Languageizer using a browser's SpeechSynthesis

I was on a flight from Brisbane to Christchurch recently with Air New Zealand. The air host was pushing the dinner service trolley down the line. He was a New Zealander and he came to a large group of Chinese passengers who didn’t have English.

Trying to ask “Will you have the Beef or the Fish?” was met with frustrated and uncomprehending eyes. He then did something intuitively clever. He opened the lid of the two different dishes and pointed to two different version of the meal and used one word for each. Beef or Fish. Genius!

He didn’t need the words albeit they were just two words. He could communicate without speaking the same language.

Similarly I was in Paris with my wife and we were ordering dinner… I would have loved a steak but I didn’t have the word “bifteck” but I did have “boeuf” so spotting a “filet de boeuf” I got something rather tasty. I didn’t need the whole language but a few words helped.

So what words would you need to get by? Not being a complete lingauphile, just have enough to get over some of the major concepts makes life easier.

I began to work on strategic groups I know from my life experience were useful.

My languages

I’ve studied Irish and German for quite a few years. I have a few years of French. A little Italian and a smattering of words from other languages.

From these you pick up some similarities. That there is a lexicon (a bunch of words) and a grammar that sticks them all together.

I’m going to stick to just the lexicon, as the grammar isn’t needed when pointing to things like the Beef or the Fish.

Using basic language concepts, we have Nouns (the name of a person, place or thing) and Verbs (actions).

Excellent my first two major groups for the Languageizer.

How was the Languageizer built?

A very dear colleague of mine said “do you have any tools to inspire a teenager in their studies of French?”.

At the same time I’d been planning a tool for close family friends on how to learn Spanish for 3 eager young students all aged under 10 and their parents.

So how to create something that will suit everyone, parents and children alike. The Languageizer was the result.

As browser technology evolves on desktop and mobile platforms it is possible to re-imagine some of the applications of the technology and make them useful for completely different challenges.

Basic Nouns

The basic nouns of any language include:

Numbers, 1 to 30
Colours, Primary Colours
Relationships, Mother, Father, Wife etc.
Ordinals 1st, 2nd, 3rd, etc.

Excellent, so there’s a target group identified.

Concepts

Thanks to a family member who specialises in childhood education, I also learnt that before a child enters primary school there are concepts they should understand.
These concepts are grouped into the Languageizer as:

spatial (location), On/Off, Forward/Backward
temporal (time), Old/Young, Always/Never
quantity (number), More/Less, All/None
quality (description), Hard/Soft, Smooth/Rough
social-emotional (feelings). Happy/Sad

Within these groupings, you then have set lists.

Days of the Week
Months of the Year
Seasons of the Year
Time of the Day

Then you have helpful key words, that are topical when travelling

Money
Direction
Travel
Food
Food from Animals
Domestic Animals

Over time I can think of plenty more groups for the Languageizer but to get started… that would help most people have a point even if they didn’t manage any grammar.

The Accent

Modern web browsers have an amazing albeit experimental feature built into them. The ability to speak!

The technology first appeared in 2014 and has been gaining support ever since.

The engine is called SpeechSynthesis and depending on the browser and the device, you’ll get more or less voices.

These voices are able to get the language specific voice intonations and nuances of a language right.

So far Google Chrome has the most amount of voices available on all devices but I’m sure that will evolve.

The first tool I built is called TalkBox… which literally allows you access to the voices of the browser.

After a few experiments I knew I could speak in foreign languages authentically.

Now to combine the lists with the voices to create the Languageizer.

The technicalities of the Languageizer

I compiled an excel spreadsheet with the data under the different headings identified above and loaded the data into a database.

A bit of powerpoint to design an interface to access the materials.

Then I combined HTML 5, CSS 3, JavaScript, Swiper.js, SpeechSynthesis, JSON, SQL and PHP to build it.

The entire of WhereWeLearn uses UTF-8 so it can handle special language characters outside of the basic ASCII.

Where next with it?

As a result of the languages being interchangeable and the data in a database, it’s possible to not just go from English to French, but any loaded language to any language.

It also means I can expand the engine to include more languages and new lists for specific areas.

Already identified are

Firstly “Common Phrases” (Good morning, Thank you)
Furthermore “Body Parts” (head, eye, finger)
Also “Party Language” (Beer, Wine, Finger Food)
Next “Anniversaries and Events” (Birthday, New Year, Christmas, QuiZinière)

The voices that Chrome current list on a desktop include as of today “Google ” :

Deutsch
US English
UK English Female
UK English Male
español
español de Estados Unidos
français
हिन्दी //Hindi
Bahasa Indonesia
italiano
日本語 //Japanese
한국의 //Korean
Nederlands
polski
português do Brasil
русский //russian
普通话（中国大陆） //Mandarin (Mainland China)
粤語（香港） //Cantonese (Hong Kong)
國語（臺灣） //Mandarin (Taiwan)

So as new groups and translations are included the Languageizer project will grow.

If there’s anything in this article you’d like to chat to me about or help me with you can contact me here or on social media.

Philip Lacey

Languageizer using a browser’s SpeechSynthesis

Current Languageizer and Language Projects on WhereWeLearn

Speaking in Tongues

My languages