Hi Miller,
Thanks for your quick answer and sorry for the noise, I posted too fast.
I wasn't seeing the obvious reason : of course I should have used [tabread4~] which I've always been using before !
Silly me ...
Best,
Joseph

Le 21/07/2025 à 15:54, pd-list-request@lists.iem.at a écrit :
Hi Joseph -

There's no resampling in tabread~, but if you're using line~ to drive it 
(for instance, via the
messages 0, 1000 44100) - well then, if the patch is running at a 
different sample rate
that will cause brutal, non-interpolating resampling.  So it could be 
simply that you're
running the patch for the first time at a different sample rate than it 
ran in before.

If you want the patch to give reasonable results at any sample rate, 
you'll have to
move to an interpolating lookup such as tabread4~.  In this way you can 
use the same
(0, 1000 44100) type message to line~ and get the same pitch and 
duration of output
as before - with a cleaner interpolating algorithm.

cheers
Miller

On 7/21/25 3:44 PM, Joseph Larralde wrote:
Hello,
I was just patching a small piano sampler with 0.55-2 and noticed that 
[tabread~] seems to be doing some internal resampling of the array it 
refers to to match pd's samplerate without transposing the sound.
I tested back to 0.52 and it is still doing it.
I could swear I always took care of correcting the sample rate myself 
in abstractions, but I have an external that does a similar thing so I 
find it an appreciable feature.
However, the resampling sounds very cheap, like linear interpolation 
or drop-sample : I notice strong artifacts with piano samples.
Is this intended or could it come from my system (Pd 0.55-2 on M3 
Ventura 13.5) ?
Has anyone expreienced this behaviour before ?
Joseph