Hi! I’m new on Analog Rytm and after few hour of search and testing i found a perfect (for me) way to batch convert audio files for the AR.
I use the free version of Wavepad (i’m on mac)
Wavepad permit the batch conversion of files, samples for the A$ (if i understand) must be: 48KHz, 16bit, Mono .wav
On wavepad, select batch, then add the files. in the step2 you have to create the batch script file, simply (in my case) add: Convert sample rate to 48000, Convert to mono and normalize.
FYI, you might not want to normalize your samples, because if your samps are normalized, using sample volumes of above 100 on the AR might introduce clipping…
Since day 1 owning the AR I haven’t converted my samples to 48000 mono, just allowing the process to be done for me during the transfer using C6.
Does anyone know if there is any downside to doing this ?
Samples I create myself or edit I have been converting to mono, 48000 and normalizing. I did notice that they are loud as f—, but I find everything is loud on the Rytm. I no complaints here, however. I tend to not crank the volume on anything no matter what machine I use for risk of drowning all the frequencies because I usually have lots playing at once.
I’ve actually grown really fond of Audacity. The first time I set hands on it was back in the day in Ubuntu linux. I remember thinking “this program looks like dog poop” and I dislike using things that have a crap UI. Then I recently installed it at work for preparing samples and instruments for the OP-1 and it’s really grown on me. It does literally everything you need. I can sample chop using a BPM plugin and sample markers to snap to. I can use free VST’s for tuning, EQ, effects whatever. Plus the handy Z button for auto zero crossing your selections is so tight. Anyways … if you can get past the ugly UI - Audacity is the bomb.
Normalizing doesn’t mean ‘maximum level’. Normalizing means bringing the maximum levels of different samples to the same level (which may be -12dBFS). It’s about making your samples comply to a norm.
Granted, often a ‘normalize’ function is actually ‘normalize to 0dBFS’ function, but usually there is a way to change the level that the samples are normalized to.
Normalizing is pretty handy, because afterwards you know what levels to expect.
If it hurts, don’t poke it. In other words, if you get unwanted distortion when using a volume of above 100, then don’t set the volume above 100!
I do not understand what you mean. A classic normalize function, as on most samplers, old and new, make the loudest peak of an audio file to be 0dBFS. And the AR gainstage is at unity at 100. Therefore, performing a callsic normalize and going above 100 will exceed 0dBFS within the ar signalpath.
It is wuite simple really. A more recent option in certain software allows one to specify the scale of normalization, so that the loudest peak of an audio file can be something else than 0dBFS, but this is a more recent ”niche” feature and should be kept out of the discussion when talking about the kind of normalization that most samplers do (ie. no parameters to config, you do it its done)
That’s only because they fixed the value that the samples are normalized to to 0dBFS.
Most serious sample editors allow you to set this value. You can normalize to for instance -6dBFS.
But i do agree that generally people think that normalization must be to 0dBFS. And it is indeed these ‘crippled’ normalize functions that made us think that.
Which particular bit of gain are you talking about? The only one that can ever go over 0dBFS is the one in the sample menu.
To be honest i haven’t tested this yet but it seems slightly strange if that one would cause DAC clipping.
Hmm, Sound Forge has had this feature for about 20 years now?
And it’s not a niche. I would even say that the 0dBFS thing is the actual niche.
I think it originates (especially in samplers) from the fact that 16 bits and lower is not that much dynamic range in the context of sound creation. Things like amp envelopes working on low level samples will have artifacts.
So there was a notion in the 80’s and 90’s that samples should be as loud as possible and fill as much bits as possible. The normalize function was used to achieve this and hence the only setting you got was to normalize to 0dbFS.
But nowadays, with 24 bit and whatnot, this isn’t a problem anymore.
Normalizing to 0dBFS is thus a historical artifact and pretty much a niche these days. There is no technical reason anymore to normalize to 0dBFS.
No, its just that this 0dBFS is old news and really not relevant anymore.
Normalize to what you need instead of to 0dBFS
Hitting 0dBFS makes convertes act badly anyway, so there is no reason to do it without good reasons.
Also, i have tested this level story, but it doesn’t check out exactly.
There is indeed distortion, but it is analog! It doesn’t come from the converter.
It’s not so much clipping but more like a rounding at the extremes.
But you’re right, something does happen if your samples are normalized to 0dBFS so this is worth knowing about for sure!
You can only remove this distortion by lowering the level on the sample page, which suggests to me this is happening either in the mixer stage (between the analog voice and the sampler voice) or on the filter input.
Reducing levels on the amp page doesn’t reduce this distortion. In fact, at maximum amp level you get even more distortion. And even more if you max out track level!
So the lesson learned is maybe that if you want to avoid distortion you need to take care of your headroom and not slam everything to max.
Question:
i just got a download from samples from mars and i get to choose:
16 bit / 44.1 khz
or
24 bit / 44.1 khz
As far as i know the rytm transfer app converts them to 16 bit / 48khz?
so which one to download?
what happens if a 44.1khz sample is converted to 48khz?
