Choosing a speech recognizer

How do choose between the available speech recognizers?

"io.spokestack.spokestack.google.GoogleSpeechRecognizer" // one of the three supported speech recognition services
// 'io.spokestack.spokestack.microsoft.AzureSpeechRecognizer'
// 'io.spokestack.spokestack.android.AndroidSpeechRecognizer'

Does it matter which one I select?

Hi ashankar,

Great question! Aside from qualitative differences in speech recognition accuracy between the three options, a key consideration is that the AndroidSpeechRecognizer is both free and on-device, while the other two require both an internet connection and API key for their respective services. For a detailed discussion of each speech recognizer, check out our ASR configuration tutorial.

Thanks for using Spokestack,
—Noel

Hello!
I’m new to spokestack, and have implemented a Spokestacktray object which works. I think it automatically uses the AndroidSpeechRecognizer because most things I say are understood in my native language.

I would like to try the spokestack ASR, I do not fully understand where to change that in my code (maybe a configuration in the object itself?)

Thanks for any help,
Lirry

Hi Lirry, thanks for using Spokestack!

Can you tell me which platform you’re using?

For Android, here are your options, and how to use them.

For iOS, [here are your options](https://spokestack.github.io/spokestack-ios/Classes/SpeechPipeline.html, and how to use them.

For React Native, check out the “Configuring Spokestack” section of our Getting Started guide.

Hope this helps!

—Noel

1 Like

Thanks for the fast reply!

I am using react native. The problem is that I followed the Tutorial for setting up the SpokestackTray from your blog, and this looks different then the initialize function explained in the Getting started guide…

Due to the vagaries in building things with cross-platform native support, we currently only allow for the native OS ASR (AndroidSpeechRecognizer/AppleSpeechRecognizer) to be used in Spokestack Tray. We will prioritize putting an option for our own Spokestack ASR into the configuration for a near-future release though! Thanks for asking.

—Noel