Load SingleCycle Waves?

[niversen]

I converted it to 16-bit, 44.1 kHz WAV and then ran it through the SamplePooler app to get the RAW. I suspect based on your comments that is the issue. If I am only going to hit another bug, I'll hold off for now. I don't own translator, and would only get it if it was a known good path to full success. The $20 I spent so far is not a big price to pay for experimentation, and I can always use them in some other wavetable synth in the future... Single-cycle waves don't "go obsolete", and they take up a negligible amount of drive space.

I'll just wait for the bug fix firmware and try again, particularly if there is some help with the navigation included.

Thanks for playing with this alongside me.

Nathanael

[koenig]

Yes, I wanted to add folders for the WAV Osc type, as I have also encountered the problems of organizing as you mention. This will be one of the items to be fixed.
We also want to have a multisample .WAV object. It's not clear to me if we need a new Osc Type or if we can stay within the .WAV type. I have to talk to Klaus again to clarify.

This is indeed a most interesting thread! I like the idea of adding wavetables to the Solaris just by putting them on the CF-card - that would be the ultimate open ended synth!

[niversen]

Koenig,

If you have Alchemy or Absynth, I'd encourage you to pick up one of the wavesets mentioned in this thread and play with them. Quite amazing. Because the Solaris has a better architecture than Alchemy, any existing patch that I have could easily be put on the Solaris and improved. If this feature can be part of the upgrade, it will make the Solaris almost infinitely expandable. A decent solution for browsing large folders of waves, and this is a killer feature. It certianly increases the distance to anything else.

[niversen]

To close this out until features can be added, I was forgetting to load the samples from the global screen. Once I did that, I was able to play the wave, but as John mentioned, it was distorted and not correct, so whatever bug that is definitely affects the ability to play single-cycle waves.

While the stability and MIDI affecting bugs would be my highest priority for fixes, for new features, the ability to work with this waveset would be my ask as it adds so much capability.

[SubliminalEffect]

I picked one of the first waves in the set.... Originally called BinAdd Arth 00000001.wav I converted it using Audacity. SamplePooler made me a .txt of the same name. This gave me an error, which I assumed was due to the filename. I changed it to BinAdd.raw on the CF card and also edited the TXT file to match the new name. This gives me a different error. Any ideas?

hello Nathanael, when you write that the Solaris Sample Pooler app gave you an error, what was the error message?

the app was written so that if you select a file that is not in the correct format (i.e., a 16-bit, 44.1kHz .WAV file), then you will get a message that is something like: "The selected sample file is not compatible for use with the Solaris synthesizer. The sample data must be in 16-bit format only."

if you select a folder that doesn't have any sample files in the correct format, then you will get a message like: "No sample files compatible with the Solaris synthesizer found in this folder."

i had purchased the Galbanum Architecture Waveforms when they were initially released but only in the Alchemy and Absynth formats. i went and purchased the 24 bit .WAV version, then converted that same BinAdd Arth 00000001.wav in Logic to 16-bit. selecting the resulting file in the Solaris Sample Pooler app didn't give me any error. it "correctly" extracts the sample file's metadata as the following:
[Pool]
name = BinAdd Arth 00000001

[Sample]
sampleindex = 1
filename = BinAdd Arth 00000001.raw
samplerate = 44100
samplelength = 2048
loopstart = 2047
loopend = 0
rootkey = 60
finetune = 0
lowkey = 0
highkey = 127

i wrote "correctly" in quotes because there is no loop information actually contained in the original sample file as supplied by Galbanum (nor the file converted with Logic). there is no issue with the filename.

since the sample date is supposed to be used as a looping single cycle, i swapped the values for loopstart and loopend. loading this as the SamplePool file and the extracted .raw file in Solaris (with OS version 1.2) didn't generate any errors and plays (and loops and transposes) correctly.

it seems that you are not getting the same successful results. perhaps it has to do with other sample files and erroneous SamplePool files? the values in the SamplePool text files can cause problems like locking up Solaris entirely while the .raw sample files are being loaded into the DSP memory...

[SubliminalEffect]

Yes, the error is because the loop start and the loop end points look swapped. The loop start should always be at zero (or near the front of the sample at least), and the loop end point somewhere down the line (at the end I would guess in your case - which is the length of the sample -1). I thought Bernard had fixed this issue in his SamplePooler app, but apparently not!

the files that Galbanum supplies in their Architecture Waveform product are single cycle samples that are meant to be looped (by you and whatever sampler or app that you use) but the files themselves contain no loop data. you (and your sampler or app) need to loop the data from first sample to the last.

the Solaris Sample Pooler app was written to look for loop data in the sample files that are selected for conversion. if no loop data is found within the sample file (as in this case with the Galbanum Architecture Waveform .WAV files), the app does not automatically assume that the entire sample file should be looped from beginning to end. the app defaults to treating such files as unlooped one-shot samples, i.e., the sample data should be played only once from beginning to end.

the question that was discussed was how to implement this one-shot behavior in the metadata within the SamplePool text files. there were varying options that were suggested (won't bore anyone with the details). after much consideration, in the case of "apparent" one-shot samples, the Solaris Sample Pooler app was written to set the loopstart to the last (valid) sample address and the loopend to the address of the first sample, even though it looks "wrong" or "flipped". this method works successfully to implement one-shot behavior. if you want the sample to have a loop though, you will need to edit the SamplePool text file subsequently to implement the loop yourself.

[SubliminalEffect]

OK, last post on this topic of samples (for tonight at least),

it seems there are SamplePool text files being shared around that have loop points that look a lot more "wrong" to my eyes than having 0 as the loopend.

if your Solaris is locking up during the loading of sample files or playing samples much differently than you expect after loading completes, i would suggest checking the SamplePool text file for a loopend value that is the same number as its samplelength value...

your mileage may vary. let us know what is or isn't working for you.

[niversen]

I did notice the loop point and adjusted that manually, so that was not my issue. I ultimately got the wave to load, and SamplePooler did its magic correctly as far as I know. Once I worked out manually loading the wave, it would play, but not correctly.

I have assumed that this is due to the bug that John mentioned, but it is also true that the samples he includes with the synth work fine, so that may not be it.

Thanks for chiming in - the SamplePooler is a nice bit of work -

[John Bowen]

We have been looking into the differences from the previous .map files (which the Solaris creates) from the v1.0 OS and the v1.1 OS .map files, and there seems to be differences. I’m thinking the routine that creates these .map files (which tell the Solaris where the samples reside in memory) is the object that didn’t get updated when Julian made the revision to the structures. It might be something as simple as an adjustment to these files that will allow us to get around the bug, and allow correct playback of samples.

john.

[niversen]

If that is the case, that would be really cool. If this is addressed, I will definitely use the feature regularly.

[minorguy]

I'm wondering what the state of samples is. I've written a GUI program for Windows that can convert various sample formats to Solaris pool/raw format and manage the pools (*). But samples don't seem quite usable to me on the Solaris right now. I wanted to experiment to try to characterize the specific issues, but it just acts oddly so I haven't. Is it better to just wait for now? I am using OS version 1.2.1.

The raw filename of the first sample doesn't show in the display (although it does in the video tutorial from a long time ago, so it must have gotten broken at some point.) Then, crazy stuff. Samples playback data from other different samples in the list, or loops don't play correctly, etc. I figured, well, the Glock samples play correctly, right? So I can use them as a guide. But they don't. If they were correct, I should be able to play the same note for each sample and hear the same note.

So I'm happy to wait. I just wondered if it's me and whether samples were usable for anyone else.

Jim

(*) I'm happy to give out the program if anyone's interested to try it.

[John Bowen]

Jim,

there’s a new bug in the sample addressing that happened with v1.2, and this has yet to be addressed. If you happen to have a older CF card with the sample data on it, well those seem to load fine - it’s just that when Julian changed the file formatting, the pointers to the sample end point were not updated, so Holger still needs to get in there and adjust for those for any newer CF cards.

[amongstmyselves]

Having fun with samples. After reading through this thread though am I correct in understanding that the bug is in end point as I do seem to get a small squeak of static at the end of a sample I've been playing with.

Steve

[John Bowen]

Having fun with samples. After reading through this thread though am I correct in understanding that the bug is in end point as I do seem to get a small squeak of static at the end of a sample I've been playing with.

Steve

Yes, well, actually both start and end loop points are offset and not being read correctly. Sometimes it’s not noticeable, because the loop is ‘silent’, but other times you will hear strange noise, or even portions of other samples. The addresses for the loop points are just not being read in correctly. There is the possibility of it loading correctly if your CF card is of the older format type, before the data format was changed; in that case, the data is in the correct location, and no loop problems are found.

[amongstmyselves]

Thanks for the clarification John. Still having a great time with the machine. Thanks again !