I see chiming clocks as a kind of sonic interface (Li, 2018). My original chiming clock reproduced the half-hour beeps produced by sysline (a now-uncommon Unix program) using Javascript and required a browser to run. At that time, WebAudioAPI was still somewhat of a “first-class object” – at least some browsers wouldn’t actively sabotage audio. Soon after I made this first applet, I wrote a second one, based on this first one, that announces night watches.
By 2020, the WebAudioAPI scene had become drastically different. Browsers now actively block audio, on the misguided idea that this could solve the “unwanted audio” problem (even though things like cookie notices had long rendered audio blocking worse than useless). Doing chiming with web applets was no longer an attractive option. Also, because two different applets were producing two kinds of unrelated chimes, the result was often “illegible” sonic output. A replacement was in order.
I chose Perl for the rewrite out of laziness. But unlike Javascript (or Python), Perl has no native support for audio. The new code currently uses mpv to play audio recordings, and multiple copies of mpv are needed to play a chime melody. This is clearly not ideal since resource usage is unpredictable so there’s both a danger of thrashing and audio distortion.
The script’s actual chime melodies have been sourced from Wikipedia (“Westminster Quarters”, 2021; “Whittington chimes”, 2021) and various out-of-copyright sources, including a couple of papers written by Starmer (1907, 1910).
The Whittington chimes actually has nothing to do with Whittington, since they were clearly already in use when Whittington allegedly heard them. According to Wilson (2005, 2006), the melody was originally called simply the “chime on 8 bells”, even though it was originally a 6-bell chime (actual composition lost). The name “Whittington” only started to take off around 1900. An 11-bell version, later composed specifically for St Mary Le Bow, was based on what was still known about the 6-bell version (Starmer, 1907, p. 5).
Because the melodies used are based on written sources, they might not correspond to reality. For example, although mechanical clocks are precision devices, their chiming mechanisms aren’t. Both grandfather clocks and clock towers exhibit irregularities in the rhythms they produce (George, 2014; Heath, 2011). The pitches produced by actual clocks might also deviate from the theoretical pitches described by scores (George, 2014).
The chime melodies for carillon mode are currently encoded in a small subset of Lilypond notation, partly because encoding melodies as arrays is cumbersome, and partly because Lilypond notation might be reasonably accessible (Zahnke, 2020). Some other sonic notifications – specifically night watch announcements and sysline beeps in synth mode – are also encoded in this notation.
Before the encoded melodies can be used, the code converts them to an internal data structure by calling the function interpret_melody. This function currently only understands the commands \relative, \drums, \tempo, \tuplex, ^\fermata, \set Staff.instrumentName and basic expressive marks (\ppp, \pp, \p, etc.). In particular, \laisserVibrer is not understood; whether to laisser vibrer is currently hard-coded as a hack. Ties are understood, but slurs are ignored. \breathe and \time are ignored.
Absolute mode is not supported,
but the scripts’s understanding of relative mode is not guaranteed to be correct.
If a score typesets correctly in Lilypond
(using make proof)
but interpret_melody converts it incorrectly
(using chimer --test-tune),
the bug is in interpret_melody.
Some very basic polyphony support is in place for “continental” chiming as described by Starmer (1910),
but it’s safest to use the existing melodies in the code as a guide.
Most percussive notes are understood and are mapped to the corresponding MIDI percussion instrument.
One thing that absolutely does not work is polyphony inside a repeat.
In real Lilypond, you can expect things like \repeat volta 12 { c1 \\ g, \\ e \\ c } to work,
but writing \repeat volta %I { c1 \\\\ g, \\\\ e \\\\ c } for hour strikes will confuse the script
and give you results that are completely wrong.
The current recommended method to generate chimes is to use fluidsynth, ideally with the Timbres of Heaven soundfont. The synth code always uses only one soundfont (either Timbres of Heaven or FluidR3); it is not currently possible to mix-and-match instruments from different soundfonts.
The result of synthesis is saved into an .oga (OGG audio) file in the data directory, for later playback using mpv.
The following were noted when synthesis was originally explored:
- pros: one subprocess only for each sink
- cons: if we have more than one sink configured, running two copies of fluidsynth at the same time will cause massive distortion. It might be possible to pre-generate a sound file and then use mpv to play that pregenerated file when an event fires.
The following options were originally explored:
- drum kit note 57 (crash cymbal 2) – sounds good
- drum kit note 52 (Chinese cymbal) – would be appropriate
- drum kit note 31 (sticks)
- drum kit note 60 (high bongo)
- drum kit note 61 (low bongo)
- drum kit note 76 (high wood block)
- drum kit note 77 (low wood block)
- drum kit note 75 (claves) – sounds good
Chimes were originally played using samples that were mostly sourced from freesound. The code still exists but has not been used in a long while so sample-based chiming might or might not still work.
The following options were originally explored:
https://freesound.org/people/strangehorizon/packs/32761/
- pros: the whole scale, public domain (cc0)
- cons: need lots of mpv subprocesses to play chime melodies
We can use gongs, but it sounds a little off
https://freesound.org/people/sandyrb/packs/6993/
- pros: fairly clear, cc-by-3.0, Canadian
- cons: possibly too loud (possible distortion), 15-semitone range not enough for Westminster chimes
https://freesound.org/people/Samulis/packs/21054/
- pros: very clear, public domain (cc0)
- cons: not enough range (only 4 notes)
https://freesound.org/people/Samulis/packs/12231/
- pros: very clear, cc-by-3.0, little to no distortion even with the f (loud) samples
- cons: not enough range in version on Freesound (will investigate full VSCO library on main site)
https://freesound.org/people/Theriavirra/sounds/269735/ https://freesound.org/people/Theriavirra/sounds/269750/
- relatively long enough
https://freesound.org/people/Theriavirra/sounds/319918/
- possibly too short
https://freesound.org/people/Theriavirra/sounds/269738/
- too short
https://freesound.org/people/Seidhepriest/sounds/234364/
- cons: nc
https://freesound.org/people/Benboncan/sounds/72699/
- pros: cc-by-3.0, very clear cuckoo sound
- cons: need to cut and stitch file for >7 cuckoo calls, sounds less natural than 35053
https://freesound.org/people/reinsamba/sounds/35053/
- pros: clear cuckoo sound, cc-by-3.0, background white noise at acceptable level (less than 121203)
- cons: cuckoo sounds too disjointed (somewhat uneven too, but disjointed is a bigger problem, as in the bird stops then restarts, a few times)
https://freesound.org/people/plantmonkey/sounds/376703/
- pros: very clear cuckoo sound, cc-by-3.0
- cons: soft
https://freesound.org/people/acclivity/sounds/19561/
- pros: very clear cuckoo sound
- cons: cc-by-nc-3.0
https://freesound.org/people/inchadney/sounds/121203/
- pros: clear cuckoo sound, cc-by-3.0
- cons: clearly perceptible background white noise
- citation: Walter, T. [Thomas]. 2011. Birds: Cuckoo.wav [Audio file]. Retrieved from https://freesound.org/people/inchadney/sounds/121203/
https://freesound.org/people/CaganCelik/sounds/435205/
- pros: clear birdsong, public domain (cc0)
- cons: doesn't sound like the stereotypical cuckoo
https://freesound.org/people/jgrzinich/sounds/364983/
- pros: clear cuckoo sound
- cons: cc-by-nc-3.0, clearly perceptible background white noise
https://freesound.org/people/RTB45/sounds/234921/
- pros: cc-by-3.0
- cons: noisy background (wind)
https://freesound.org/people/SiliconeSound/sounds/475049/
- pros: public domain (cc0)
- cons: clearly perceptible background white noise
https://freesound.org/people/soundbytez/sounds/111077/
- pros: cc-by-3.0
- cons: very loud when starting, doesn't sound like stereotypical cuckoo
https://freesound.org/people/MeicalJones/sounds/474062/
- pros: public domain (cc0)
- cons: clearly perceptible background white noise
https://freesound.org/people/Artemis_R_Swann/sounds/517480/
- pros: public domain (cc0)
- cons: soft
https://freesound.org/people/Martineerok/sounds/242710/
- pros: cc-by-3.0
- cons: clearly perceptible background white noise, soft
https://freesound.org/people/Pashee/sounds/193354/
- cons: cc-by-nc-3.0, very loud background white noise (wind?)
https://freesound.org/people/mike_stranks/sounds/397575/
- cons: cc-by-nc-3.0, other birds much louder than the cuckoo
George, C. (2014, March 2). Seth Thomas 8 Bell Sonora Clock [Video file]. Retrieved May 21, 2024, from https://www.youtube.com/watch?v=XJJHVLjVd2I
Harrison, D. (2000, October). Tolling Time. Music Theory Online, 6(4), Article 6.4.1. Retrieved September 10, 2021, from https://www.mtosmt.org/issues/mto.00.6.4/mto.00.6.4.harrison.html
Heath, G. (2011, June 26). St Mary Le Bow, London's Bow Bells - Clock chimes [Video file]. Retrieved May 21, 2024, from https://www.youtube.com/watch?v=rk5NvXNPUPs
Li, A. (2018, January 8). sysline beeps are music, or the multisensory Unix text mode interface [Weblog post]. Retrieved from http://incd.ambroseli.ca/node/308
Oliver, R. (n.d.). Clock Chime Tunes: History and Examples You Can Hear. Retrieved September 10, 2021, from https://www.clockguy.com/SiteRelated/SiteReferencePages/ClockChimeTunes.html
Pearce, C. W. (1903, December 1). Occasional notes [Letter to the editor]. Musical Times, 44(730), 790. Retrieved from https://ia600708.us.archive.org/view_archive.php?archive=/22/items/crossref-pre-1909-scholarly-works/10.2307%252F903901.zip&file=10.2307%252F904253.pdf
Starmer, W. W. (1907, November 19). Chimes. Proceedings of the Musical Association, 34, 1–24. Retrieved from https://ia800708.us.archive.org/view_archive.php?archive=/22/items/crossref-pre-1909-scholarly-works/10.1093%252Fjhered%252Fos-3.1.263.zip&file=10.1093%252Fjrma%252F34.1.1.pdf
Starmer, W. W. (1910, April 19). Continental chimes and chime tunes. Proceedings of the Musical Association, 36(1), 93–107. Retrieved from https://ia800708.us.archive.org/view_archive.php?archive=/22/items/crossref-pre-1909-scholarly-works/10.1093%252Fjhered%252Fos-3.1.263.zip&file=10.1093%252Fjrma%252F36.1.93.pdf
Westminster Quarters. (2021). Wikipedia. Retrieved August 18, 2021, from https://en.wikipedia.org/w/index.php?title=Westminster_Quarters&oldid=1038660380
Whittington chimes. (2021). Wikipedia. Retrieved August 18, 2021, from https://en.wikipedia.org/w/index.php?title=Whittington_chimes&oldid=1039122189
Wilson, S. [@SrWilson]. (2005, February 12). The actual melody which became later known as the whittington chimes is much older than the westminster chimes [Reply to forum post]. Retrieved September 1, 2021, from https://mb.nawcc.org/threads/when-did-the-whittington-chime-get-developed.17426/
Wilson, S. [@SrWilson]. (2006, August 9). I think I know what your clock does [Reply to forum post]. Retrieved September 1, 2021, from https://mb.nawcc.org/threads/jauch-grandfather-movement-with-11-note-chime.27915/
Zahnke, T. (2020, January 25). Hello, I am a totally blind composer, and I'v tweaked LilyPond so it's accessible for me [Facebook post]. Retrieved September 10, 2021, from https://www.facebook.com/groups/gnulilypond/permalink/10157912670783529/
Zimmerman, C. S. (n.d.). Great Bells of the British Isles. Retrieved September 28, 2021, from http://www.towerbells.org/data/GBGreatBells.html