Speech API issue

[Chris Wong]

Hi All,

I am building an android mobile application and need to use the API for translate speech to text, but it seems like there have limitation of 50 requests per day. Do we have any way to solve this issue? or use other API for my mobile application without limitation. Thank you!

Chris

[Thiago Farina]

On Wed, May 7, 2014 at 1:59 AM, Chris Wong <onet...@gmail.com> wrote:

Hi All,

I am building an android mobile application and need to use the API for translate speech to text, but it seems like there have limitation of 50 requests per day. Do we have any way to solve this issue? or use other API for my mobile application without limitation. Thank you!

You may have to purchase it. I don't know the details.

I copied some of the guys that I think will know more.

Btw, did you read http://www.chromium.org/developers/how-tos/api-keys?

Thanks,

--

Thiago Farina

[One Taxi]

Dear Thiago,

Thanks for your quick respond. I know how to use it but just concern about only 50 requests per day, that would be great if you can have anyone to get back to me about it, I also would like to know if it require to purchase the service for extra requests per day. thank you for your great help.

Chris

[Primiano Tucci]

Hmm wait. Let's be sure we get the context right.

You are using webkitSpeechRecognition()in a HTML/JS page/site which you access using Chrome for Android?

If this is the case, the situation is a bit different.

On Chrome Desktop, the Speech API implementation (at least the client part) is fully baked in Chrome [1] and hits directly the Google Speech WebService. This means that if you bundle your own version of Chrome, you need an API key as explained here.

On Android, instead, the Speech API implementation [2] is essentially a shim layer towards the Android API (android.speech.RecognitionService). The audio is captured / processed / exchanged entirely by the Android framework. The story here gets a bit more complex. The short version is that AFAIK no API keys should be required on Android, unless:

(a) you're using a custom Android rom, and the rom doens't have the right API keys.

(b very unlikely) The backend that implements (inside the Android rom) the android.speech.RecognitionService is customized (the vendor can replace it) and has some weird limitation like the one you're experiencing AND you are forcing the Web Speech API to use the non-standard backend (i.e. forcing the enable-speech-recognition switch).

Long story:

Android defines the API for speech recognition, hence the contract interface but doesn't necessarily provide a backend implementation. On most Android devices the backend that implements the android.speech.RecognitionService is provided by the Voice Search / Now app. The implementation of web speech inside Chrome for Android looks for that package.

Now, I might miss something at this point, but as far as I know, the Android API shouldn't be subjected on any limitation. Also 50 req/day sounds a very low limit. Can you provide more context?

[1] speech_recognizer_impl.cc, google_streaming_remote_engine.cc

[2] speech_recognizer_impl_android.cc, SpeechRecognition.java

--
--
Chromium Developers mailing list: chromi...@chromium.org
View archives, change email options, or unsubscribe:
http://groups.google.com/a/chromium.org/group/chromium-dev

[Chris Wong]

Hi Primiano, 

Thanks for your reply. Actually we already built a mobile application which require to store the voice clip on the server, and then from the windows server sending out request to google via speech API for translate the speech to text, and then server push both of the voice clip and translated text back to the mobile application, so we can't use the android API to do the speech to text translation as we need to store the voice clip on server and push it back to the mobile device.

it was working fine in last two months but somehow we found out the application doesn't work anymore because we hit the 50 req/day limitation, so we are seeking a way to resolve this issue. we don't know whether google can offer us more quota for the daily request or need to pay for extra request per day. really appreciate if you can share with us any idea about it. thank you very much!

Chris

[Primiano Tucci]

So this has nothing to do with the Web Speech API in chrome?

Am I understanding correctly?

Sounds like you're reinventing Web Speech. I'm confused.

[Chris Wong]

Sorry, I may posted on a wrong group. but do you have any idea about my issue? I don't know where should i post it to. sorry again

Chris

[Torne (Richard Coles)]

If there's no link to purchase more quota, this means it's not available, as far as I know.

This is not intended to be a public API.

[Primiano Tucci]

So, just to be clear, your question has nothing to do with Chrome.

You're just trying to hit the same webservice that Chrome Desktop uses in a product you're developing.

If you log in into the https://console.developers.google.com/ under the APIs page there is an entry for Speech API, which tells you:

- The free quota is "50 req/day"

- "The Speech API allows developers to access Google speech-recognition services. It is only available for development and personal use."

Which essentially means that those APIs are not mean to be used for anything else. They're only meant to be used by Chrome and its developers. period.

[Chris Wong]

thank you! I got it now. 

[Chris Wong]

I got you. Many thanks for all of you help!! thanks again!!

[Arun Nijhawan]

Hey Chris! I am trying to develop a mobile app that does something similar - sends speech to server, converts to text, does some processing, then returns to phone. I'm fairly new to cloud computing and having trouble and would really benefit from your guidance through the steps. (where to store wav file, how to write/call script, where does output go).

I am running a heroku server (and have tried google app engine) in which I am trying to run a python script to take a wave file from storage, send it to google speech, and return the text. Would be happy to chat over phone or email, arun.nij...@gmail.com. Thanks!

[Kajal Chauhan]

hi, Chris 

how much time the api take for audio processing?