Speed Correction Tools

[Dan Gould]

OK, DIME people and other audio experts:

How do you pitch/speed correct recordings that are fast/slow aka sharp/flat?

My Goldwave has a time warp option to slow/speed sampling rate as a percentage of the original ... and I am trying first to match the stage announcement from earlier in the tape to determine how much correction is "right".

Any better options/software suggested???

As always, thanks in advance.

[JSngry]

What are you using to reference the correct key to make sure that it really is correct?

For corrections, would ratios help as a starter?

[Dan Gould]

Only using my ears to recognize that our special someone tenorist sounds a little more boyish on the last stage announcement than on the first. Was going to alter pitch via time warp to match the vocal, then apply to subsequent tune and see what we get.

I am also having a hard time spotting an exact moment of speed change; tape speed didn't change on the flip as his stage announcements are still closer to basso profundo on set 2.

[Dan Gould]

On 10/16/2020 at 3:09 PM, JSngry said:

What are you using to reference the correct key to make sure that it really is correct?

For corrections, would ratios help as a starter?

If its a blues in F but it sounds like F#, does a bear crap in the woods?

No seriously, can you express the adjustment of F# to F on a percentage or ratio basis?

[JSngry]

That would be down a half-step, so -16:15 = - (16/15)= -1.07%...am i doing that right?

[Daniel A]

Rather, you should stretch the file to 1,0594623228... times its original length to make it sound a halftone lower. 

I'm not enough of a mathematician to tell whether the ratios in the table above are correct, but if you divide any two pitches a halftone apart you get 1,059... etc

https://pages.mtu.edu/~suits/notefreqs.html

Then, maybe the recording is not exactly a halftone sharp, so there will be some trial and error anyway. 

Edited October 26, 2020 by Daniel A

[Dan Gould]

5 minutes ago, JSngry said:

That would be down a half-step, so -16:15 = - (16/15)= -1.07%...am i doing that right?

Well slower speed lowers the pitch, so 15/16 = .9375 would work in terms of my audio software.

You know I sent you the original and speed corrected for a different track. What do your ears say the correct pitch would be? The change I applied was more than a full step actually (.85) -  Allen hears the corrected as in the key of B Flat and sounds right to him.

3 minutes ago, Daniel A said:

Rather, you should stretch the file to 1,0594623228... times its original length to make it sound a halftone lower. 

I'm not enough of a mathematician to tell whether the ratios in the table above are correct, but if you divide any two pitches a halftone apart you get 1,059... etc

https://pages.mtu.edu/~suits/notefreqs.html

Then, maybe the recording is not exactly a halftone sharp, so there will be some trial and error anyway. 

I think its the nomenclature of my software. .9375 will make the recording longer, and lower the pitch.  

[JSngry]

2 hours ago, Dan Gould said:

Well slower speed lowers the pitch, so 15/16 = .9375 would work in terms of my audio software.

 

Since an octave has a frequency ratio of 2, a half-step has a frequency ratio of 2^(1/12), or approximately 1.0595.

So 100%-1.5095 % = 98.4905% or .9849

Sound right, at least in terms of math?

I'm too old to think this hard...

 

 

[Daniel A]

1/1,0595=0,9438 🙂

Edited October 27, 2020 by Daniel A

[Dan Gould]

Fact is I have two options with my software. I can make the recording longer, by a factor of 1.0595, or I can make the recording slower, and I appear to have three different calculated percentages for that option. I guess if I go slower, the percentage that duplicates the new timing at 1.0595 would be the right one.

 

Place your bets!

And thanks gents.