Project Guidance

Hello my Siniors.

am a student beginner in project development, especially projects based on hardware, i wantend your guidance please? i want to create a system based on microcontroller that will unable users or just people to communicate together without any matter, like in the class students who don't know english, and a lecture who doesn't know their local languages, can use the system to assimilate the class. means a system which will need headset with mic, a smartphone wifi connected and micro controller

The "universal translator" is still science fiction.

However, using very large, very fast computers and machine learning techniques, people are making respectable progress toward that goal. This is not an Arduino project.

How many different written words will you need, in both languages, will you need to make your design work?

I will write as much as possible words

This is not a correct answer

One of the first problems you will discover is not all languages use the 26 letters of the Latin alphabet that English speaking people use.
If you decide to only use English words and Latin alphabet, you might make an array of the words using a fixed length of the longest word you will need. When putting words together to make a sentence, you can easily remove any spaces at the and of shorter words.

The very FIRST thing a student should learn is "systems analysis". That means exploring in detail the requirements of your proposal. That also includes the amount of storage needed, as Delta_G just demonstrated for you.
The second thing is to estimate the throughput speed of your project. Given some input, how much time will be needed to give output.

Another important starting point for engineering is figuring out what others have done already, and how they're doing it.

The closest existing example to your project that I know of is Youtube's auto-translated subtitles. That's speech to text plus text translation. I think this is pretty much state of the art when it comes to this field.

Important observations:

  • YT is using computer farms resembling supercomputers
  • automatic subtitles are never available for new uploads, only for older uploads: this suggests it's not done anywhere near to real time
  • while usable to get the gist of what's being said most of the time, the transcripts let alone the translations are far from perfect

So that shows you roughly what you're up against.

Of course you could always start looking for a babelfsh to stick in your ear. The real one has unfortunately died.

what should it be, please? cause i wanted to only use realtime translation and between two languages include english, and would need a headset of caurse with mic, a smartphone and other equipments which i don't know now, if arduino, for there you said this is not an arduino project, so, could you guide me please?

the other language wich will be translated with english uses sames characters as english and am maybe my responses was too short but, i would like to write word, as many as possible, to enrich this project and if possible to use online language library, i will do that, any directive please?

Have you done the analysis to determine how many minutes/hours will be needed to do your translation of single sentence?

try this: find the lyrics to a song you know well. I used "For what It's worth" by Buffalo Springfield. Copy those lyrics. paste them into a translation program. I chose Indonesian for no particular reason

Translate. Cut and paste that translation back into the translation program, on the foreign language to English side. Marvel at the results.

Now imagine explaining how a resonant circuit works in English, then translating that with a computer. you would need an AI that was part Stephen Hawking and part Carl Sagan

Here is a well known sentence: " Four score and seven years ago our fathers brought forth, upon this continent, a new nation, conceived in liberty, and dedicated to the proposition that all men are created equal.". Since French and Spanish both use the Latin alphabet, how much time will your program need to translate that sentence into one of the two languages?

Currently not possible, with audio output, using computers.

The state of the art is probably OpenAI's Whisper, which is open source, and can produce English textual translations of audio input in about 50 different languages, using a large computer with a powerful GPU. Whisper is far from real time, though.

You can experiment with Whisper for free using Google Colaboratory.

Capture

Good example. : )

Back-in-th-day as OCR was forcing its way onto official documents, but the private world enjoyed BOTH fonts; Pica and Elite, I had to "train" scanner software to recognize every word in standard fonts and the new sans font, including "A" and "a" ... painful. Ten years later, scanners know the time of day I paid. Nice. Just wanted to say... "Train" again.

Anyone can help me to start, and what should I use as fundamental devices to start, and other things like synonyms or whatever, I will know how to handle them.

Thanks and waiting for your comment and help please.

@mpzo2024
Isn't everything above a help? Or are you waiting for someone to start writing code for you?

The opinion of the community, it seems to me, is quite definite - this cannot be done on a standalone microcontroller at the current level of development.
So what kind of help are you waiting for if the problem has no solution?

please, if it cannot be done by arduino, could you suggest me something else where can it be feasible?
thanks.

You can begin with Python on your PC. Start small and discover the logic you need to translate word-for-word using two languages you are familiar with. Slowly increase the number of words.