Google-cloud-node: Adding Speech_Context to the Speech API stops Recogniztion

[LibertyBeta]

It seems that the Speech Context config breaks the recognition stream.

Environment details

• OS: *nix(Raspian, Ubuntu, etc)
• Node.js version: 7.5
• npm version:
• google-cloud-node version: 0.6

Steps to reproduce

1. require google-cloud
2. Start Recognition Stream with a config values of
   const request = { config: { encoding: encoding, sampleRate: sampleRate, speech_context: ['one', 'two] }, interimResults: true };
   
   
   
   1. .pipe() audio stream to the Recognition Stream
   
   

[jmuk]

Looks like your code specifying speech_context, but that should be speechContext?
Did you know what happens on the recognition streams? Is there any errors reported?

[stephenplusplus]

@LibertyBeta I believe @jmuk caught the issue with speech_context vs speechContext, but please let us know if this is still an issue.

[arnoldad]

Actually, this does not seem to be resolved (at least in version 0.9.0 from npm)-trying to supply the speech context any which way (see below request config) results in the following error for me:

Interestingly, passing the additional "profanityFilter" config param works fine.

  const request = {
    config: {
      encoding: encoding,
      sampleRateHertz: sampleRateHertz,
      profanityFilter: true,
      // speechContext: {"phrases": ["dog", "cat"]},
      //speech_context: {"phrases": ["dog", "cat"]},
      //speechContext: ["dog", "cat"],
      //speech_context: ["dog", "cat"],
      languageCode: languageCode
    }
  };

Error: .google.cloud.speech.v1.RecognitionConfig#speechContext is not a field: undefined
    at Error (native)
    at MessagePrototype.set (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:2490:35)
    at MessagePrototype.set (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:2483:38)
    at Message (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:2411:34)
    at Element.ProtoBuf.Reflect.ElementPrototype.verifyValue (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:1925:28)
    at T.ProtoBuf.Reflect.FieldPrototype.verifyValue (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:3499:33)
    at MessagePrototype.set (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:2493:59)
    at MessagePrototype.set (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:2483:38)
    at Message (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:2411:34)
    at Element.ProtoBuf.Reflect.ElementPrototype.verifyValue (/Users/arnolda/GitHub/nodejs-docs-samples/speech/node_modules/protobufjs/dist/protobuf.js:1925:28)

[stephenplusplus]

We started using v1 of the Speech API (compared to v1beta1 previously), and it looks like it expects an array for speech_context now: https://github.com/googleapis/googleapis/blob/2b761d5190e3de4764fb58c312286ae043aec02d/google/cloud/speech/v1/cloud_speech.proto#L215

That would mean:

const request = {
  config: {
    encoding: encoding,
    sampleRateHertz, sampleRateHertz,
    profanityFilter: true,
    speechContexts: [
      {
        phrases: ["dog", "cat"]
      }
    ],
    languageCode: languageCode
  }
}

Sorry for the trouble. I opened an issue to fix our links in the docs: https://github.com/GoogleCloudPlatform/google-cloud-node/issues/2219

[arnoldad]

Thanks for the quick reply. I can confirm that the configuration you supplied works as expected with the @google-cloud/speech NPM module version 0.9.0