen with Windows Media Player, but there are separate settings for the video playback engine in Big Box than in LaunchBox, which I expect is what you're running into. Make sure Big Box is set to Windows Media Player specifically.

Hi @bran479, what are you seeing happen in Big Box?

Thank you @Overload86. I am working on this.

Thanks @Wutex. Unfortunately the method we're using just waits for the initial process to exit, which isn't fail-proof. Sometimes emulators will launch one process from another, which is very difficult to detect. I don't have an immediate way to solve this problem, but I'd appreciate if you could report it using Issues/Requests in the header.

Keep in mind though that I'm doubtful that this will make much of a difference for @smallsbig's 50,000 games. Curious to know if it helps any, though.

It may be quite a challenge to get 50,000 games working well with voice recognition. To be honest, it might not be possible. That said, we do have some options like only making it work for visible games and such, so there are some solutions, but we might have to rethink some things. I just tested with polymorph's XML (15,000 games). It was taking something like 30 seconds to load the voice recognition stuff, which obviously is no good. I switched it up to only recognize full game titles for now, and it now loads up on my machine in about 4 seconds. There's a new beta out with the update that should at least make it usable. I'll be doing more work soon to see how to improve it further.

This is being worked on by @higan, ckp. If videos are not playing in WMP, it's relatively easy to solve this issue by installing some codecs such as the K-Lite codec pack.