But the Tuner Says....

[harrisonreed]

Abstract:

https://asa.scitation.org/doi/10.1121/1.5131244

Full article:

https://www.google.com/url?sa=t&source= ... 2963312343

Next time you're in the stratosphere wondering why the tuner says you're sharp, or wondering why the piccolos sound so bad even though the conductor is holding a tuner in their face.

I was working on a project today, running audio through melodyne to try and clean up some of the chords. It was funny - setting everything up lock on to the piano roll (12Tet) sounded TERRIBLE. Not just, "oh, use just intonation and lower the third should fix it" terrible, but horrendously out of tune. [the following turned out to be incorrect] It turns out the raw audio that was being played way up high was already in tune with the bass trombone octaves below it (which was right on the money in 12Tet), but only when the upper part was 20-30 ¢ sharp (as recorded).

EDIT::: It turns out I'm an idiot who forgot what pitch convention we were in. There was still a spread, but it was more like 15c, not 30.

Crazy!

[robcat2075]

Well, this is a big can of worms isn't it?

Not only do you need a stretch but it's a different stretch among different instruments, different dynamics, different octaves and different listeners!

The chart on p. 16 is alarming.

I can imagine this phenomenon has been the seed for many knock-down-drag-out fights over who was out of tune, even before there were tuning machines.

I'll note that many software tuners allow you to choose a stretch or even define your own.

[Doubler]

I wondered why playing to a tuner didn't sound right musically, frustratingly so in upper registers. This explains it well. It also suggests a different way of using a tuner, more as a way of analysis and less as an example/goal. Thanks for posting this information, harrisonreed!

[BurckhardtS]

Stretched octaves sounding better to our ears is not news, and it's nice to see a study done on it depending on the instruments... time to read!

I've heard so many people say 'there's in tune or out of tune, nothing inbetween', including seasoned professionals and that's just not really accurate. Context determines everything. If you're playing with a piano and trying to play perfect just intonation, it's gonna sound horrible. If you are playing chords in a section setting and don't adjust, something is not going to sound right. If you play a melody exposed using just intonation, it's also not going to sound good either.

I think it's important to learn all of it - good tuner pitch, good melodic tuning, good equal temperament, and good intonation for drones/chords, and learn how it sounds, why it sounds good, and then play what 'sounds right' for the context. The context can change within the ensemble, piece, or phrase, too!

[robcat2075]

Note the difference between a piano stretch, the formula stretch, and the listener-perceived stretch.
.

Curve.jpg

[harrisonreed]

I have read another article that suggests this is due to the way a bone in our ear vibrates (also why there is a variability in preference - humans can be incredibly sensitive to pitch but it'll be within the frame of their imperfect hardware).

I think what is critical here is that this stretch is an average, so as you go up in a melody it will stretch up with that average curve, but in and around that average you still have to temper -- pythagorean for the moving lines and intervals, and just intonation for held notes in chords

[harrisonreed]

I was thinking more about this. Typically as trombonists we will tune to A220 or 221 and not A440 (second space in the treble clef). According to this we probably are tuning about 5c too sharp if we tune exactly to A220. But we do it. And upper instruments do not stand a chance in dictating pitch against the base of a pyramid. You can imagine a tuba is even sharper against this theoretical railsback ish curve if they tune to their tuner. So this already will force the upper winds and piccolos higher. The oboe should be playing an A440, but it will sound flat against an orchestra below it playing a 12tet A220.

I was working more on my project and it indeed sounds in tune but the stuff in the staff is falling within a few cents of 12Tet (accounting for just intonation) and the stuff around A440 and above is "sharp". I can see where the upper winds would just get pushed way sharper having to play with trombones that are centering their curve on A220.

[Doug Elliott]

I don't understand exactly what those curves represent, except I looked up Railsback and see that corresponds to how pianos are tuned.

In any case, the differences are small or nonexistent between 220 and 440, so I think you're exaggerating the potential problem.

[SwissTbone]

Tuners are always right. But they always lie. Does that help?

[harrisonreed]

Tuners are always right. But they always lie. Does that help?

I just thought it was an interesting article. I never use a tuner. It doesn't tell you anything useful

[harrisonreed]

I don't understand exactly what those curves represent, except I looked up Railsback and see that corresponds to how pianos are tuned.

In any case, the differences are small or nonexistent between 220 and 440, so I think you're exaggerating the potential problem.

I think you're right. A couple cents might not change much that ears won't fix. What about the tuba, tuning another octave lower? They should be solidly flat in order to sound correct (according to this article) but tuba players will tune to their tuner as well.

[Doug Elliott]

A2 (110hz), A3 (220hz), and A4 (440hz) are all virtually in a flat line on that graph.

[WilliamLang]

we might be talking about the difference between psychoacoustics (or perceived pitch) vs. scientific absolute pitch. i think most people should use a tuner more for middle Bb each day, but not for specific notes in ensemble playing as much. rather use a drone for intonation work and develop the ear as much as possible.

[harrisonreed]

A2 (110hz), A3 (220hz), and A4 (440hz) are all virtually in a flat line on that graph.

You're looking at the dotted line from a grand piano tuning program on a piano strobe. The study results are the grey shaded line that dips well below everywhere left of A4. A3 is shown as sounding best 8c flat, A2 is shown as 15c flat.

I'm not saying this study is right. It seems like it might be now that I've got advanced pitch/chord software. But if it is right, and tubas tune to A110 instead of 15c below it, and the strings and clarinets tune to A4, but are tricked by the sharp sounding tuba, 15c is a big shift when you're already hearing A6 20c sharp.

It is probably a complete coincidence, but A442 is about 8c higher than A440, the same tuning difference between preferred values for A4 (the oboe) and A3 (the trombones) according to that chart.

[AndrewMeronek]

Reading through the introduction, I would think one of the first things to bring up should be to clarify if he's talking about stretching intervals of pitches sounded simultaneously or separately. But I'm not finding that detail. We know this can make a huge difference.

[robcat2075]

The lede of the study is that the perceived stretch for orchestral instruments is not the same as the conventional piano stretch and it is not virtually flat.

This is the chart from pg. 16
This is the result of what their listeners perceived to be properly tuned octaves over different spans. "Subjective Octaves" they called it.
.

Chartpg16.jpg

.
8 cents per octave (for brass) might not seem to be a lot but over the 3 octaves a trombone choir typically plays or the 5 octaves a brass section plays... that can't be insignificant.

The study notes the variation of human perception. This chart plots all the listener responses for all instruments. even the dense clusters near the centers make for a rather wide span of opinion.

Fig2.jpg

[Kdanielsen]

I wonder if this is why playing along with MIDI orchestra (Alessi’s “orchestra machine”) feels off to me sometimes. I haven’t been able to put my finger on it but something with the tuning feels weird. Maybe it’s the lack of spread?

[WilliamLang]

orchestras aren't meant to be in equal temperament, so it's going to sound weird no matter how good the sample library is

[Kdanielsen]

orchestras aren't meant to be in equal temperament, so it's going to sound weird no matter how good the sample library is

Yes definitely. I kind of assume it's going to feel a bit like playing with piano, but there is another level of weirdness that is very distracting.

The sample library used isn't great. I wonder if there are sample libraries (or insert computer words) that account for spread. I know there have been some that do just intonation on the fly.

[Doug Elliott]

OK, I didn't look at the actual study but that all makes perfect sense in terms of perceived pitch. And the spread of opinions.

Slightly different but related subject:
What I want to see studied is the perceived pitch differences between right and left ears. I'm quite sure there is often a difference, and the greater the difference, the more problems in pitch recognition and pitch matching - I think that's probably related to people being "tone deaf" - confusion due to hearing radically different pitches in each ear.

[Burgerbob]

OK, I didn't look at the actual study but that all makes perfect sense in terms of perceived pitch. And the spread of opinions.

Slightly different but related subject:
What I want to see studied is the perceived pitch differences between right and left ears. I'm quite sure there is often a difference, and the greater the difference, the more problems in pitch recognition and pitch matching - I think that's probably related to people being "tone deaf" - confusion due to hearing radically different pitches in each ear.

And even more important for trombonists and some string players, who are going to hear their instrument with one ear predominately.

[Doug Elliott]

Exactly

[harrisonreed]

So, first, I have to admit a mistake. When we recorded the project I was working on, we were tuned up the way we always do (since we work with musicians in Japan frequently), so we were at A444. When I went to work on the project initially, the software was set to A440. No wonder locking everything in sounded god awful. Not only does 12TET not sound right, but the entire recording was distorted from a pitch shift. Sorry about that. However, I did notice that the bass seemed to be flatter than the upper parts as I analyzed where we were playing at.

So I tried an experiment with these different pitch conventions, and I think they sort of make me question this study. For reference, here is the original audio I recorded, with us playing without "corrections":



Here it is in 12TET, artificially edited to have every note locked on exactly to true twelve tone, equal temperment:



Here it is in mechanically perfect 5-limit just intonation, where the notes are again artificially locked on where math says they are "perfect":



Here is is, finally, with the tuning scale altered to center on the graph we've all been looking at, where A4 is in tune, and A3 is 8c flat, and A2 is 15c flat, and A5 is 8c sharp, etc. Again this is artificially locked on, so it's in equal temperment, but stretched to the curve from the graph:



I think the one based on the graph is the weakest. It sounds OK, and might sound even better if I figured out how to temper each note WITHIN the stretched tuning, but I don't think the study was on to anything as special as I thought it was.

I think I prefer the unaltered version, although the 5-limit is also very nice to my ears.

[Doug Elliott]

The last file needs permission

[harrisonreed]

The last file needs permission

Think I fixed it!

[elmsandr]

OK, I didn't look at the actual study but that all makes perfect sense in terms of perceived pitch. And the spread of opinions.

Slightly different but related subject:
What I want to see studied is the perceived pitch differences between right and left ears. I'm quite sure there is often a difference, and the greater the difference, the more problems in pitch recognition and pitch matching - I think that's probably related to people being "tone deaf" - confusion due to hearing radically different pitches in each ear.

Just a fun anecdote here about this.. I played bass bone in a band with a drum set where we set up with me right next to the kit. Horrible, but I put in an earplug on the side facing the drums and everything worked fine. Generally got great feedback from leader and other players, bass player in particular telling me that he was happy with my positioning relative to him as I was a ‘rock’ for him for intonation. Then our physical space was re-vamped and we had to mirror the group and put the drum set on my other side. Fine, I thought, just move the earplug to the other ear. Ugh. I haven’t spent so many months playing so badly out of tune. Got a lot of confused looks from the leader and the same bass player. Tried a lot of things, but could never quite figure it out. Even leaving out the earplug only did so much to help, and then my ears hurt.

Unfortunately, that gig ended a long time ago and isn’t likely to return, but it remains a painful memory for me on how to trust or not trust my own ears. Always be aware of the setup and where you are hearing from. Even for your personal hearing and reactions, intonation is a team sport.

Cheers,
Andy

[BurckhardtS]

Another thing that I thought of is the pitch differences depending on what side of the bell you're on. There are times I feel like I'm pretty close and then the pitch I hear back on the recording is definitely not what I heard come out of my horn.

Andy, I've had a similar experience where playing bass bone in a very good big band, where I was right next to the drum set on my right and I was having issues locking in the pitch with the bass player, but when I was on the other side of the band, it was much easier to tune.

[Doubler]

I think I prefer the unaltered version, although the 5-limit is also very nice to my ears.

Not that anyone asked...

Original sounded best - balanced and resonant.
Perfect equal temperament sounded thin.
Mechanically perfect 5-limit - bass sounded overly resonant.
Stretched sounded messy.

I'd guess what happens between the ears is the critical factor here.

[Doug Elliott]

On my crappy computer speakers I couldn't hear any difference

[BurckhardtS]

It was all pretty close. First one was best overall, perfect ET sounded good except on a couple of long notes and chords. Mechanically perfect I had some issues with melodic intervals in the melody when it didn't land on chords, but the chords sounded great. The last one I think had both issues with some chords and also a few weird melodic intervals. I'm being nitpicky, because it really was close.

[harrisonreed]

Yes, the mechanically perfect 5-limit takes away the melodic/linear/pythagorean tempering of the moving lines.

[AndrewMeronek]

A trombone ensemble doesn't really cover the range discussed in the paper, though. You'd need to add a couple of more octaves.

[harrisonreed]

A trombone ensemble doesn't really cover the range discussed in the paper, though. You'd need to add a couple of more octaves.

Within the range covered in that recording the chart gives about a 25c spread. It sounded off in the bass part.