Jump to content

How to Define When a Singer Hits a Wrong Note

Rate this topic


to_the_sun
 Share

Recommended Posts

I’m interested in doing real-time analysis on vocals in order to rate how "in tune" they are, using a programming language called Max/MSP, which I've been working with for a couple of years now. Something that’s sophisticated enough to be able to point out when the singer hits a wrong 
note.


I'm on this forum to address a more basic question before I can begin and that's how exactly to quantify "when the singer hits a wrong note". How would a singing professional, like a vocal coach, define a missed note? I'm talking as far as pitch only, not tempo or anything. It may seem like an obvious question, but it's really not.

There are plenty of programs out there already that detect pitch and perhaps compare it to a reference melody in a karaoke sort of fashion, but that’s not exactly what I’m asking. My question is: how does one take completely freeform input and decide if there are mistakes in it?

As a test, I thought I could use a clip of some a cappella singing, something professional that definitely doesn’t have any off notes. However when I run my pitch tracking analysis on it, it seems that the singer spends an equal amount of time in the gray area between pitches as near or directly on them. I suppose this is because the singer’s voice slides all over the place and rarely stays completely stationary for long. So what exactly makes a note sound bad, when a skilled singer is using vibrato and sliding through all sorts of off key pitches all the time anyway?

My hypothesis is that it’s the moments that the pitch does stay stationary that count; the mind doesn’t really pin anything down as being on or off key as long as it keeps moving. Kind of like how you can vibrato your way out of a shaky landing at a pitch and still make it sound alright.

Is this a sound assumption to make? Any other insights into these questions?

Link to comment
Share on other sites

    That is a hard question and is not going to have any solid answers that can be graphed by software. A human voice creates partials, certain frequencies that are different from the main pitch. Sort of like the idea of musical chords. Root, 3rd, 5th ...... A voice can sound off pitch because some of the partials are not harmonious with the music being played in the background or the fundamental pitch itself.

   For Acapella the association between the next adjacent pitches would determine whether the note sounds OFF or not, Along with whether the partials produced by a voice is "Tuned" to itself. Most of this is regulated by intuition by the singer.

   Western music in particular uses a system of "Just" tuning. Which is a compromise between "mathematical true frequency harmonics" and a division into a twelve tone scale. To put it plainly a G chord is going to have a different "Sound" and feel than a C chord because the mathematical relation between the individual notes is not exact. Some of those notes will not be perfectly "In Tune" with the others.   The G note that resonates with the G Chord may be slightly different with the G note that resonates with the C chord.

   This is one of the things that bugs me about "Pitch correction". The pitch is not supposed to be mathematically perfect.

 

Link to comment
Share on other sites

I think determination of proper pitch is best described as a cloud of pitches that the singer hits during the duration of the note they're singing (pretty much what you said). So there is probably a sort of "center" of the note being sung that sits inside the cloud. But most people do not have perfect pitch or even extremely good relative pitch, and the level of pitch awareness really varies among people. The idea of "on pitch" is basically an individual preference for the size constraint of that pitch cloud when singing or listening to others sing (or play violin or tune a guitar or whatever). Some people can tolerate a really large variance in pitch that's considered a certain note (F4) but others may be pickier, thus preferring a tiny rather precise cloud of acceptable pitches for a certain note.

You can find a note a singer is singing in a song like a A4 somewhere in a song, and then from when the singer starts the note till when they end, if you took list of all the different pitches (like if you want a A4 to be 440hz, but they probably went to like 438-442) they hit, you could basically compile a cloud of pitches and see how much variance there is in their pitch on that note.

Giving credit to Sadolin (CVT) for the graphic:

JWTb4BN.png

I hope that helps confirm your beliefs. You were basically thinking of using the scientific method to try to figure out the above graphic.

Link to comment
Share on other sites

I would like to add that it should be specified if you want to detect out of pitch or out of key. Because many times during a performance singers choose not to sing a given note and go for a substitution that even though in the same key is not the original note recorded.

Link to comment
Share on other sites

Thanks for all the responses.

Most pitch tracking methods will isolate/estimate the fundamental and use that, however I could take partials into account with an FFT analysis or the like if necessary. I don't think it will be though, for now at least.

Yes, I should specify that what I mean is "out of key". The subjective nature of music is fascinating to me and I understand that a key is a completely relative thing. I intend to acquire the scale being used at any given time heuristically, as the singer goes along, by assembling every frequency reading into a graph. This reveals a sort of "spectrum" unique to each scale and is sensitive to things like just tuning or intonation drift. For example, this image is of the "spectrum" of one of my test songs which is in C minor.

From there I need to determine key moments at which to compare the current frequency reading with the peaks in the "spectrum", deriving a rating based on a sliding scale of proximity, not as a black and white matter of being in key or out of key.

Would you agree with my intuition that the moments when tuning matters most are the moments of relative stability in pitch?

Untitled.png

Link to comment
Share on other sites

 I have been known to be pitchy, so i may not be the best to respond. If I feel that I need to work on something I may hone in on the beginning pitch and which ever pitch seems to be the defining note in a phrase. Usually the one held for the longer time.

  I concentrate more on the cadence and timing of a phrase rather than pitch by pitch. I usually adjust pitch more or less by "volume" or "inflection". Let the rhythm or emotional intent guide the pitch. Only when things sound "OFF" do I go back and really work on getting the pitch correct.

Link to comment
Share on other sites

Thanks MDEW, those are exactly the sort of expert insights I was looking for.

I suppose there is no one thing that determines the "important" moments in a vocal melody. I think onsets in volume play a greater role that I was originally thinking. Like you said, the beginning note of the phrase and perhaps others that are emphasized. It's the smooth melismas that will be the trickiest to identify.

Link to comment
Share on other sites

Also, like you were saying about cadence and timing, I've already written and am perfecting a program that determines your rhythm, and it's true that shaky or off-key notes are rare when you're in a good rhythm. So a lot of what I'm doing now is just a matter of fine-tuning.

Link to comment
Share on other sites

well with any song there is a key.. Some modulate and move the key but for the most part there is a specific key. Now in that specific key there are notes that are in and notes that are out of the key. Unless its some free form jazz/or type of music and you want to create dissonance..  Now what makes a singer out of key or sound out of key is when you have a note over another note or chord which is not in the key of the song or they land flat or sharp of the notes in the key they were intending to hit.  We hear it vibrate in an off oscillation in our ears and it makes your head turn sideways like talking to a puppy.. thats out of key or out of tune...

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...