Hi folks! Great package and really easy to use, kudo's for that!

I was just logging audio as well and noted that the durations of the audio being saved did not correspond to the duration of the audio displayed in Tensorboard. The examples in examples/Audios.jl seem to have a duration in Tensorboard of 11 seconds. It's like the sampling rate does not matter and that it defaults to something like 8 kHz.