VST Code generation issues with element-wise multiplication

2022 11 11
1 个回答
更新时间：2022 11 11
2 次查看（30 天）

0 个投票

The following code produces audible discontinuities when compiled to a VST:
gains = [0.5, 0.725];
out = in .* gains;
But this works fine
gains = [0.5, 0.725];
out = [in(:,1) * gains(1), in(:,2) * gains(2)];
Is this a known issue with code generation?

请先登录，再回答此问题。

回答（1 个）

VBBV
VBBV 2022-11-11
Ran in:

0 个投票

Ran in:
gains = [0.5, 0.725];
in = rand(8)
in = 8×8
0.8849 0.9486 0.2943 0.5330 0.5551 0.2483 0.9058 0.8506 0.6508 0.8914 0.3171 0.4493 0.6775 0.0840 0.3678 0.6322 0.2078 0.4190 0.9428 0.2469 0.3487 0.5543 0.0761 0.7525 0.8023 0.4535 0.7250 0.9433 0.3312 0.5522 0.5628 0.8056 0.2804 0.6671 0.6911 0.2097 0.8895 0.6512 0.3580 0.4607 0.2275 0.3705 0.9019 0.3368 0.4651 0.0667 0.7093 0.3143 0.8744 0.2188 0.9651 0.2883 0.7966 0.2366 0.2595 0.6238 0.6302 0.3316 0.5598 0.3719 0.0362 0.0559 0.2016 0.5517
out = [in(:,1) * gains(1), in(:,2) * gains(2)]
out = 8×2
0.4425 0.6877 0.3254 0.6463 0.1039 0.3038 0.4011 0.3288 0.1402 0.4837 0.1137 0.2686 0.4372 0.1587 0.3151 0.2404
out = in .* gains % check the matrix multiplication rule
Arrays have incompatible sizes for this operation.

3 个评论
显示 1更早的评论 隐藏 1更早的评论

VBBV
VBBV 2022-11-11
编辑：VBBV 2022-11-11
when the cols of 1st matrix is equal to rows of 2nd matrix then it works fine (not exceeding the maximum array size value)
Robert
Robert 2022-11-11
Ran in:
Ran in:
Thanks for the quick response. I agree with this, but the default VST is 2 in, 2 out. This code should work if that were true:
gains = [0.5, 0.75];
in = [ones(5,1), ones(5,1)*2];
out = in .* gains
out = 5×2
0.5000 1.5000 0.5000 1.5000 0.5000 1.5000 0.5000 1.5000 0.5000 1.5000
Is it possible that the VST doesn't always default to stereo like the documentation says?
"By default, both the input to and output from the process method have two channels (columns). The number of input rows (frame size) passed to process is determined by the environment in which it is run. The output must have the same number of rows as the input.", found here:
https://www.mathworks.com/help/audio/gs/audio-plugins-in-matlab.html
Jimmy Lapierre
Jimmy Lapierre 2022-11-11
I was able to reproduce the issue. I will investigate further.
>> p=loadAudioPlugin('GainTest.dll')
p =
VST plugin 'GainTest' 2 in, 2 out
>> process(p,ones(4,2))
ans =
0.5000 0.7250
1.0000 1.0000
1.0000 1.0000
1.0000 1.0000
Please use the workaround in the meantime.
p.s. modifying the input in-place is allowed, so it could look like this:
function x = process(~,x)
gains = [0.5, 0.725];
x(:,1) = gains(1)*x(:,1);
x(:,2) = gains(2)*x(:,2);
end

请先登录，再进行评论。

类别

在 帮助中心 和 File Exchange 中查找有关 Audio Plugin Creation and Hosting 的更多信息

标签

提问:

Robert
Robert
2022-11-11

评论：

Jimmy Lapierre
Jimmy Lapierre
2022-11-11

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by