microphone input stream

[staminna]

Hi

Why isn't there streaming support?
I am looking at https://github.com/gillesdemey/node-record-lpcm16

Thanks

[Hgergely]

You can do it for yourself. You can start a sub-process in node with
"arecord" and pipe it instead of file input
I have done it. it worked.

On 21 July 2016 at 17:43, Jorge notifications@github.com wrote:

Hi

Why isn't there streaming support?
I am looking at https://github.com/gillesdemey/node-record-lpcm16

Thanks

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#157,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AFyOcZEfv4rKSdUujKVNmZAvHd5q1Hm2ks5qX6GugaJpZM4JR_j9
.

[jmdobry]

@staminna Which sample are you talking about?

[Hgergely]

Iam sure he is talking about streaming.js

On Thu, 21 Jul 2016, 17:51 Jason Dobry, notifications@github.com wrote:

@staminna https://github.com/staminna Which sample are you talking
about?

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#157 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AFyOcXO4d-AD7jYsRJB84kyAPdrcIrNuks5qX6OPgaJpZM4JR_j9
.

[jmdobry]

Then I don't understand the question. speech/recognize_streaming.js is an example of using the Speech API's bi-directional streaming gRPC method. streamingRecognize() returns a duplex stream object.

[Hgergely]

This guy wants an example of how it can be used with a hardware audio input
( microphone) instead of audio file. I understand the question but I think
he must understand : Firstly that this topic is not about all ways how the
api can be used. Secondly this nodejs example is environment independent
and cannot provide a environment dependent solution like microphone usage.
Microphone is used totally different ways depending on the op system under
node.

On Thu, 21 Jul 2016, 18:29 Jason Dobry, notifications@github.com wrote:

Then I don't understand the question. speech/recognize_streaming.js is an
example of using the Speech API's bi-directional streaming gRPC method.
streamingRecognize() returns a duplex stream object.

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#157 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AFyOcSN7NokBw5tyXnD4JlTqQ6rE4dAHks5qX6yBgaJpZM4JR_j9
.

[jmdobry]

Makes sense. I think the one of the biggest takeaways from the sample is that streamingRecognize() returns a duplex stream object, which means that one can do the following (borrowing from the sample):

var grpcStream = speechService.streamingRecognize();
grpcStream.write(streamingRecognizeRequest);

readableStream
  .pipe(toRecognizeRequest)
  .pipe(grpcStream);

where readableStream is any Node.js readable stream, e.g. file stream, http stream, recording stream from node-record-lpcm16, or whatever.

[Hgergely]

Yeas thats what I did. Simply replaced the readablestream with a audio
input stream provided by linux arecord data. It works like charm but node
record lmpcm16 can be an even better choice. Thanks Jason.

On Thu, 21 Jul 2016, 19:53 Jason Dobry, notifications@github.com wrote:

Makes sense. I think the one of the biggest takeaways from the sample that
we have is that streamingRecognize() returns a duplex stream object,
which means that one can do the following (borrowing from the sample):

var grpcStream = speechService.streamingRecognize();grpcStream.write(streamingRecognizeRequest);

readableStream
.pipe(toRecognizeRequest)
.pipe(grpcStream);

where readableStream is any Node.js readable stream, e.g. file stream,
http stream, recording stream from node-record-lpcm16, or whatever.

—
You are receiving this because you commented.

Reply to this email directly, view it on GitHub
#157 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AFyOcU1bNA02ye2ifBD3YvqbzI9q2Nblks5qX8A3gaJpZM4JR_j9
.

[staminna]

@jmdobry Where should this go?
I keep getting grpcStream.write(streamingRecognizeRequest);
ReferenceError: streamingRecognizeRequest is not defined

[jmdobry]

Can you show all of your code?

[staminna]

http://pastebin.com/puXvAr5r

[jmdobry]

@staminna Here's a more concise example you can try. Save it to a file and run it. You don't have to pass any arguments to the file.

// Copyright 2016, Google, Inc.
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//    http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

'use strict';

var async = require('async');
var fs = require('fs');
var path = require('path');
var grpc = require('grpc');
var googleProtoFiles = require('google-proto-files');
var googleAuth = require('google-auto-auth');
var Transform = require('stream').Transform;
var record = require('node-record-lpcm16');

var PROTO_ROOT_DIR = googleProtoFiles('..');
var protoDescriptor = grpc.load({
  root: PROTO_ROOT_DIR,
  file: path.relative(PROTO_ROOT_DIR, googleProtoFiles.speech.v1beta1)
}, 'proto', {
  binaryAsBase64: true,
  convertFieldsToCamelCase: true
});
var speechProto = protoDescriptor.google.cloud.speech.v1beta1;

function getSpeechService (callback) {
  var googleAuthClient = googleAuth({
    scopes: [
      'https://www.googleapis.com/auth/cloud-platform'
    ]
  });

  googleAuthClient.getAuthClient(function (err, authClient) {
    if (err) {
      return callback(err);
    }

    var credentials = grpc.credentials.combineChannelCredentials(
      grpc.credentials.createSsl(),
      grpc.credentials.createFromGoogleCredential(authClient)
    );

    console.log('Loading speech service...');
    var stub = new speechProto.Speech('speech.googleapis.com', credentials);
    return callback(null, stub);
  });
}

async.waterfall([
  function (cb) {
    getSpeechService(cb);
  },
  function sendRequest (speechService, cb) {
    console.log('Analyzing speech...');
    var call = speechService.streamingRecognize()
      .on('error', cb)
      .on('data', function (recognizeResponse) {
        console.log(JSON.stringify(recognizeResponse, null, 2));
      })
      .on('end', cb);

    // Write the initial recognize reqeust
    call.write({
      streamingConfig: {
        config: {
          encoding: 'LINEAR16',
          sampleRate: 16000
        },
        interimResults: true,
        singleUtterance: false
      }
    });

    var toRecognizeRequest = new Transform({ objectMode: true });
    toRecognizeRequest._transform = function (chunk, encoding, done) {
      done(null, {
        audioContent: chunk
      });
    };

    // Start recording
    var readableMicrophoneStream = record.start({
      sampleRate : 16000,
      verbose : true
    });

    // Stream the microphone audio to the Speech API
    readableMicrophoneStream
      .pipe(toRecognizeRequest)
      .pipe(call);
  }
], function (err) {
  if (err) {
    return console.error(err);
  }
  console.log('DONE');
});

[staminna]

I can't seem to make this loop once JSON returns isFinal true. Can someone help me with this? I should be able to continue from there.

[jmdobry]

I'm not sure I understand. Perhaps use recursion to restart your stream?

[staminna]

I had all sorts of problems with recursive calls exceeding memory on using node-async-loop. This example only listens once, and exits node when it's done ( only once). I would like it to keep listening at all times and not exit once is 'DONE'.