Get WhatsApp voice notifications & translations in native languages with Tasker

If you are fortunate enough to be bilingual, you’ll quickly understand the shortcomings of Android OS (others too) when it comes to speech. True, the options for most popular languages are there, however Android does very little to help to switch from one another like it does for the keyboard input. This is surprising considering the entire Google Translate API is available to us, mortals. I decided to take the advantage of that and improve one of my car profiles WhatsApp Voice Notifications. In that tutorial, I’d assign the spoken language to the contact name. Google’s text to speech system (TTS) would set the language based on the nationality of my contact. It worked well enough until some of my friends would send me lines in other than assigned, languages.

Thanks to the Google Translate API I can not only change the language of the text-to-speak engine on the fly but also incorporate on-the-fly translations. The best thing about it is that the delay isn’t significant. The whole system works great and I can enjoy the spoken notifications in my car yet again.

Google Translate API & WhatsApp Voice Notifications

To determine the language and translate the text, you are going to need the API key for the Google Translate service. Log in to the Google Console, enable Google Translate API and generate an API key. You are going to need this to authenticate the HTTP calls.

WhatsApp Voice Notifications Types

I’m going to show you how to achieve this on WhatsApp, but the same can be applied to other messengers that store the text of the message in the Android notification. The Tasker profile will have 2 modes.

1. Capture the language of the message and set a correct TTS
2. Translate the message to your preferred language

In the next write up, I will also show you, how to achieve this with Google Home.

Querying language with Tasker

A1: HTTP Post [ Server:Port:https://translation.googleapis.com/language/translate/v2/detect?key=%TranslateAPI 
	Path: Data / File:
	q=%par1 
	Cookies: User Agent: Timeout:10 
	Content Type:application/x-www-form-urlencoded 
	Output File:Download/lang.txt 
	Trust Any Certificate:On ] 
A2: AutoTools Json Read [ Configuration:Input Format: Json
	Simple Mode: true
	Json: /storage/emulated/0/Download/lang.txt
	Fields: language
	Separator: , Timeout (Seconds):60 ] 
A3: Return [ Value:%language Stop:On ]

To get the language info, we have to submit the test sample using Google Translate API and HTTP Post request. The URL has to be authorised with your API key, and contain the JSON object with the message sample.

https://translation.googleapis.com/language/translate/v2/detect?key=YOUR_API

The message sample is submitted under the “q” key and to do this in Tasker simply add:

q=enter_text_to_sample

in your data field. Once the request is submitted, you will receive a JSON (want to know more about JSON in Tasker – tutorial here)formatted object which looks like this:

{  "data": {"detections": [
      [{  "confidence": 1,
          "isReliable": false,
          "language": "en"
      }]
    ]      }
}

Of course, we are interested in the language detection and the code that comes with it. You can look up all the ISO language codes here:

To make our life complicated, the TTS engine codes for individual languages do not correspond with the ISO standard (why? dialects and localisations) so each TTS setting will have to come with defined IF condition. It’s ok, I’d rather do this that way as realistically speaking, you will only need 2-5 languages at best. If you want to create a full conversion table in Tasker, be my guest!

Since response comes as a JSON, I’m going to use AutoTools JSON Read action to extract the value and identify the language. At this stage, all I need is to set the variable %language, which will determine the TTS engine later on using IF statements. The correct Say Tasker action is selected, and the sample text is spoken with the properly picked TTS engine.

Translating a message with Tasker

A1: HTTP Post [ Server:Port:https://translation.googleapis.com/language/translate/v2/?key=%TranslateAPI
	Path: Data / File:
	q=%par1
	target=en
	Cookies: User Agent: Timeout:10 
	Content Type:application/x-www-form-urlencoded 
	Output File:Download/translated.txt 
	Trust Any Certificate:On ] 


A2: [X] HTTP Post [ Server:Port:https://translation.googleapis.com/language/translate/v2/?key=%TranslateAPI 
	Path: Data / File:
	q=%par1
	target=zh-CN 
	Cookies: User Agent: Timeout:10 
	Content Type:application/x-www-form-urlencoded 
	Output File:Download/translated.txt 
	Trust Any Certificate:On ] 


A3: [X] HTTP Post [ Server:Port:https://translation.googleapis.com/language/translate/v2/?key=%TranslateAPI 
	Path: Data / File:
	q=%par1
	target=pl 
	Cookies: User Agent: Timeout:10 
	Content Type:application/x-www-form-urlencoded 
	Output File:Download/translated.txt 
	Trust Any Certificate:On ] 
A4: AutoTools Json Read [ Configuration:Input Format: Json
	Simple Mode: true
	Json: /storage/emulated/0/Download/translated.txt
	Fields: translatedText
	Separator: , Timeout (Seconds):60 ] 
A5: Return [ Value:%translatedtext Stop:On ] 

The second option assumes that we are going to translate every message to a default language. For the sake of this tutorial I will use English, but feel free to pick your own native language instead.

To submit the data to Google Translate API, once again, I’m going to send an HTTP Post to the authenticated URL:

https://translation.googleapis.com/language/translate/v2?key=YOUR_API

This time the data has to contain the sample text and the language we are going to translate the text to.

q=your_sample_message
target=language (ISO language code)

The return message is a JSON response which looks like this:

{  
    "data":{  
         "translations":[  
            {  
               "translatedText":"This is just a test",
               "detectedSourceLanguage":"pl"
            }
         ]
      }
   }

In the same manner, AutoTools JON Read action will extract the value translatedtext and set it in %translatedtext for a later use.

International WhatsApp Voice Notifications

Profile: WhatsApp Translations 
	Event: AutoNotification Intercept [ Configuration:Event Behaviour: true
			Notification Type: Only Created Notifications
			Has Reply Action: true
			Notification Apps: WhatsApp
			Notification App: WhatsApp
			Package Name: com.whatsapp ]
			Enter: WhatsApp Spoken International 

	
	A1: If [ %WhatsAppMode eq 0 ]
	A2: Perform Task [ Name:WhatsApp Get Language 
		Priority:%priority 
		Parameter 1 (%par1):%antext 
		Parameter 2 (%par2): 
		Return Value Variable:%sayit Stop:Off ] 
	A3: Say [ Text:%antext Engine:Voice:default:default
			Stream:3 Pitch:5 Speed:5 Respect Audio Focus:On Network:Off 
			Continue Task Immediately:Off ] 
			If [ %sayit ~ en ]
	A4: Say [ Text:%antext Engine:Voice:com.google.android.tts:pol-pol 
			Stream:3 Pitch:5 Speed:5 Respect Audio Focus:On Network:Off 
			Continue Task Immediately:Off ] 
			If [ %sayit ~ pl ]
	A5: Say [ Text:%antext Engine:Voice:com.google.android.tts:zho-chn 
			Stream:3 Pitch:5 Speed:5 Respect Audio Focus:On Network:Off 
			Continue Task Immediately:Off ] 
			If [ %sayit ~ zh-CN ]
	A6: End If 
	
	
	A7: If [ %WhatsAppMode eq 1 ]
	A8: Perform Task [ Name:WhatsApp Translate 
		Priority:%priority 
		Parameter 1 (%par1):%antext 
		Parameter 2 (%par2): 
		Return Value Variable:%sayit Stop:Off ] 
		
	
	A9: Say [ Text:%sayit Engine:Voice:com.google.android.tts:eng-gbr
		Stream:3 Pitch:5 Speed:5 Respect Audio Focus:On Network:Off
		Continue Task Immediately:Off ] 
		
	
	A10: [X] Say [ Text:%sayit Engine:Voice:com.google.android.tts:pol-pol
		Stream:3 Pitch:5 Speed:5 Respect Audio Focus:On Network:Off
		Continue Task Immediately:Off ] 
		
	
	A11: [X] Say [ Text:%sayit Engine:Voice:com.google.android.tts:zho-chn
		Stream:3 Pitch:5 Speed:5 Respect Audio Focus:On Network:Off
		Continue Task Immediately:Off ] 
	A12: End If 

We have the full know-how to create the on-the-fly adjustments. The profile is very simple actually. I’m going to capture WhatsApp notifications using AutoNotification (intercept Whatsapp notifications with a “reply”), and then use Say action after the language or translation has been determined.

To get the language or translation, I will use the Perform Task action and send the %antext as %par1. The language detection task will return the %language code and the translate task will return the %translatedtext to the main task.

The returned value will be passed over to the Say action with a specified TTS either by the IF condition (%sayit) or predefined language setting for your translations.

A1: Profile Status [ Name:WhatsApp Translations Set:On ] 
A2: Variable Add [ Name:%WhatsAppMode Value:1 Wrap Around:2 ] 
A3: Flash [ Text:Native language mode enabled Long:On ] 
	If [ %WhatsAppMode eq 0 ]
A4: Flash [ Text:Translating to another language Long:On ] 
	If [ %WhatsAppMode eq 1 ]

If you want to hardcode your option into the task (and you don’t need another one) – feel free to do so. I’m going to wrap both actions in an additional IF condition which will determine what behaviour I want from my profile. A variable %WhatsAppMode becomes 1 or 0 thanks to the Variable Add action with a wrap around 2.

A toggle will set a global variable and flash the confirmation of the selected behaviour.  And with that done, you can decide for yourself how you want to enable it. You can trigger the profile from a home shortcut, voice command or the notification bar tiles.

Conclusion

This is a much-improved version of my previous voice notification profile. WhatsApp Voice Notifications are not perfect, as foreign words in a single message won’t get the same treatment, but as long as your contact sticks to a single language per sentence – everything should be fine. The translated data can be also used in custom notifications which would have the translated messages. This is something I’m not going to cover in this tutorial.  Be aware I have not included any exceptions for URL handling. If your friend sends you a link, this profile will read it out!

Lastly, I will be looking into adding this to my Google Home Notifications too. This way messages received to your phone can be passed over, translated and spoken by the smart speaker.

Project Download

Download project files here. Bear in mind that Patreon supporters have early access to project files and videos.

Download