Brainstorm 🧠
Since I was born, my nanny has been like my grandmother. My nanny is a bit of everything for me, she is a mother, a grandmother, a friend and above all a person I can trust. My nanny is Russian but I never really had the chance to learn Russian. When the war broke out between Ukraine and Russia she was in Krasnadar. For a while I thought I wasn't going to see her again for a long time so I wondered how I was going to contact her and not only contact her but also make her feel better . I thought that if I called her and instead of my voice a person who spoke perfect Russian answered, it would surely make her laugh. So for several months I had this idea of having a superpower. And when we came across this hackathon, the idea came very naturally. So that's how Bonsaï and I came up with the idea of this direct translator.
What it does 💻
Translation 667 is a translation tool. It provides real-time voice translation. Once the program has been downloaded and the instructions followed, the user is able to communicate in English and be heard in the language chosen by the user. Indeed, in order to make the project accessible to all, we have facilitated its integration into discord and thus be able to communicate with foreign people even if we do not know their language. Of course the project has some limitations, Symbl.ai for example is only able to translate voice to text from English and our translation API currently only supports about fifteen languages even though they are internationally spoken.
How we built it 🔧
Everything was coded in Python :
--> Sound capture : Sounddevice Python library
--> Real time speech to text : Symbl.ai API
--> Text translation : Deepl API
--> Text to speech : Polly API from AWS
Challenges we ran into 🔥
During these two days we faced many challenges :
--> Firstly, we had to do a lot of research as we were not experts in the field. This information went through a lot of rigorous reading of the documentation.
--> In addition, one of the most challenging factors was the live translation. It is much more difficult to translate a text quickly when the person being translated is speaking continuously.
--> Moreover, we learned to use AWS, which is an extremely powerful tool, but difficult to learn, but well worth the effort since it allows us to translate with a voice that is extremely similar to that of a human being.
Accomplishments that we're proud of 💪
--> It was Aries first hackathon ever !
--> The real time aspect of the project
--> The use of the Polly API which, as almost every AWS API, was a tough challenge to understand and implement
What we learned ✍️
Aries : Github, How an hackathon works (as well as team working) - How to read complex documentations - Web requests (Deepl API) - Symbl.ai - Polly, AWS API
Bonsaï : The way an audio is structures at a mathematical level - How to manipulate audio chunks in real time - Computer devices management - Symbl.ai - Polly, AWS API
Startup potential ✨
The project is very simple to understand because it consists of allowing a person to express himself in a language of which he does not know a single word. We think that it is extremely useful on the one hand because it allows one to contact people in a professional environment that one could not before, and on the other hand in the social environment because it allows one to break the language barrier.
In the future, we can imagine that Symbol a.i. will allow us to translate texts in all languages and in both directions, which would allow us to patent these technologies and therefore to create companies that will help people !
Built With
- amazon-web-services
- deepl
- python
- symbl.ai


Log in or sign up for Devpost to join the conversation.