In this work, the application of audio watermarking in large scale short-form video platforms is explored by addressing challenges particularly focusing on minimizing watermark audibility while maximizing detectability. Experimental results are presented, discussing approaches to improve imperceptibility such as using a mixing gain for the watermark signal and applying it only on speech segments. The experiments also examine the impact of multiple audio encodings and music mixing on watermark detectability, proposing solutions to enhance robustness.