[Release] Google Home Assistant Relay Driver (v1 and v2)- text to speech on Google Home devices

They worked fine for me and many others at the time when they were written. Assuming you're using the exact same versions of Raspian, NPM, etc... that I documented, it should still work fine. Did you start with a clean install of Raspian? Or are you using an existing RPi that might have had other stuff installed on it already?

I haven't tried rebuilding one of these in quite some time, though. So things could be out of date.

Yes, I tried on an existing RPi with other stuff. Hate to go out and get another just for this but maybe that's the only way. Get stuck getting these kind of errors when I do the NPM Install:
> grpc@1.8.0 install /home/pi/assistant-relay/node_modules/grpc

> node-pre-gyp install --fallback-to-build --library=static_library

node-pre-gyp ERR! Tried to download(403): https://storage.googleapis.com/grpc-precompiled-binaries/node/grpc/v1.8.0/node-v72-linux-arm-glibc.tar.gz

node-pre-gyp ERR! Pre-built binaries not found for grpc@1.8.0 and node@12.1.0 (node-v72 ABI, glibc) (falling back to source compile with node-gyp)

node-pre-gyp ERR! Tried to download(undefined): https://storage.googleapis.com/grpc-precompiled-binaries/node/grpc/v1.8.0/node-v72-linux-arm-glibc.tar.gz

node-pre-gyp ERR! Pre-built binaries not found for grpc@1.8.0 and node@12.1.0 (node-v72 ABI, glibc) (falling back to source compile with node-gyp)

Then it only gets worse.

Let me take a look at all the versions . I though I updated everything to the latest but maybe there's something off.

Node has to be at v 8.x. Newer versions do not work.

Instead of a new RPi, you could just start with a new microSD card to see if a nice clean install gets things up and runnning. Then figure out how to get the old stuff running again on the new card, or if you want another RPi altogether.

That would explain it. I went to far up on the Node version. Which version of Node do you use? I must have missed it in the installation instructions. :frowning:

My bad.

Node 8.x... - per my MS Word Document

curl  -sL https://deb.nodesource.com/setup_8.x | sudo -E bash -

sudo apt-get install -y nodejs

THanks. I jumped to so many installs and readme on github I completely missed the word doc you have. Few things had to update but it's working now.

Also as still weak in RPi, had to figure out how to downgrade haha. Anyways, got it :slight_smile: I'll try it out now. Thanks a bunch!

1 Like

By the way, is there a document to describe what commands you can use with [CC]? It works to turn off/on switches, turn volume up/down on the google device, got that from the search. However can it do everything as if we talked to it? For example play music on a chrome cast?

tried looking around but didn't see anyone talk about that. Haven't slept in awhile though :slight_smile:

Some CC commands work, others do not. For example, IIRC, issuing a command to "Set Volume on Home Hub to 50%" probably will not work. But if you use a Google Home Mini and issue the same voice command, your Home Hub (assuming you have one) will change its volume.

So, it's hit and miss.

That's too bad. Would be great if Google made it 100% work as if you spoke it. I even tried to see if it would play a Google Home routine which would allow you to customize (it did not).
So far only benefit I see of the [CC] function would be if Google Home has an integration that Hubitat does not have yet.

Was really hoping to play music/spotify type thing with [CC]. Back to the drawing board looking for a good (non-Sonos) spotify URI player

Have you tried using the built-in Chromecast Integration?

Yes, I have that installed as well. (also tried to port in the cast-web, work in progress).

For the Chromecast integration, it works for TTS on my Minis, Display, and chromecasts. TTS does not work on my Google Home for some reason. :stuck_out_tongue:
However not sure how to play a Spotify URL. I'll play around with it. I assume spotify URI will not work though.
Also tried Logitech Media, not working.

If all else fails, try Echo and Alexa devices/tabs :sweat_smile:

But for this integration (Google Home Relay), might prove handy for those hard to intergrate devices that Google assistant has.

I skimmed through this thread and haven’t been able to find and answer to my question. Does assistant relay allow for sending a text notification to google assistant or does it only speak messages through on of their speakers?

Sure here is the node server. Just backup your assistant.js file and replace with this. Fair warning I take no responsibility if something breaks. Towards the end of the file are a few lines you need to modify.

var heUri = '/apps/api/2832/trigger/setGlobalVariable=test2:' + gaResponse + '?access_token=461f117c-95ad-4d52-8136-12345678789'

this line needs to be modified with your accesstoken and the name of the global variable you want to set. Also the app id for your RM trigger.

Also just below this line you need to set the ip of your hubitat hub.

Let me know how you get on :slight_smile:

const async = require('async');
const GoogleAssistant = require('google-assistant');
const FileReader = require('wav').Reader;
const FileWriter = require('wav').FileWriter;
//const wavFileInfo = require('wav-file-info');

const terminalImage = require('terminal-image');
const path = require('path');
const fs = require("fs");
var tou8 = require('buffer-to-uint8array');

var http = require("http");

const defaultAudio = false;
let returnAudio;
let gConfig;

var playbackWriter = new FileWriter('server/playback.wav', {
  sampleRate: 24000,
  channels: 1

const inputFiles = [
  `${path.resolve(__dirname, 'broadcast.wav')}`,
  `${path.resolve(__dirname, 'response.wav')}`

var self = module.exports = {
  setupAssistant: function() {
    return new Promise((resolve, reject) => {
      let users = []
      async.forEachOfLimit(global.config.users, 1, function(i, k, cb){
        let auth = i;
        global.config.assistants[k] = new GoogleAssistant(i)
        let assistant = global.config.assistants[k];
        assistant.on('ready', () => cb());
        assistant.on('error', (e) => {
          console.log(`❌ Assistant Error when activating user ${k}. Trying next user ❌ \n`);
          return cb();
      }, function(err){
        if(err) return reject(err.message);
        (async() => {
          console.log(await terminalImage.file('./icon.png'))
          console.log(`Assistant Relay is now setup and running for${users.map(u => ` ${u}`)} \n`)
          console.log(`You can now visit ${global.config.baseUrl} in a browser, or send POST requests to it`);


        if (!global.config.muteStartup) {
		self.sendTextInput('broadcast Assistant Relay is now setup and running')


  sendTextInput: function(text, n, converse) {
    return new Promise((resolve, reject) => {
      console.log(`Received command ${text} \n`);
      if(converse) returnAudio = true;
      // set the conversation query to text
      global.config.conversation.textQuery = text;
      const assistant = self.setUser(n)

       assistant.start(global.config.conversation, (conversation) => {
          return self.startConversation(conversation)
          .then((data) => {
          .catch((err) => {

  sendAudioInput: function() {
    let raw = []
    const assistant = self.setUser('greg');
    fs.readFile(`${path.resolve(__dirname, 'response.wav')}`, (err, file) => {
      if(err) console.log(err)
      assistant.start(global.config.conversation, (conversation) => {
         return self.startConversation(conversation, file)
         .then((data) => {
         .catch((err) => {

  startConversation: function(conversation, file) {

    let response = {};
    const fileStream = self.outputFileStream();
    return new Promise((resolve, reject) => {
        .on('audio-data', data => {
          // set a random parameter on audio url to prevent caching
          response.audio = `http://${global.config.baseUrl}/audio?v=${Math.floor(Math.random() *  100)}`
        .on('response', (text) => {
          if (text) {  sendReq(text)

            console.log(`Google Assistant: ${text} \n`)

            response.response = text;
            if(returnAudio) {
              self.sendTextInput(`broadcast ${text}`, null, gConfig);
              returnAudio = false;
//dit it here
        .on('end-of-utterance', () => {
          console.log("Done speaking")
        .on('transcription', (data) => {
        .on('volume-percent', percent => {
          console.log(`Volume has been set to ${percent} \n`)
          response.volume = `New Volume Percent is ${percent}`;
        .on('device-action', action => {
          console.log(`Device Action: ${action} \n`)
          response.action = `Device Action is ${action}`;
        .on('ended', (error, continueConversation) => {
          if (error) {
            console.log('Conversation Ended Error:', error);
            response.success = false;
          } else if (continueConversation) {
            response.success = true;
            console.log('Continue the conversation... somehow \n');
          } else {
            response.success = true;
            console.log('Conversation Complete \n');

        .on('error', (error) => {
          console.log(`Something went wrong: ${error}`)
          response.success = false;
          response.error = error;

  setUser: function(n) {
    // set default assistant to first user
    assistant = Object.keys(global.config.assistants)[0];
    assistant = global.config.assistants[`${assistant}`];

    // check to see if user passed exists
    if(n) {
      const users = Object.keys(global.config.users);
      if(!users.includes(n.toLowerCase())) {
        console.log(`User not found, using ${Object.keys(global.config.assistants)[0]} \n`)
      } else {
        n = n.toLowerCase();
        console.log(`User specified was ${n} \n`)
        assistant = global.config.assistants[`${n}`]
    } else {
      console.log(`No user specified, using ${Object.keys(global.config.assistants)[0]} \n`)

    return assistant;

  outputFileStream: function() {
    return new FileWriter(path.resolve(__dirname, 'response.wav'), {
      sampleRate: global.config.conversation.audio.sampleRateOut,
      channels: 1

  joinAudio: function() {
    if(!inputFiles.length) {

    currentFile = inputFiles.shift()
    let stream = fs.createReadStream(currentFile);
    stream.pipe(playbackWriter, {end: false});
    stream.on('end', () => {
        console.log(currentFile, "appended")


function sendReq(text){
//console.log (text)
var gaResponse = text.replace(/\?/g, "")
var gaResponse = gaResponse.replace(/\:/g, "")

var heUri = '/apps/api/2832/trigger/setGlobalVariable=test2:' + gaResponse + '?access_token=461f117c-95ad-4d52-8136-12345678789'
var uriencode = encodeURI(heUri)
console.log ("uri: " + uriencode)
var options = {
  hostname: '',
  //port: 80,
  path: uriencode,
  method: 'GET',
  headers: {
      'Content-Type': 'application/xml',

var req = http.request(options, function(res) {
  console.log('Status: ' + res.statusCode);
  console.log('Headers: ' + JSON.stringify(res.headers));
  res.on('data', function (body) {
    console.log('Body: ' + body);
req.on('error', function(e) {
  console.log('problem with request: ' + e.message);


Not sure what you’re asking? You can use assistant-relay to perform Broadcast TTS on your Google Home Devices. You can also use Custom Commands using the [CC] prefix.

Not sure what you mean by “sending a text notification”? Do you mean use it for Push Notifications to your phone? If so, not that I am aware of.

If you want to send a push notification to your phone, there is the Join implementation that I currently use. Or you can use theIFTTT notification setup.

@ogiewon Up to now I've been running v1 of Assistant Relay with no issues. I'm rebuilding my node.js server on a newer Mac capable of newer versions of Xcode and therefor, newer versions of node.

Can you confirm, is 8.11.3 the newest version of node.js I can use or are there newer versions still that are also compatible with Assistant Relay?

Will I gain any advantage with Assistant Relay v2 or just a lot of pain in the installation and setup? I know you've explained this before. Sorry, but I cannot find where you answered that for me.

Great questions... as far as I know, you must use Node v8.x.x. I am not aware of anyone using Node 9 or 10 successfully.

I would recommend switching to Assistant-Relay v2, as that version is receiving Greg Hesp’s attention these days. You might even be able to simply change the driver on Hubitat from my v1 to my v2 driver to prevent having to create a new device and therefore modify all of your automations. This is just a guess, as I have never tried it. It should work though, since the only real difference is the format for the http requests to the NodeJS AR server.

1 Like

Perfect. Thanks so much Dan!

Thanks! I got this installed on my Raspberry Pi and it is working great.
I wonder how I can issue a custom command to Google Assistant, as I was speaking "Ok google, play some music". Tried to to this with [CC] and [CCC] prefixes but I get no response.
TTS works fine.

I would love to have Google play a random music when I arrive home!

That's because the command you are issuing it to is the Assistant Relay server. Not one of your Google Home speakers. If you issue that command to a Google Home speaker, that speaker is the one that starts playing, correct?