A google cloud speech node module to comunicate with the API that is compatible with Electron.
The official module, @google-cloud/speech, relies on grpc as a dependencie and there is an issue using the official module in electron .
This is the error I would get when trying to integrat @google-cloud/speech in electron for adding google cloud speech as a STT option for autoEdit.
Uncaught Error: The module '/Users/pietropassarelli/Dropbox/CODE/Vox/TBVE/autoEdit_v2/node_modules/grpc/src/node/extension_binary/grpc_node.node'
was compiled against a different Node.js version using
NODE_MODULE_VERSION 48. This version of Node.js requires
NODE_MODULE_VERSION 53. Please try re-compiling or re-installing
the module (for instance, using `npm rebuild` or`npm install`).
at process.module.(anonymous function) [as dlopen] (ELECTRON_ASAR.js:173:20)
at Object.Module._extensions..node (module.js:598:18)
at Object.module.(anonymous function) [as .node] (ELECTRON_ASAR.js:173:20)
at Module.load (module.js:488:32)
at tryModuleLoad (module.js:447:12)
at Function.Module._load (module.js:439:3)
at Module.require (module.js:498:17)
at require (internal/module.js:20:19)
at Object.<anonymous> (/Users/pietropassarelli/Dropbox/CODE/Vox/TBVE/autoEdit_v2/node_modules/grpc/src/node/src/grpc_extension.js:45:15)
at Object.<anonymous> (/Users/pietropassarelli/Dropbox/CODE/Vox/TBVE/autoEdit_v2/node_modules/grpc/src/node/src/grpc_extension.js:49:3)Official documentation has more details on what are the features that requrie grpc. eg Streaming Recognition.
However both Synchronous Recognition and Asynchronous Recognition can be used with REST.
Follow these instrucitons from officinal documetnation
If your client application does not use OAuth 2.0, then it must include an API key when it calls an API that's enabled within a Google Cloud Platform project. The application passes this key into all API requests as a key=API_key parameter.
To create your application's API key:
- Go to the Cloud Platform Console.
- From the projects list, select a project or create a new one.
- If the APIs & services page isn't already open, open the left side menu and select APIs & services.
- On the left, choose Credentials.
- Click Create credentials and then select API key.
To use the actual module, it is not opinionated about where you store your API key, as this is passed to the class constructor when creating a new instance.
However to try the test/example I am using .env file in the root of the project. This is in .gitignore to avoid accidentally pushing it to github. So you need to create one as follows. This follows the practice of storing sensistive information as enviroment variables
GSPEECHKEY=your_google_speech_api_key_here
See example.js
Add the module
npm install https://github.com/pietrop/gcloud_speech_electron -saveUse as follow
'use strict';
//Get google API keys here https://support.google.com/cloud/answer/6158862?hl=en
//require('dotenv').config();
//const APIKEY= process.env.GSPEECHKEY;
const APIKEY= `your_google_api_keys_here`;
var Speech = require('gcloud_speech_electron');
//var Speech = require('./index.js');
var sampleAudio = './brooklyn.wav';
var language = 'en-US';
var configSTT = {
"encoding": 'LINEAR16',
"sampleRateHertz": 16000,
// for supported languages see https://cloud.google.com/speech/docs/languages
"languageCode": language,
"maxAlternatives": 1,
"profanityFilter": false,
"enableWordTimeOffsets": true
};
var client = new Speech(APIKEY);
client.recognize(sampleAudio,configSTT, function(error, response){
if(error){
console.error("There has been an error");
//console.error(JSON.stringify(error.error.message, null, 2));
// Do something with the Error
console.error(JSON.stringify(error,null,2));
}else{
// Do something with the results
console.log(JSON.stringify(response,null,2));
}
});Which would return the following json, with word accurate timings, because we have enableWordTimeOffsets set to true.
{
"results": [
{
"alternatives": [
{
"transcript": "how old is the Brooklyn Bridge",
"confidence": 0.987629,
"words": [
{
"startTime": "0s",
"endTime": "0.300s",
"word": "how"
},
{
"startTime": "0.300s",
"endTime": "0.600s",
"word": "old"
},
{
"startTime": "0.600s",
"endTime": "0.800s",
"word": "is"
},
{
"startTime": "0.800s",
"endTime": "0.900s",
"word": "the"
},
{
"startTime": "0.900s",
"endTime": "1.100s",
"word": "Brooklyn"
},
{
"startTime": "1.100s",
"endTime": "1.500s",
"word": "Bridge"
}
]
}
]
}
]
}Example error if you mistype one of the required parameters eg LINEAR16----.
{
"error": {
"code": 400,
"message": "Invalid value at 'config.encoding' (TYPE_ENUM), \"LINEAR16----\"",
"status": "INVALID_ARGUMENT",
"details": [
{
"@type": "type.googleapis.com/google.rpc.BadRequest",
"fieldViolations": [
{
"field": "config.encoding",
"description": "Invalid value at 'config.encoding' (TYPE_ENUM), \"LINEAR16----\""
}
]
}
]
}
}{
"error": {
"code": 400,
"message": "Request payload size exceeds the limit: 10485760 bytes.",
"status": "INVALID_ARGUMENT"
}
}There seems to be a 10.48576 MB file size limit, when I tried to implement longRunningRecognizewhich I have not found in the documentation.
Tests are incomplete but for now you can run a minimal one with
npm test
- replace callback with promises to be closer to official module and make it easier to swap when needed (eg using electron)
- add more details to documentation
- make more comprehensive tests with
jest - Find out how to test API SDK with jest, do you need to mock the API?
- add long
longRunningRecognizeoption to the module, with 15 sec polling to check when transcription is completed. (this raises limit to 80min, however file size limit to 10.48 mb makes it tricky to implement);