Let me guess, you want to work on something which has risks as well as rewards and you like doing it? or you are in a situation where you want to decide between an early-stage startup or a corporate. I’d choose startup any day!
Hi, My name is Jaskaran Singh, and I’m a Deep Learning Engineer at Dubverse.ai. I joined dubverse back in February 2021 right after it was registered 😛. I had met Varshul Gupta (Founder/Mentor) before dubverse through LinkedIn via our shared interest in AI. So when he told me about a new opportunity that includes working on the latest AI, I said yes 😃 . This blog is about my journey at dubverse.ai
Initial Stage/Early First year
I joined dubverse during my last semester of Btech as an intern, well everyone was an intern (and a paid one). We only had a jupyter notebook (python IDE) as our MVP which was made interactive with buttons and tools (We still use it), you enter a URL of the youtube video you want to dub and select the target language and follow the process to get the final dubbed video.
During this time we were working in silo mode, I was working on a speech synthesis problem which included reading research papers, coming up with different hypotheses, implementing them, training the model, getting the results, and validating them to improve further. This Iteration cycle is the central theme of learning at dubverse.
We were also experimenting with the use-cases that we had for the TTS technology, one of them was around voice cloning, we got in contact with a YouTuber and cloned his voice so that it would be easier and cost-effective to produce videos. We designed the business model and got the prototype ready to test, though it didn’t work out because of the current limitation in the TTS technology, it was a great experience.
Around May I had my final exams finished for the last semester. This was a critical time for me to make a big decision (TCS vs Dubverse). I wanted to work on AI but being a first employee of an early-stage startup was too risky to consider as well as meant getting more opportunities and learning curve, but it became clear when we got our First Client 👊 now alongside research I was working on the product too, building and shipping features to complete our first major dubbing project. I was working directly on the MVP which was impacting the final result of the dubbing quality!
This was an amazing experience and I decided to continue my journey at Dubverse and became the First core member👐
Around Dec we got our core team… https://dubverse.ai/team/
Each one of them is a beast in their field. We have weekly meetings where we discuss everyone’s progress and the overall product progress so there is learning across Engineering, AI, Growth and Product development, and the overall alignment of the company is similar.
P.S. we party a lot 😸
We did a Product hunt launch in January 2022 and got an amazing response from the PH community. During the launch, we did tweets, twitter spaces, and explored different platforms. This launch put us on the internet and we started talking with our potential users.
This helped us discover many use-cases and ideas that would help dubverse grow.
This is the first major AI project. Creating multiple Speakers (TTS) with the help of voice-over artists for every Indian and International Language. This project had many iterations and is still evolving.
The clip above is of Mr Shyaam (Dubverse Speaker), he is a good pal 😛.
Research (n) Development
Dubverse is a core AI startup. Research is always happening here. we like to call it “lab to product path”. Now we are dedicating our efforts to Speech, Natural language processing and Visual processing as well. We are constantly looking for good talent from interns to advisors. Dubbing is a complex problem that requires complex solutions across domains and hence we are constantly researching to improve the experience.