|
|
|
a computer calculates silence
Thursday, January 18 2018
[REDACTED]
We needed to pick up Neville's new, stronger antibiotics today, so I wrapped it into a visit to the Tibetan Center thrift store. There wasn't much for me there except an esoteric USB cable and a spool of thin steel wire, but sometimes it's good to know I'm not missing out on some amazing thing I wouldn't even know existed if I didn't run across it there.
Yesterday (using a USB audio dongle attached to a laptop) I'd digitally copied some audio off an ancient cassette tape from the early 1970s. The tape featured a tiny two-year-old Gretchen and her brother Brian being interviewed by her parents[REDACTED]. The tape had a terrible audio quality, but today I was able to clean it up by filtering out an oppressive noisy low-end, and punching up the audio frequencies. Using Audacity's Noise Removal Effect, I was able to make it sound almost as if it had been recorded in a silent studio. The effect works by using a sample of the background noise from what would be, in a noiseless recording, a silent section of the recorded material. Using internal logic based on this training material (the mechanism might even involve machine learning), it then turns all of that noise into silence. Gretchen had asked for a digital copy of this old tape as a birthday present, and I was trying to do it right.
This evening's weekly video happy hour went for a moderately long time, but it didn't set any record (that would've required extending into the wee hours of the next day). I was pretty drunk by the time it concluded. In my drunken state, I painted a greyscale version of a Neville painting I'd done twice before. The idea was to maybe give it to Gretchen as an additional gift tomorrow, but even in my drunken state, I knew the quality was insufficient. She already has a much better version of this painting in color.
The black and white painting of Neville kind of works in the context of better ones.
[REDACTED]
For linking purposes this article's URL is:
http://asecular.com/blog.php?180118 feedback
previous | next |
