Quick Sampler “Optimise”

[GuyBorlander]

Hey! I have been writing a sub bass line for an acapella I ripped from off of an old dance record - the baseline is in key but there is one note on the bass line that feels slightly off I came to the conclusion that the vocal has been shifted slightly higher in pitch

I dropped the vocal into quick sampler with the optimise option and it has deducted the vocal by -35cents… the bass line also now sounds perfect and none of the notes feel slightly off.

1. Am I right by saying that the optimise function examines the file and then gets it perfectly in key to whatever key closely matches the file… how accurate is it? 

Now I know this vocal is definitely slightly pitched up I’m thinking of manually deducting cents to this vocal by ear and see if I can get an even better result-  but this optimise function seems bang on and I will probably leave what it has done. I’m pretty blown away by how good it is.

2. How was this even designed in the first place? 

thanks for your response as usual guys.

kind regards,

joe

[des99]

6 minutes ago, GuyBorlander said:

1. Am I right by saying that the optimise function examines the file and then gets it perfectly in key to whatever key closely matches the file… how accurate is it? 

Yes, it'll analyse the tuning, and likely if it's a few percent off of a note, it will pitch it slightly accordingly. As for accuracy, it probably depends on the audio. You can use a tuner plugin to check the original and the optimised version if you want to examine it.

6 minutes ago, GuyBorlander said:

2. How was this even designed in the first place? 

It comes from Keymap Pro, which was an (awesome) independent EXS24 sample editor, until the company and the developer were brought into Apple. It had all kinds of automatic tools to do pitch analysis and automatic tuning, automatic looping, normalising and (a lot of!) other functions. A lot of that tech is in Sampler and Quick Sampler now, and other tech from the same developer is also in Logic, like Autosampler...

[GuyBorlander]

4 hours ago, des99 said:

You can use a tuner plugin to check the original and the optimised version if you want to examine it.

the audio is from an acapella b side ripped from vinyl highest quality possible - it’s a clean recording good performance too. 

what tuner plugin can I do this? Maybe Next time I can do this first and then just manually deducted cents in quick sampler by ear.

Thanks for the reply

[des99]

28 minutes ago, GuyBorlander said:

what tuner plugin can I do this?

Logic's "Tuner".

[GuyBorlander]

2 hours ago, des99 said:

Logic's "Tuner".

Will this find the accurate key of the entire vocal sample or will it just lock into the last note of the sample as it plays out? I’ve used it whilst tuning individual notes before but not to detect the key of an entire sample.

[des99]

It’s a tuner, it gives you a pitch readout of the audio that is playing through it, in the same way a physical tuner does.

[David Nahmani]

44 minutes ago, GuyBorlander said:

Will this find the accurate key of the entire vocal sample or will it just lock into the last note of the sample as it plays out? I’ve used it whilst tuning individual notes before but not to detect the key of an entire sample.

It gives you a real time reading, so if the sample's pitch changes during the length of the sample, the needle will move to reflect this. 

[GuyBorlander]

the tuner plugin is giving me a real time reading which is great but it doest tell me the key of the sample so I can't use it for what I'm trying to do - also I've realised that the quick sampler optmize setting is actually inconsistent every time I drag and drop in the audio file; I'm just going to have to use my ears to pitch this sample against my sub bass - nothing else seems to be working better than just using my ears. not sure if there is a solution to this.

Kind regards,

Joe

[des99]

The tuner gives you the pitch of a note, which is what you are looking for to make sure the sample is in tune. The "key" is a different concept, as is changing the musical key or material played/recorded in a different key.

There is nothing wrong with using your ears. The listener to your music isn't running it through meters to check how accurate in pitch it is, so the ears should *always* be the final arbiter.

[GuyBorlander]

5 minutes ago, des99 said:

The tuner gives you the pitch of a note, which is what you are looking for to make sure the sample is in tune. 

the tuner has been giving me a real time reading as the sample is playing through but it isn't telling me how much the entire vocal has been pitch shifted away from a perfect AFlat which is the key of the track.  it is clear to me that this vocal has been pitch shifted up a few cents because my bass line written in the ket of AFlat sounds slightly off.

 

There is one note that is sung in the vocal which sounds particually off during a certain part of my written bass line - I cut this note out of the audio sample and played it on a new audio channel with the tuner plugin attached; the results are inconsistent every single time... its surprising because its just half a second worth of audio information.

(the real time reading of the entire vocal sample being played having inconsistencies is understandable, but surely not with half a second of one note being played.)

I use my ears initially and finally of course that is what I'm supposed to do - but it would be great to have consistent results from these meters.

[des99]

Just now, GuyBorlander said:

the tuner has been giving me a real time reading as the sample is playing through but it isn't telling me how much the entire vocal has been pitch shifted away from a perfect AFlat which is the key of the track.  it is clear to me that this vocal has been pitch shifted up a few cents because my bass line written in the ket of AFlat sounds slightly off.

A vocal pitch will fluctuate anyway, from note to note, and even during a note. The note it's closest to will register as the note value, and how far away the audio is from that note value is displayed on the meter. That will tell you exactly how far of A-flat the current note is.

3 minutes ago, GuyBorlander said:

There is one note that is sung in the vocal which sounds particually off during a certain part of my written bass line

Sure, maybe that note is more off-pitch than other notes. This is not abnormal in a vocal performance. A tuner is not going to give you an overall average reading of how off-pitch an entire vocal performance is - it's simply going to tell you the pitch of the current note, and is really designed in the same way as a guitar tuner. If you stick Mariah Carey gymnastics into it, it's probably not going to help you very much.

4 minutes ago, GuyBorlander said:

its surprising because its just half a second worth of audio information.

The tuner might not have a long enough note to lock onto the pitch of a short section in your example. You could try flex pitch, (or Melodyne), for better tuning by individual note features.

[GuyBorlander]

"The tuner might not have a long enough note to lock onto the pitch of a short section in your example. You could try flex pitch, (or Melodyne), for better tuning by individual note features."

great idea I will extend the note with polyphonic time stretch so it doesn't change the pitch and will let you know if it works.

is there an explanation for the optimise setting on quick sampler giving me inconsistent results? one time it detected a G3 and adjusted cents by -35 and then when I dragged the same sample in another time it detected GSHARP3 and adjusted cents by +29...

[des99]

11 minutes ago, GuyBorlander said:

is there an explanation for the optimise setting on quick sampler giving me inconsistent results? one time it detected a G3 and adjusted cents by -35 and then when I dragged the same sample in another time it detected GSHARP3 and adjusted cents by +29...

No, I don't have any details of the algorithm, nor know what audio you're dragging in that the algorithm is working on.

[GuyBorlander]

just letting you know that when I drag a perfectly cropped version of the audio sample where it starts immediately and cuts off right when it finishes gives me the GSHARP3 +29 cents result but when I import a version with a gap of silence about a second before the sample starts and after it finishes gives me the G3 -35 cents result.

Maybe the algorithm is more "accurate" when there is a a gap of silence before and after giving it time to "think" when scanning. The version with the gap before and after which gives the G3 -35 cents result sounds perfect against the bass line but the GSHARP3 +29 cents result is out of tune completely... maybe something to get the developers to look into.

Kind regards,

Joe

[oscwilde]

6 hours ago, GuyBorlander said:

Maybe the algorithm is more "accurate" when there is a a gap of silence before and after giving it time to "think" when scanning.

Not a realtime process, so time to think shouldn't come into it.
I'd imagine that the root pitch of the initial transient and/or fundamental tone (strongest harmonic) is being analyzed.....and averaged.
In a drum strike, for example, the actual stick noise is part of that hit, which would be a different transient to the actual "drum" sound.
I think you'd get consistent results with both the "gap" and also by setting the anchor slightly into the waveform.
This ^^^^^^^ is all somewhat academic - as long as you've got what you need, that's all that matters. 😃

[GuyBorlander]

11 hours ago, oscwilde said:

setting the anchor slightly into the waveform. 

Thanks for the reply - I’ve not heard of term anchor before what do you mean?

Cheers

[oscwilde]

19 minutes ago, GuyBorlander said:

I’ve not heard of term anchor

Absolute start point of the audio file.
https://support.apple.com/en-au/guide/logicpro/lgcp2158cc79/mac

[GuyBorlander]

14 hours ago, oscwilde said:

I think you'd get consistent results with both the "gap" and also by setting the anchor slightly into the waveform.
 

I don’t get exactly what you mean by this. The version that works for me in regards to it being seemingly perfectly in key with my baseline is the version with the gap before and after , I just print this gap of silence before the beginning and after the Audio file ends. I understand the concept of an anchor now but I cant visualise what you mean exactly with setting the anchor into the wave form? 
 

thanks for your patience,

Edited February 2 by GuyBorlander

[oscwilde]

Zoom in on the waveform in the Audio editor.
Move the anchor to the right so that it is NOT just before the (visual) waveform....but is actually placed after the (visual) waveform startpoint.

In essence, you're changing the point where waveform playback starts.....and before you go to the "but won't that cut off the start of the waveform" thought - this is WAY too short to be audible/for your brain to process it. Sure, you can see it....but no-one will "hear" it - not even those golden ears who claim they can hear pin drops from 500 yards or can pick up a millisecond of latency. 😊

This won't work for "every" type of sound/sample....but in the drum hit example, there's (potentially) a second transient...stick and drumhead.
Vocalists often have an audible inhalation or a plosive then the note, etc.

Whether or not this is significant enough to be recognised as a discrete transient by QS/Sampler's analysis is dependent on the audio material.

As a trick employed by the ancient.....when total hardware sampler memory was measured in 0.8 to 15 seconds...to ensure a totally "tight" hit of say a kick drum you'd sample at C3, then transpose down an octave or more to edit the sample start at the slower playback speed. These edits were generally done solely by ear or with 8 bit graphics on machines like the Fairlight (that no-one could afford).
When played back at normal speed, the kick would be punchy/bang on.