WEBVTT NOTE Transcript 2024-04-11 from FLAC by https://elvery.net/prototypes/transcribe/ on my MacBook Air M1 on maximum quality setting. NOTE Body edited. 00:00.000 --> 00:10.200 Hi, I'm Damon Hart-Davis and this is a mini Earth Notes podcast episode for 29th October 2019. 00:10.200 --> 00:16.900 Move over Handel! Having tried "audifying" various chunks of date in a fairly simplistic way, 00:16.900 --> 00:24.400 here I try my hand at "sonifying" some water temperature data at a slightly more abstract level 00:24.400 --> 00:29.200 via a MIDI soundtrack and a slightly larger set of tools. 00:29.200 --> 00:34.900 Rather than turn data values directly into amplitude at a very high rate to make a tone, 00:34.900 --> 00:39.800 I'm now more slowly turning those values into note pitches. 00:39.800 --> 00:48.300 For this, I turn my data first into a CSV (Comma-Separated Values) file such as Excel might load. 00:48.300 --> 00:54.800 Then I convert it into a version with regular cadence even when the data is not itself perfectly regular. 00:54.800 --> 01:05.700 Then I process that file into a MIDI (Musical Instrument Digital Interface) file with each data point a single note. 01:05.700 --> 01:12.500 I have Apple's GarageBand on my Mac transform that into a (large WAV) audio file. 01:12.500 --> 01:21.200 Finally, Audacity exports into the final FLAC (lossless) and MP3 (lossy) audio files. 01:21.200 --> 01:26.200 Whew! About a half dozen steps, though most are fairly mechanical. 01:26.200 --> 01:33.100 The step of converting to MIDI from the even-cadence data is the one with the most scope for fun, probably. 01:33.100 --> 01:38.800 The first pass was the simplest thing I could get working, and for under a hundred lines of shell and awk script, 01:38.800 --> 01:41.800 I don't think that it's too bad. 01:41.800 --> 01:49.100 For my first attempt, I take about a decade of (eventually) monthly readings of my kitchen tap water temperature 01:49.100 --> 01:52.100 and turn it into about 30 seconds of audio. 01:52.100 --> 01:55.200 Maybe it's water electronica. 01:55.200 --> 02:00.200 (I have been advised that 70 to 140 beats per minute is normal music tempo, 02:00.200 --> 02:08.900 but I have made something like 120 data points for 30 seconds, so maybe 240 bpm?) 02:08.900 --> 02:13.900 [music] 02:13.900 --> 02:42.900 [music] 02:42.900 --> 02:47.900 That didn't seem to actually have the instrument voice loaded, which is fine. 02:47.900 --> 02:52.900 This time then, with feeling, and a dollop of fake harp. 02:52.900 --> 03:21.900 [music] 03:21.900 --> 03:27.900 [music] 03:27.900 --> 03:32.900 And then, with four copies of the harp track overlaid and slightly delayed. 03:32.900 --> 03:38.900 The more delayed copies are quieter, as something like echo or reverb. 03:38.900 --> 03:44.900 This is all done in GarageBand from the previous harp MIDI file. 03:44.900 --> 03:54.900 [music] 03:54.900 --> 04:19.899 [music] 04:26.900 --> 04:29.900 So, my first adventures in audification and sonfication via MIDI files generated from the source data. The world is my oyster. 04:29.900 --> 04:37.900 Another idea that I'd like to try is using outside temperature for pitch and taking volume ('velocity') from gas demand. 04:37.900 --> 04:43.900 Another is to create various monthly series - e.g. battery flow, electricity demand, PV generation - 04:43.900 --> 04:48.900 and give them their own instruments and arrange them on a virtual stage with LR panning. 04:48.900 --> 04:56.900 There's more on my Earth Notes Web site at Earth.Org.UK.