Low disk performance of SpectraLayers (also 9)

Nevertheless, I think I’m just going to wait until SL10.

@Gades That’s not what I repro here.
Windows 11, Intel Core i7-9750H CPU @ 2.60GHz, 16GB RAM, NVME SSD.
1:00:07 long file, 48Khz, 24bit, wav format

Time to open:
RX9: ~2s
WaveLab Pro 11: ~2s
SL9: ~3s

Time to process the whole file with RX9 Spectral De-Noise VST3:
WaveLab Pro 11: 1:15
SL9: 2:45

Although it’s a bit slower in SL9, keep in mind SL9 is applying this VST3 spectrally, which is not the case in WaveLab. Regarding RX9 it doesn’t support VST3 plugins so no comparison was made (the same process built-in natively and not as VST3 would lead to biased results).

Yes, I’m aware about RX and VST3, that’s why I measured the time also in Acon Acoustica. Besides, it does not explain such large differences in the times measured by me - 26s instead of 4s? 21min instead of 2min? There is something terribly wrong!
For instance, why SL9 do some “conversion” while opening single file? Even if the project settings are the same like in the wav file? This is the main time of the opening - the “conversion”.
21 minutes of de-noising is the fact. I repeated this three times, every time it was 20-22 minutes in SL9…

VST3 support is nice, but I’m not willing to sacrifice overall performance to it. I definitely see improvements in SL9 over SL8, but the (admittedly very limited) user feedback about SL9 that I’ve seen here and at Gearspace has persuaded me to wait another generation before upgrading.

I’m happy with the upgrade, but the slow disc operations were my complaint also in SL8 and I was hoping that the announced performance increase would improve also this aspect.

Converting from a format such as mp3/wav/… to raw, unformatted 32bit float data.

Can you provide a couple log files (Help > Support Info > Logs Folder) ?

Sure, here you are: Log file - loading and de-noising

That is what most probably is happening here.
Spectralayers converts everything it is working on to a hi definition format, so every operation it does is better.
At the audio world, the level of precision one is able to achieve with 24 bits fixed files is outstanding already. 32 bits is several levels above and then to have that ample field of intervention extended to ‘floating point’ bits is another sky entirely…

Placing issues and tasks in perspective, what one is able to do nowadays 2022 with Spectralayers is a decades long held dream already come truth. If in order to do that, it takes 15 seconds longer to do simple file operations than wanna be programs, so be it. In fact, let it be. Please.

Yes, as found me saying at another thread: Please, Robin do not divert your time and efforts in mentioned time efficiencies if your development Gantt charts are already populated with new features and optimizations focused on the synergetic spectral and de/layering tools Spectralayers has been increasingly providing us. For one this is my vote.

About the conversion: I tried also 32-bit float point 48kHz WAV imported into 48kHz project. There was also conversion - it should be?

@Nspace - why are you whining in this thread? :wink: I’m curious why loading/processing is so slow, because I often process large files and I spend huge amount of time on waiting. I have a lot of 32-bit float point files (because I record them) and I know about advantages of this format. Why Acoustica, RX, WaveLab can process these files fast, but SpectraLayers cannot? Maybe for you 15 seconds on loading, 20 minutes on processing is nothing, but I need to deliver processed files to my client and I see that I’m able to do it faster in other applications. However I don’t want to, because SL9 has some unique features I need…

Is it wrong to ask about something like that on the forum? I thought that was what it was for. It was similar performance problem with Match EQ function in SL9 - maybe the same way can Stainberg improve loading or processing by VST3? For one this is my vote :wink:

@Gades Can you open a terminal and type the following command line:
"C:\Program Files\Steinberg\SpectraLayers 9\Plugins\Files\ffmpeg.exe" -i "F:\Gades\Tata\01\tata_01_konrad.wav" -y -vn -f f32le "C:\Users\Rutterkin\AppData\Roaming\Steinberg\SpectraLayers 9.0\Cache\layer(1b709119).pcm"
and let me know how long it takes to fully execute it ? You can then delete C:\Users\Rutterkin\AppData\Roaming\Steinberg\SpectraLayers 9.0\Cache\layer(1b709119).pcm

NB: even if your wav file is 32bit float, SL still needs to convert it from formatted data (wav) to unformatted (pcm).

1 Like

Sure. Here is the result of the command line execution: CMD Execution

It took 50 seconds.

@Gades. I welcome your numbers & measurements centered posts. Plus when you ask for confirmation. More so even when you ask for reasons and express your curiosity.

Robin, above just addressed most of these points, further providing test results… and importantly, pointing to the format Spectralayers audio engine works at as well as some specific differences at play…
That is all quite interesting and meritorious in itself: your questions, test and results together with Robin’s answer on the same tenor with further specification on the actual comparison conditions, detailing each software.
Oh yes! I welcome all that.
Reading the attached log, it seems you are applying de-noise RX process thru Spectralayers recently launched VST3 chain… Odd to compare that with the actual performance off said other softwares. If this thread title is your aim, one would expect you compare each software in their own devices.

But again it might be your angle, in my case I don’t just yet expect to use Spectralayers as a my do-it-all center of operations. SL is my hi-tech unique specialized tool. One that is the depository of my advanced tricks and findings. These are either my deep sea/deep space eyes or my most powerful microscope and tweezers into where nothing else reaches.

1 Like

However, that just said is not all.
Just beside your numbers and partial description of your rapid tests, also when rather contrasting results and explanations are given to you, come your adjectives.
In your own words:
"maybe only for me these times are sooo slow… :wink: “, “There is something terribly wrong!”, plus the one post above where only emotional qualifiers articulate its content: "I’m happy with the upgrade, but (…) were my complaint also in SL8 and I was hoping that (…) would improve also this aspect.”
Then you delve further that rabbit hole: “@Nspace - why are you whining in this thread? :wink:” …

So, who is having the emotional lapse here? I regularly have no issues with reading and even exchanging emotional barter, it just seems odd to be called such when those words come by precisely who’s doing such…

I’m emotional because of I hoped that SL9 will be faster than SL8. There is no trial version, so I had to buy an upgrade.

BTW, Match EQ had a performance problem which was resolved immedietely. So I did measurements to be sure I’m right - SL9 IS slower than other software. I chose Spectral De-noise VST3, because I use this de-noiser in every application and I can compare processing times.

I got the point about necessary 32-bit raw conversion and I understand this. But I do not agree that it is something of little importance - yes, that may be subjective, but I don’t think the performance is important only for me.

I don’t want to argue, because that’s not what I created the thread for. So, if you want to vote or complain, please start a separate thread.

2 Likes

Oh brother. Lol.

Oh Yes! That is rather allegoric way to qualify Robin’s work with Spectralayers…

Until you dive deeper and find about his research with image/video reconstruction algorithms.
Not entirely unlike this week’s noticeable advancement at a parallel field of research, literally being applied at Space or subcellular technologies :

https://www.laserfocuswold.com/science-research/article/14277750/new-algorithm-speeds-image-reconstruction

There is some chance such developments come later to affect what we do with these audio tools.

"the team developed a rapid joint space and frequency reconstruction (JSFR) SIM method, which provides improved image reconstruction speed and suppression of the out-of-focus background in thick cells.

How does JSFR-SIM work?
Most SIM reconstruction algorithms are based on the Wiener filter protocol, which essentially filters out noise from a corrupted signal to provide an estimate of the underlying signal of interest and optimizes contrast in the available signal-to-noise ratio. The complex workflow of Wiener-SIM makes super-resolution reconstruction extremely time-consuming, according to the researchers, which limits the utility of SR-SIM as a live-cell imaging modality.

JSFR-SIM is a concise reconstruction protocol to rapidly obtain optically sectioned, super-resolution images of thick specimens by combining spatial domain processing with the conventional workflow implemented in the frequency domain.
Most Fourier-domain operations are replaced with simple multiplication and summation calculations in real space.

“The main advantage of JSFR-SIM is its unrivaled reconstruction speed,” Lei says. The team found that the execution time of the reconstruction is reduced to millisecond level with the new algorithm, “which is measured to be 80x faster than the widely used Wiener-SIM. Critically, the speed increase does not come at the expense of image quality.”

///Spectralayers foundation comes from image research, has demonstrated is able to incorporate machine learning from others and own initiatives (Torch), and has navigated the business modelling + firm associations stormy waters while retaining its maneuvering ability.
Robin Lobel is at a very good position to keep delivering the best innovation with this software.
*And according to this specific news your speed optimization quest might be eased as well.

Dude… We know you are wise, brilliant, smart, eloquent poet, but really, this is a simple thread about SL disk performance. You litter it with your long unrelated comments and you make Robin’s job difficult. So again, please, create your own thread about Robin or beauty of SL and write there.

I know Robin and I appreciate his work very much, all we do.

I think Gades is providing completely reasonable feedback. It’s not like he’s insulting Robin’s work.

Let’s focus on trying to figure out whether this is something to fix or if there’s something going on with his system that could be causing this.

1 Like

Thanks. ffmpeg shouldn’t take that long to convert it. So there’s either something unusual with the file, or your system, or that version of ffmpeg running on your system.

1/ Would you be able to upload your wav file and send me a link to it in PM (or by email to contact [at] divideframe (dot) com) so I could check locally ?

2/ Can you download the latest version of ffmpeg (from here) and run the same command as in my previous post, but replacing the path to ffmpeg to the path where you downloaded the latest ffmpeg, and let me know how long it took (+log) ?

1 Like

Wow, Robin, I think we found the source of the problem - here is log, but time is about 4-5s instead of 50s! :smiley:
Do I need new ffmpeg.dll to improve SL9? I replaced “c:\Program Files\Steinberg\SpectraLayers 9\Plugins\Files\ffmpeg.exe” with the new one, but SL9 is still slow. From where can I dowload dll version? :slight_smile: