Converting sampling frequency: best method?

DAW's, Audio, MIDI, VST's and all other music software.

Moderators: Derek, parametric, Saul

User avatar
EXer
Member
Member
Posts: 169
Joined: Tue Dec 11, 2012 6:15 pm
France

Converting sampling frequency: best method?

Unread post by EXer » Fri Oct 06, 2017 8:49 pm

I have extracted samples from a Roland S-550 sampler file and I have saved them to .wav files.

Now I want to use them in my EX5, but their sampling frequency is 30 kHz so I have to convert them to 44,1 kHz (I have tried them 'as is' but they sound horribly flat Image ).

My DAW asks me which interpolation method I want to use: linear, spline, 2nd degree polynomial, 3rd degree polynomial, 4th degree polynomial, narrow band, or none.

► Which method should I use to get the best possible result, given that those samples are single notes of a solo harpsichord?

It also asks me if I want to use anti-aliasing. I guess that for upsampling anti-aliasing is not needed?

Thanks.
EX5 EX5 EX5R FS1R TX81Z TX81Z TX7 A4000
EX5Tech Member #101
User avatar
SeaGtGruff
Global Moderator
Global Moderator
Posts: 662
Joined: Wed Jan 29, 2014 4:24 am
United States of America

Re: Converting sampling frequency: best method?

Unread post by SeaGtGruff » Sat Oct 07, 2017 8:28 am

Hmm, on a guess, I'd think that a polynomial would be more accurate, and the higher the degree the better, so I'd try 4th degree polynomial and see how that works. If you don't like the results, try one of the other options. I'd say that "none" should give you the closest thing to the original 30 kHz in a 44.1 kHz form, since the sampled values would just be repeated as needed rather than having the computer try to guess (interpolate) what the missing values might be.
Michael Rideout
Current keyboards: Yamaha YPT-400, PSR-E433, PSR-E443, PSR-EW400, MX49 BK
Current controllers: M-Audio Axiom 61-II
Previous keyboards: Farfisa Matador 611; Casio CTK-710
2112
Member
Member
Posts: 16
Joined: Sat Oct 27, 2018 9:04 pm
United States of America

Re: Converting sampling frequency: best method?

Unread post by 2112 » Sun Oct 28, 2018 1:25 am

EXer wrote:
Fri Oct 06, 2017 8:49 pm
but their sampling frequency is 30 kHz so I have to convert them to 44,1 kHz (I have tried them 'as is' but they sound horribly flat ).
I presume that when you wrote "horribly flat" you've meant "lost a lot of treble", not that "pitch shifted down".

In that case use 4-th degree polynomial interpolation. On an undamaged signal it sounds "glassy" and that glassiness may put back some sizzle that you may be looking for.

In the past there were lawsuits in the USA related to the upsampling technology between Sony and a boutique manufacturer of CD-players called Wadia. Lots of re-sampling algorithm since then were intentionally incorrectly implemented to avoid getting tangled in those patent disputes. Try your resampling algorithms on something you know very well by ear, like speech or singing by a friend or relative.
Post Reply

Return to “Music Software Forum”