Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
What's the best way to encode samples?
#1
I've never really managed to get decent quality sample playback from my NerdSeq. The samples that are supplied sound great, as do the Sonic Voltage ones.
I've been chatting to some of my other Nerdseq mates, and they haven't managed to converst samples that sound good, either.
What settings should I be using in Audacity on the PC?
Reply
#2
Might seem counter intuitive but try converting your samples to 8 bit mono unsigned raw. The lofi'ness gives it some extra punch.
Reply
#3
Depending on what you want to archive play around with the dither setting in audacity.
https://manual.audacityteam.org/man/dither.html
Some explaination here https://forum.audacityteam.org/viewtopic.php?t=111209

If you don't care about the lo-fi ness and don't want to use that as an effect on purpose you can also use a brick wall low pass on the sound so dithering has got more space to work in.
Reply
#4
Some weeks back I wrote some code, using SoX, to help process some purchased samples into a few different forms for loading into NerdSEQ. 8 bit vs 16 bit, and each in undithered and dithered versions, so each sample was turned into four samples. Most things didn’t sound much better in 16 bit on the NerdSEQ, though it does help some. Most things sounded better without dithering rather than with, but it was very heavily dependent on the sample. Long story short: consider generating a few variations of each sample so you can try them on the NerdSEQ.

Also, and though this might just be my NerdSEQ I expect it would also be true for others: expect some extra noise during preview (likely from sd card access) that won’t be present during normal sample playback.
Reply
#5
I actually converted the on sd card samples using winamp (wave writer) into 8 bit raw.

There will always be some noise since we are dealing with 8bit quality and some will sound better some less.
PLEASE use the search function if something have been asked or discussed before.
Every (unnessesary) forum support means less time to develop! But of course, i am here to help!  Smile
Reply
#6
I spent a couple of evenings researching and trying to get the best sound out of samples I created with my setup. In my experience, 8bit raw undithered and unsigned was best for most uses.

I found that snares, cymbals, hats, basically anything noisy sounds great.

Kicks are problematic though. Does anyone have a trick for getting clean results from decaying sounds?

I found that samples are noisiest when there is any silence or the level drops below -12db or so.

I haven't had a chance to test out the stock samples because I purchased my unit used and without an SD card. Is there a good sounding kick there?

And a question for the makers here - are the samples converted to 8bit raw when loaded, or are 16bit samples still 16bit?
Reply
#7
Internally converted to 8 bit.
The noise is a result of that since the LSB is already audible. You should avoid long low volume releases.
PLEASE use the search function if something have been asked or discussed before.
Every (unnessesary) forum support means less time to develop! But of course, i am here to help!  Smile
Reply
#8
(08-30-2020, 05:49 AM)mvdirty Wrote: Some weeks back I wrote some code, using SoX, to help process some.... 
Hi, could you share your Sox code please? 
I'm just diving into this 8bit raw conversion and I'm having issues with zipper distortion printed over the samples..
Reply
#9
(11-01-2021, 11:10 PM)radar23 Wrote:
(08-30-2020, 05:49 AM)mvdirty Wrote: Some weeks back I wrote some code, using SoX, to help process some.... 
Hi, could you share your Sox code please? 
I'm just diving into this 8bit raw conversion and I'm having issues with zipper distortion printed over the samples..

You probably just go the best with the original wave file. Most
conversions do the same as the NerdSEQ does internally.
PLEASE use the search function if something have been asked or discussed before.
Every (unnessesary) forum support means less time to develop! But of course, i am here to help!  Smile
Reply
#10
(11-01-2021, 11:23 PM)XORadmin Wrote: You probably just go the best with the original wave file. Most
conversions do the same as the NerdSEQ does internally.

Very true. The NerdSEQ actually does a pretty nice conversion on most loaded samples.

In my case, I scripted a solution because I wanted to try different dithering options and sample rates for various samples, was renaming and reorganizing the files via script, and had hundreds of files to process (resulting in thousands.)

Unless in need of dithering (and/or doing a bunch of re-org) then simply loading the samples is usually fine.

(11-01-2021, 11:10 PM)radar23 Wrote:
(08-30-2020, 05:49 AM)mvdirty Wrote: Some weeks back I wrote some code, using SoX, to help process some.... 
Hi, could you share your Sox code please? 
I'm just diving into this 8bit raw conversion and I'm having issues with zipper distortion printed over the samples..

I don’t have it in front of me at the moment but it was mostly file organization scripting with a bit of sox invocation along the way. The sox operations themselves were bog standard, setting output bit depth and sample rate, and I had the script generate variations with and without dithering. The sox docs should have the command line examples you need, but let me know if you can’t find them and I might be able to dig up the code.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)