Android Question New Google Cloud Speech Recognition

Rick Harris

Well-Known Member
Licensed User
Longtime User
Google is currently working on a new (cloud based) speech recognition API: https://cloud.google.com/speech/

This looks very promissing, although it seems Google intends to charge for its use once the definite version is released.
Who can / will port this API to B4A?
 
Last edited:

Rick Harris

Well-Known Member
Licensed User
Longtime User
You can access any online API with HttpUtils2.
Have you examined whether all methods of this API will thus become accessible?

Why are there now no less than 3 libraries in the Forum that access the current Google speech API (by XverhelstX, VPires and recently Stevel05)? Apparently using HttpUtils2 was not sufficient (or maybe I am wrong).
 
Upvote 0

DonManfred

Expert
Licensed User
Longtime User
Upvote 0

Rick Harris

Well-Known Member
Licensed User
Longtime User
Does this help if the code is NOT java?
No idea, you tell me! I have no experience with Github projects (it looks like Java to me)
I never understood Java and am just a "simple" Assembler, VB6 and B4A developer. You apparently do have such knowledge in view of the many wrappers you have made. Nice work by the way.

PS. I am surprised (and somewhat sad) that not more discussion is taking place in this Forum about the latest speech recognition technologies. A lot of new developments are going on like Google Cloud SR API, but also Houndify, Pocketsfinx, MSN activities and many more.
My aim is to extend my elaborate personal assistant app to enable it to listen continuously (like the new "Google Home" box) and with much better speech recognition quality (in particular in the Dutch language).
 
Last edited:
Upvote 0

Rick Harris

Well-Known Member
Licensed User
Longtime User
It does not help.
Pity this Forum does not offer a "Don't like" button... I still don't know what you mean, but suggest that we halt this discussion. If you were interested in further pursuing the New Google Cloud Speech Recognition possibilities, which I am trying to give a push, you would have reacted differently. I can't help it that I can't judge whether the Github project is as useless as you seem to be implying.
 
Upvote 0

DonManfred

Expert
Licensed User
Longtime User
I still don't know what you mean
You have posted the info that a example is available. In the Language "Buffer-Protokoll" which is a native neutral programming language.
In fact; it it not java; so no way to translate it to b4a

That´s the reason why it dont help. Search a Java-Example-App (or java library) for this.
This can be wrapped..

Pity this Forum does not offer a "Don't like" button...
You are right. I´ll put you back onto my ignorelist so i´ll dont answer to your threads.
 
Upvote 0

Rusty

Well-Known Member
Licensed User
Longtime User
Did anyone get this to work?
If so do you have any sample code I can look at?
Thanks
Rusty
 
Upvote 0

JohnC

Expert
Licensed User
Longtime User
Upvote 0

canalrun

Well-Known Member
Licensed User
Longtime User
Since it appears there is a charge to use this API (See the "Resources - Pricing" button), I doubt this api will be very useful to B4x developers because it would then require us to charge our users every time our app uses the API.

I see the original posts from this thread are three years old. Real time speech transcription is a strong interest of mine.

I played with Google's API about a year ago, but at that time they still did not support true real-time recognition, and I was dissuaded by the fee – you have to absorb that, pass it on to the customer, or have the customer create their own account.

I also played with Microsoft's speech API. They also charged a fee.

A problem with true real-time transcription is that you have to continuously send audio to the API or somehow detect start and stop of speech and send only the "busy" portion. Sending continuous speech over the Internet to an API quickly becomes "not real time".

I posted the source, library, and example for something I call CRGoogVR to this forum. You can search for and find it somewhere. It uses the free Android/Google API that's oriented toward non-real-time web searches. I created an app (see the link below in my signature) that uses this library to perform pseudo-real-time recognition.

Google has somewhat recently come out with a free app "Live Transcribe" that does real-time transcription and most likely uses their API. I still like my app better, for several reasons :D.

Barry.
 
Upvote 0
Top