On-hold Audio recording change on tempo

[DaveC426913]

TL;DR Summary

Can an audio clip change its tempo because of compression?

Once every couple of months I have to call the Ministry of Health about fifty times over the course of a day. Their hold music is an electronic ditty that loops every minute-ish, so I hear it a lot. It's very bad quality. I don't know if it's originally digital or if the original was recorded on tape and digitized. (It is conceivable that it is quite old since this is the Health Ministry which is usually decades behind in technology. I believe some of our headless utility programs run on MS-DOS.)

Anyway, in the recording, the tempo of the music wavers slightly. It speeds up a fraction for just a second. (It's always in the same place.) It does not have a corresponding rise in pitch. If you were tapping out the beat to this, you'd have to tap a litter faster for a second to stay on beat.

I'm wondering where this change in tempo might have crept in. One hypothesis is that it was originally on tape and the tape stretched, but that would cause a slowing of the tempo and a drop in pitch, so that can't be right.

Another hypothesis is the heavy digital compression. Can heavy digital compression cause a slight change in tempo?
This is a part of the clip just for reference:
* I'm sorry it doesn't have the part where it changes tempo. I'll have to try again when the line is busier.
** I think it's pretty faint. I used OpenShot Studio to process it for uploading, which have some audio bugs in it.

 

[jedishrfu]

Some questions to ask:

- does it occur at a certain time of day? or only at a certain time in the music?

I'm thinking that certain times in the day the line frequency of AC systems rises for clock time corrections. Perhaps that affects the playback.

https://hackaday.com/2018/03/29/ask-hackaday-is-your-clock-tied-to-mains-frequency/

Another time of day condition is that people perceive music tempo as faster at night.

https://yonamariemusic.com/yona/blog/216/4-reasons-why-music-can-sound-faster-at-night

and then there's a deliberate change of tempo by the artis called tempo rubatot:

https://en.wikipedia.org/wiki/Tempo_rubato

Some radio stations or DJs may speed up songs slightly to give the impression of more energy in their broadcast or to match other tempos the DJ is trying to mix in.

https://arstechnica.com/civis/threads/am-i-crazy-or-do-some-stations-play-music-faster.587149/

https://pulsemusic.proboards.com/thread/58007/hit-radio-station-speed-songs

I recall hearing a song while waiting where a hiccup occurred at a specific point and suspected it was due to a tape loop splice either real or looping of a digital audio track.

 

[DaveC426913]

- does it occur at a certain time of day? or only at a certain time in the music?

Nono. It's definitely part of the recording. I'm on hold for up to ten minutes at a time, so I hear the loop ten times in a row.

and then there's a deliberate change of tempo by the artis called tempo rubatot:

Only if it were a fumble. It's only for a second or two.

Some radio stations or DJs may speed up songs slightly to give the impression of more energy in their broadcast or to

No. Again, it's only for a few seconds. And, if you listen to the clip, you can tell the music is pretty repetitive, so a couple of beats being a little faster is a bit jarring.

I recall hearing a song while waiting where a hiccup occurred at a specific point and suspected it was due to a tape loop splice either real or looping of a digital audio track.

Don't get me started on that.
I have a ringtone on my phone that's quite rhythmic. It's like 16 beats over 16 seconds, but the ringtone is, like, a 15 1/2 second loop. Grrr! They could have sped the ditty up by 5% or made the ringtone loop a half second longer.

 

[jedishrfu]

I looked for compression issues and couldnt find anything.

I did think that perhaps while doing the decompression there was a predictable garbage collection event.

Ive seen that with Java on some application runs.

 

[jedishrfu]

There was a story about the early days of audio compression tech where a particular very popular song failed bigtime causing the developers to rethink the algorithm.

I think it was a Suzanne Vega archipela song that was the culprit.

 

[Tom.G]

Ahh! I hear clip ends with that ubiquitous "Your call is important to us..." 'So we are putting you back on hold.'

(and here I thought it was just an American thing)

 

[Algr]

Many hold messages are downright sadistic. I got stuck on one once where every 20 seconds it would sound like someone was answering the phone and saying hello, only for the next words to be "...your call is very important to us...". And then it would dump you back into the same 20 second music loop. It seemed designed to maximize frustration and prevent you from doing anything productive while waiting! Horrible!

---

From an audio perspective, changing a song's tempo without changing its pitch is a complicated thing to do, and I doubt it could happen by accident. It could happen if the music was some kind of MIDI file, but I see no reason why hold music would function like that. An error while the hold music was being recorded is possible, but then why would the effect be intermittent. I suppose that sadism is as good an answer as any - they just want people to give up.

 

[jedishrfu]

As an aside, I found the Suzanne Vega connection. It was her Tom's Diner song being used as the testcase for the MP3 algorithm.

https://www.npr.org/sections/therec.../the-mp3-a-history-of-innovation-and-betrayal

The key takeaway was that some sounds were removed if they were overlaid by other sounds capitalizing on limitations of human hearing.

and this Verge article shows you what you're missing when the Tom's Diner song is converted to MP3.

https://www.theverge.com/2015/2/19/8068923/mp3-compression-ghost-suzanne-vega-toms-diner

 

[pbuk]

The key takeaway was that some sounds were removed

That's why it's called "lossy compression"!

Note there is an error in the linked article:

Maguire salvaged the sounds from "Tom's Diner" as it was shrunk from an uncompressed .wav file to a 320kbps MP3.

... if you read the original article https://www.theghostinthemp3.com/theghostinthemp3.html you will see that the source used was a 128kbps MP3.