Sunday, October 5, 2014

Google Voice Engine tests

The V2 API for google voice translation seems to be using the URL https://www.google.com/speech-api/v2/recognize
The output is having following characteristics 

output : json. xml not supported 
lang: any locale 
key: Developer should get the key from Developer console 
app : optional parameter . passing this returns some additional reruns values 
client : is an optional parameter again, normally application seems can use chorome as value. 

An Example can be given like this: 

First of all, we need to have a recording utility, and that can be obtained using the folioing few commands

Install SOX 
On OS X with homebreew installed, 
brew install sox 

Recording Audio

rec —encoding singed-integer —bits 16 —channels 1 —rate 16000 test.wav

Send the request
curl -X post —data-binary @“audio/hello.wav” —header ‘Content-Type:audio/l16;rate=16000;’ https://www.google.com/speech-api/v2/recognize?output=json&lag=en-us&key=

When google is 100% sure about the translation, the resonse is something like below 

{
   "result":[
      {
         "alternative":[
            {
               "transcript":"good morning Google how are you feeling today"
            }
         ],
         "final":true
      }
   ],
   "result_index":0
}

when google is unsure, the response will be something like below 

{
  "result":[
    {
      "alternative":[
        {
          "transcript":"this is a test",
          "confidence":0.97321892
        },
        {
          "transcript":"this is a test for"
        }
      ],
      "final":true
    }
  ],
  "result_index":0
}

References


https://github.com/gillesdemey/google-speech-v2

No comments:

Post a Comment