Community Blog Buy Now

Misty Community Forum

Speech Recognition & Database Questions

Hi there, I am very new to Misty.

I have problem of using StartKeyPhraseRecognition in Misty’s API Explorer. It only works once when I switch Misty on. Then it never work again even I issue StopKeyPhraseRecognition. I have to turn Misty off and then switch her on to make it to work again.

The console says “DevTools failed to parse SourceMap: ***”.
Sorry. It does not allow me to include a link in here.

Again, this is not on my machine, it is on Misty’s API Explorer.

Is there a way that I can just leave Misty on as a doorman and have a conversation with anybody who pops into the office? Or do I have to manually start the skill everytime to impress someone?

Many thanks.

Hey, @aria.zhenzhenli, welcome to the community forums, and thanks for posting your question here!

If you just started using your Misty, you may need to apply the Android patch that we released earlier this year. This patch fixes an audio recording bug that matches exactly the behavior you’ve described. You can find a link to download the patch and installation instructions here: 2020.01.07: Misty II Android Patch for Improved SLAM Performance

If you’ve already applied the patch and are still seeing this issue, let us know, and we’ll keep digging for the root cause :slight_smile:

As for having Misty be a doorman who can have a conversation with anyone who comes into your building: that’s a great idea. Misty can stay powered on for as long as you want, and you can set your skills to run for as short or long a time as you need. You can even set a skill to start automatically every time Misty boots up. If you haven’t done so already, consider checking out the Hello World tutorial series. It’s a great introduction to writing skills with Misty’s JavaScript SDK. Or, if you prefer C#, you can read about Misty’s .NET SDK in the developer docs. I’d also recommend checking out the Misty-Concierge-Template, which is an open-source skill template that lets Misty have conversations with people by using Dialogflow (Google’s text-to-speech service) and any other APIs that you want to integrate. (In other words, it’s already pretty close to something that can make Misty a useful doorman :robot: :slight_smile: )

If you have questions about applying the patch or developing for Misty, please feel free to post them here!

Many thanks Johnathan. I’ll try to see if it works.

Best regards for your warm welcome.

Hi Jonathan, thank you very much. I have installed the patch and restarted Misty as you suggested. Misty is working everytime now. Have a great day! Best regards, Aria

1 Like

Thanks for the update @aria.zhenzhenli, I’m glad to hear that applying the patch worked!

Hi Jonathan,

Do you know if Misty can do speech recognition internally? Or, does it have to rely on Internet to send the audio file to a third party for transcription?

I can make speech recognition available on a Misty app (as a web app) using WebKit, but I like it to be a Misty skill running independently to the Internet. Is it possible?

Also, is there a way to store a sqlite database in Misty’s internal memory and run queries on it? Or, does Misty have a database available in her ready to use?

Many thanks.

Best regards,

Misty uses an internal engine for all voice activity detection and speech capture activities, but for speech-to-text and natural language processing, you’ll have to send the audio that you capture to a third-party service.

It is possible to set up a SQLite database in a C# skill. C# skills are background applications for the Universal Windows Platform (UWP) that run on Windows IoT core on Misty’s 410 processor. So in theory, almost anything you can do in a background application, you can do on Misty. We’ve not built any examples of this particular functionality ourselves, but if you’re interested, this documentation from Microsoft looks like a good starting point for using a SQLite database in a UWP app. And check out Misty’s developer documentation for more information about Misty’s .NET SDK.

Misty does have built-in shared data stores that you can read from and write to with JavaScript and C# skills. If you’re using the JavaScript SDK, this is going to be the best option for storing long term data and sharing data across skills. Here is a link to the developer documentation on the subject. However, if you’re only writing/reading data across C# skills, it may be more effective to write your own database.

Thanks a lot Jonathan. That is absolutely awesome.

Best regards,

1 Like

3 posts were split to a new topic: “Speak” command not working