Prelude:
Had my eyes on this for quite a while.
Am going to have to walk away quite disappointed as it is extremely rudimentary for what it is.

Honestly I could have made this myself. Heck I'm already half way there.
But let's not digress. On to the details.

The audio Issue:
The software records voice clips as .wav files. And uses them to compare the audio stream by polling it ever few seconds.
To understand how BAD this is. You need to understand that even with cleaning up the audio of these files from static background noise it still has trouble understanding me.
Yet endlessly detected my background static as "Menu" being said.

The reason it does not hear me is it compares small audio clips to the stream. Resulting in RANDOM clips which the dev kindly lets you listen back to.
Saying "Open" was impossible for me as it would attempt detection on either "o" or "pen".
Obviously my audio clip of me saying "Open" does not match with me saying "pen" causing the software to not be confident.

Word Chains:
"You can chain words." However due to the instability of what I just said. It won't be detecting word chains.
I've yet to get it to understand "open map" while "map" would trigger at random times and so does "menu" due to static remotely sounding like it. In fact it seems to hone in on the static as speech despite being barely perceptible in the audio samples.

It does let you use any language as they are clips of you talking which is good.
Setting things up is "easy" which is true but there is absolutely no "advanced" features which could mitigate these issues.

Unless your microphone audio is absolutely studio quality clean I do not see this software in any usable capacity.

Accuracy:
The reason detections of audio are so quick is that the software does not handle audio appropriately.
Every few ms a poll should be taken to attempt to check against the library of audio clips for comparison.
This is slow. But is ACCURATE. Sadly this software does the opposite which is FAST but INACCURATE.

Despite their advertised accuracy. Which is only achievable in a clean audio setup, with a practically no noise room and absolutely no background noise.

HOWEVER

The one use case:
The software does boast a few modes which allow you to play with the detection.
The only useful one seems to be the "Push to talk" mode.

The reason this is the case is as soon as you press it starts to record which lets you avoid the "polling" issue described above.
The accuracy with this PTT mode is VERY good. So if you want a voice recognition software the is extremely accurate and push to talk. This DOES seems to get the job done in that VERY SPECIFIC case.

That being said. Unless your intent IS to have a voice control context button. This software is easily overshadowed by things like voiceattack. However those rely on microsoft voice recognition which honestly should be avoided like the plague.

Conclusion:
Unfortunately I can't say this software is worth more than £1.
I got it on a deal for £9.89 and feel absolutely ripped off as this is extremely CRUDE for what it is advertised.

£44 for standard price... That is scam levels price. This software is absolutely NOT worth that price.
Additionally the price of this software has not come down as far as I can tell so don't bother waiting or hoping.

Also the "Automatic AI Training" which "Trains the AI"?
What A.I.? As far as I can see there is NO data training profile saved anywhere.
This is not an A.I. and if you believe there is. you are incorrect.

It does say "training" when you start capture. But gives NO indication of if you are supposed to talk. Or shut up.
Is it training on your voice? Or capturing a sample of background noise for noise cancelling.

Absolutely nothing is explained beyond help cues and there is no tutorial to indicate how you SHOULD behave during that section.

Make of all this what you will. I hope the DEV will address these concerns from one Dev to Another, this simply... Will not do.

[b]Prelude:[/b] Had my eyes on this for quite a while. Am going to have to walk away quite disappointed as it is extremely rudimentary for what it is. Honestly I could have made this myself. Heck I'm already half way there. But let's not digress. On to the details. [b] The audio Issue: [/b] The software records voice clips as .wav files. And uses them to compare the audio stream by polling it ever few seconds. To understand how BAD this is. You need to understand that even with cleaning up the audio of these files from static background noise it still has trouble understanding me. Yet endlessly detected my background static as "Menu" being said. The reason it does not hear me is it compares small audio clips to the stream. Resulting in RANDOM clips which the dev kindly lets you listen back to. Saying "Open" was impossible for me as it would attempt detection on either "o" or "pen". Obviously my audio clip of me saying "Open" does not match with me saying "pen" causing the software to not be confident. [b] Word Chains: [/b] "You can chain words." However due to the instability of what I just said. It won't be detecting word chains. I've yet to get it to understand "open map" while "map" would trigger at random times and so does "menu" due to static remotely sounding like it. In fact it seems to hone in on the static as speech despite being barely perceptible in the audio samples. It does let you use any language as they are clips of you talking which is good. Setting things up is "easy" which is true but there is absolutely no "advanced" features which could mitigate these issues. Unless your microphone audio is absolutely studio quality clean I do not see this software in any usable capacity. [b] Accuracy: [/b] The reason detections of audio are so quick is that the software does not handle audio appropriately. Every few ms a poll should be taken to attempt to check against the library of audio clips for comparison. This is slow. But is ACCURATE. Sadly this software does the opposite which is FAST but INACCURATE. Despite their advertised accuracy. Which is only achievable in a clean audio setup, with a practically no noise room and absolutely no background noise. HOWEVER [b] The one use case: [/b] The software does boast a few modes which allow you to play with the detection. The only useful one seems to be the "Push to talk" mode. The reason this is the case is as soon as you press it starts to record which lets you avoid the "polling" issue described above. The accuracy with this PTT mode is VERY good. So if you want a voice recognition software the is extremely accurate and push to talk. This DOES seems to get the job done in that VERY SPECIFIC case. That being said. Unless your intent IS to have a voice control context button. This software is easily overshadowed by things like voiceattack. However those rely on microsoft voice recognition which honestly should be avoided like the plague. [b] Conclusion: [/b] Unfortunately I can't say this software is worth more than £1. I got it on a deal for £9.89 and feel absolutely ripped off as this is extremely CRUDE for what it is advertised. £44 for standard price... That is scam levels price. This software is absolutely [b]NOT[/b] worth that price. Additionally the price of this software has not come down as far as I can tell so don't bother waiting or hoping. Also the "Automatic AI Training" which "Trains the AI"? What A.I.? As far as I can see there is NO data training profile saved anywhere. This is not an A.I. and if you believe there is. you are incorrect. It does say "training" when you start capture. But gives NO indication of if you are supposed to talk. Or shut up. Is it training on your voice? Or capturing a sample of background noise for noise cancelling. Absolutely nothing is explained beyond help cues and there is no tutorial to indicate how you SHOULD behave during that section. Make of all this what you will. I hope the DEV will address these concerns from one Dev to Another, this simply... Will not do.