“Trim Silence” not working with Auto Sampler

OhulahanBass

Title says it all. I’ve been doing allot of Auto Sampling over last few years and looking at my auto samples it doesn’t look like Trim Silence has worked for me going back to to 2.0.

Anyone else having this issue? Is there a known fix?

darkwaves

NI's definition of silence is annoyingly literal.

For example:

My hardware setup connects to my desktop setup via adat. There's a noise floor, but it's really, really low.

4 tests. All 4 normalize output. Hold the note for a second, record for 10 seconds. The first has no trim.

10 seconds as expected.

The second has a gate on my hardware. This is coming from a yamaha modx, into a motu 8a (where the gate is) and then into my desktop interface (rme ucx). Any noise after the gate closes is just the ADAT connection.

Machine cut off .02 seconds. If I normalize just the silence here; you'll see it's not actually "silent":

The final test is using a gate on my RME. In other words, cutting out even the noise from the ADAT connection:

Notice: 1.21s. Maschine cut it because it's "true silence".

Local gate without trim. I try to normalize the silence. If I found something, I move to the right and try again. At some point, I get the following:

It's a horrible error message, but it's basically saying, "You can't normalize silence"

The point here:

NI is defining silence literally. It's annoying, but they've always done this. It's the reason people have exports that have a couple minutes long tail of "silence". NI needs to define "silence" as something like -90db (or let us define it)

This concludes my TED Talk.

darkwaves

This would be easy to "fix" if we could sample "through" maschine. For example, I should be able to point the autosampler at pad 5. Set the audio input and midi output to my synth. Have a gate on the pad and then sample; but they don't give the option for this type of setup.

I suspect the only "fix" is handling with a gate at the audio interface level (which isn't an option if using just a mk3).

This annoys me so much that I posted 2x. lol

OhulahanBass

I am mainly sampling hardware for multi-sampled instruments. I am mostly annoyed at the 8-16ms of "silence" caused by latency at the beginning of all my samples. I have to do the math and adjust the "start point" in the sampler plugin and hope there is an available interval (.1, .2, .4%) that is close to what I need for that sound. I wish they had a threshold for "trim silence".

@darkwaves, it seems the only workaround in the current state of the maschine is to resample while running the original output through a gate plugin. Only issue, I have resampled multisamples before and witnessed a bit of distortion at the attack of each sample. I'll see if Mainstage has any of the same issues; I haven't used Mainstage:-( since Maschine released the auto-sampler.

Modern latency compensation would be good enough for what I am trying to do. The odd thing is the gap of silence at the beginning doesn't decrease with buffer or increased sample rate.

darkwaves

Interesting. I didn't realize it also trimmed the start. I jacked my buffer up to 2048 and tried the same tests.

With trim:

Without trim:

Trim - no gate - "silence" normalized:

Seems it's the same issue I'm complaining about. If NI would just have a little chill with their definition of "silence".

That said; you could probably batch process in a proper wave editor.

OhulahanBass

Interesting, got a wave editor to suggest?

darkwaves

Cubase has an option to strip silence; though I've never looked into the batch options. You'd be able to load all the files in a track, run the trim silence function and then export. I'm not sure how it would handle the file names, but that would be the direction I'd go simply because I use Cubase.

If I needed to do it often, I'd probably look into buying wavelab (because I'm a steinberg fanboi)

https://archive.steinberg.help/wavelab_pro/v11/en/wavelab/topics/auto_split/auto_split_dialog_audio_files_r.html

I have some ancient version of audacity installed. It's not nearly as nice of an interface as the one in Cubase, but it worked to trim my test file. It seems the process there would be setting up a macro and then running it against all the files. I didn't take my testing that far,,,

(both trim both start and end)

OhulahanBass

I'll look into it and report back. Im looking at thousands of files that need to maintain their file names, so some form of batch processing is required. It seems Twisted Wave may be the ticket for us non Cubase people.

It all started out as me sampling a few sounds from my Boomstar before I put it on sale and now I have a library of ~ 120 sounds, most of them multi-samples and some with different velocity layers. I guess I got addicted to sampling. Once I figure this out, either with a batch processor or a different multisampler, I will continue with all my hardware synths and a few wild ideas I have.