It’s almost a week since I’ve been diagnosed with COVID-19, and currently I’ve been admitted into the low risk center in Hospital Kuala Lumpur. Fortunately none of my close contacts have the diagnosis as well so far; they all need to be quarantined for 10 days.
The only productive activity has been exercise (once the headaches went away). I managed to upload a new Crossy Road Castle video:
If you came here just for the video, here you go:
This was a prerecorded video played for this webinar. Its audio was however, distorted throughout from likely clipping. The version uploaded on YouTube had EQ work done to reduce the distortion; investing in denoising is enough expense. Reminder to everyone including myself whenever recording audio: always check the gain and avoid clipping!
Without this, you would definitely learn what clipping is.
As usual, don’t forget to like the video and subscribe to the channel.
If you came here just for the video, here you go (please be patient with the audio, as it gets better in the middle):
Production notes below.
As I said earlier, the audio does get better in the middle, and this was the best that I could fix. A lot of stuff I could tell during editing the original footage given to me: > 1 mic streaming, air conditioning noise, use of noise gating were those I picked up. Without spending further on further audio plugins that would in theory fix some of the problems, I used what I had. I identified 6 segments of different recording settings and did a combination of denoising/high and low pass filtering/EQ’ing (this seemed to attenuate both double mic’ing and leftover noise)/compression (had to push the gain really high as the voice levels became very low trying to extract any good audio quality).
Heavy EQ’ing to remove wind noise “protected” by the noise gate and remove effect of 2 mics working simultaneously + the need to boost things back up after all of the attenuation
Also experimented with moving labels to keep up with moving objects during scanning; staying still scanning while talking isn’t easy. I won’t be using this technique ever, as it’s dizzying and distracting to follow the moving scan and the corresponding labels. Luckily there were spots where the speaker finished a sentence while scanning, allowing me to interject with freeze frames and have static labels, which is much easier to follow. I guess in the future any demos must keep this in mind if there’s a desire to label things in post: keep the probe still (as much as possible) when pointing things out, then shift only once finished to show other views.
I’m not sure if my Takstar SGC-598 or Boya BY-MM1 (with the dead cat) would be able to overcome the heavy wind noise there. Maybe a RODE NTG supercardioid? That RODE though, is much, much, much more expensive.
As usual, don’t forget to like the video and subscribe to the channel.