.jpg)
Upon Chabaka's request, I collaborated with the Tunisian National Television Team (our client) to come up with a concept that we can work as a collaboration. I pitched some ideas, among which a series where we use AI to upscale, colorize & restore historic footage & audio recording of historical Tunisian artists & use that to create a documentary. The following is the pilot:
All of this was done in 3 weeks, right on time for the ASBU conference on AI and its use in journalism.
It was very challenging to work with the videos I received from the archive but Topaz was powerful enough to change them drastically. I then used an AI colorizer on a cloud GPU to colorize the videos. It wasn't perfect, but it was huge compared to the initial state.
For the audio, I cleaned up the artists' voice and cloned it, then made him sing a more recent Tunisian classic while keeping his expression.
I also used another TTS to generate the narration for the documentary & video generators to create scenes for the music video part.
The most challenging part was definitely dealing with the Arabic language in TTS, 11labs just couldn't understand the punctuation and nuances of the language which meant I had to experiment with different methods and re-render many times to get a correct result
For the project launch announcement, I created this promo to be played during the conference.
The clients loved it so much and want to go forward with this series. They're currently gathering a team to work on it. We displayed the episode to showcase the collaboration in ASBU conference, which opened up the possibility to get a lot of new customers.
Here's the interview link
