but remember it won't be a full fledged electribe sx internal sample memory is limited to ~500kB and you have to flash the samples from the SD card there, which is not possible during playback. So it's a nice addition for some sample fun but won't satisfy all your sample needs.
but since it can be used as a waveform in any oscillator you are still able to mix the sample with other oscillators and mangle them with fm and filtering
- you need a folder on the sd card named "samples" - put some 16-bit mono wavefiles in there (not more than 490kB!) - on the load screen there is a new option "load samples" - confirm with 'ok', make sure the seq is not running or the upload will fail - sample upload starts... this may take up to 30 sec. (not optimised yet) - when you see the "load sample - ok" screen again the upload is finished - the oscillators have some extra waveforms now "S1 to "Sn" where n is the number of samples uploaded that's it.
Successfully loaded 13 samples of various lengths (12k to 47k in size)
some samples are particularly 'clicky' perhaps due to zero crossings? Changing their Coa parameter can avoid this at various settings but not all. It isn't a gradual thing, single steps in Coa introduce a click, and the next step it might disappear for several values, and then return again. Not a major issue, as I can just find Coa levels where the sample sounds good ... but maybe there is a fix?
Do you have recommendations on what the 'ideal' sample characteristics should be when preparing them?
eg for each .wav,: better there is a short silent gap before the sample starts & after the sample ends ? Or maybe no gap?
A few of my samples didn't work, I'm sure it's me...others I had to make sure the decay and slope were up so that they could be heard...probably the "clicky" mokomo is talking about. If nothing else, 606 hi hats and a "real" 808 cowbell are worth this!
the thing is it is a very naiv sampler implementation... no resampling or anything fancy. just plain interpolation but i have to adjust the sample engine a little. at the moment there is a rather short maximum length as well..
samples that did not work - they didn't show up in the menu or were loaded and just produced noise?
here's a sox batch converter script for a windows bat file to drop .wavs onto (to ensure they are 16bit mono 44.1k with silence trimmed off the end) ... you'll need the sox.exe folder in your path:
cd %~dp0
mkdir LXRconverted
FOR %%A IN (%*) DO sox %%A -b 16 "LXRconverted/%%~nxA" rate 44100 remix - reverse silence 1 1 0.1% reverse
the kHz should not matter. It just means the pitch range of the sample will be off compared to 44.1kHz samples. You are free to use 22.05kHz samples to save some space and pitch them twice as fast
When you mentioned that you'd try this, I was expecting to see an implementation in a few months, definitely not this soon.
It happily ate all the samples I threw at it from the first try. The longer samples (crash and ride) are indeed cut short.
At first it seemed like some waveforms weren't working, but I figured out that it loads anything from the sample folder, regardless of the file format. So the empty waveform slots I have now are merely the hidden id files that my mac created when I copied the samples to the sd card. Perhaps a file extension check is in order...
It would also be helpful if the first 3 characters of the filename are used to name the waveforms, rather than s1... sN
Excellent work Julian, this opens up many new possibilities!
Rvk
EDIT: Apparently we were both loading a 707 kit at the same time, what are the odds!
Do you have recommendations on what the 'ideal' sample characteristics should be when preparing them?
eg for each .wav,: better there is a short silent gap before the sample starts & after the sample ends ? Or maybe no gap?
definitely no gap. there is no processing involved from the side of the LXR. The sample data is just transfered to the microcontrollers flash memory 1:1 so any gap before a sound will result in latency. silence after the sample just wastes memory.
another thing to keep in mind. you are only aber to flash new samples about 10.000 times before the flash is starting to wear out. With one sample upload per day thats 27.39 years.
@Revok most of the playback infrastructure was already there. I just needed a .wav file parser and modify the flash upload routine from the bootloader a bit + after all the pre-order bureaucracy I was motivated to do some coding again sometimes you just have a feeling that it's a good time to code and things go 10x as fast
the kHz should not matter. It just means the pitch range of the sample will be off compared to 44.1kHz samples. You are free to use 22.05kHz samples to save some space and pitch them twice as fast
Trimmed the same set of samples down to remove more of the tail (used the FS1R folder from the links posted earlier in this thread) and put them to 22.05k instead of 44/1k. For some reason there is now far less clickiness and they sound very well on the LXR.
It would be useful that there is a different naming convention than S1 ...Sn that reads from the filenames.
Maybe 'Uab' where U is static for User and then the ab is the first two characters of the file name so we can use set the filename's first character for say, K for Kick and the second character as a number 0-9.
Also, it would be handy that you can have user kits in subfolders within the 'samples' folder, and select from these folder names using the 'Load Samples' menu.
Comments
internal sample memory is limited to ~500kB
and you have to flash the samples from the SD card there, which is not possible during playback.
So it's a nice addition for some sample fun but won't satisfy all your sample needs.
I would not rather replace a sampler than spice up hats and snares...
got to dig out my breakbeat collection
and don't forget about these
http://warbeats.com/Community/Forums/aft/57
https://www.dropbox.com/s/m7fq66m0e69tjit/FIRMWARE.BIN
- put some 16-bit mono wavefiles in there (not more than 490kB!)
- on the load screen there is a new option "load samples"
- confirm with 'ok', make sure the seq is not running or the upload will fail
- sample upload starts... this may take up to 30 sec. (not optimised yet)
- when you see the "load sample - ok" screen again the upload is finished
- the oscillators have some extra waveforms now "S1 to "Sn" where n is the number of samples uploaded
that's it.
:-S
but i have to adjust the sample engine a little. at the moment there is a rather short maximum length as well..
samples that did not work - they didn't show up in the menu or were loaded and just produced noise?
It just means the pitch range of the sample will be off compared to 44.1kHz samples.
You are free to use 22.05kHz samples to save some space and pitch them twice as fast
there is no processing involved from the side of the LXR.
The sample data is just transfered to the microcontrollers flash memory 1:1
so any gap before a sound will result in latency. silence after the sample just wastes memory.
another thing to keep in mind. you are only aber to flash new samples about 10.000 times before the flash is starting to wear out.
Trimmed silence and converted 24bit to 16 in Renoise from Goldbaby & Driven Machine Drums packs.
All worked first time and sound great - no clicks, pops or fizz
Well maybe just a little fizz on the hats
Bravo!