SPECWEB (Now 2.2)

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I really like what it does but I’m still having the issue with a crackling/distortion in the LCR. If i could find a way to make that disappear this would be 💯 winner
 
Hey Ron,

Glad you are having some good results, at least with drag and drop.

The ability to upmix and play in real time is going to depend on the processing power/speed of your computer. There are some things to try to help mitigate that, at the possible expense of quality of the upmix. However, you could play live. with settings that allow you to hear without problems, then note down the settings and then upmix the song using drag and drop for maximum quality.

You are supposed to be able to save your settings as an ini file for later use, but I recently discovered that that feature isn't working in 2.2. It will be fixed in the next version, which I hope to get out later this month or in January at the latest.

In order to decrease the load on your computer for live upmix/playback, here are some flags to try (one at a time, or in combination):

-u1 will turn off the default 2x oversample, which in turn also affects some of the below FFT default settings.​
-N8192 will set the FFT size to 8192 bins. You could also try 4096,​
-W2 (or or 1) sets the FFT overlap factor. Default is 3. For a given oversample, or FFT size, this would probably have the biggest impact on CPU load, but also the biggest impact of results quality.​

Of course all of those settings have their equivalents in the ini file, should you prefer.

FYI SpecWeb uses threads and parallel tasks so a CPU with more cores will perform better than one with fewer.

The Sonic Anomaly "Unlimited" VST is used (by default) to remaster the upmix before final level adjustment in "Zag". It happens in the background (meaning the GUI doesn't pop open, unless you ask it to) during drag and drop, as the 2nd step. If you feel it is getting in the way of live play you can turn it off with:

-v0 (zero)​
But again in that case you might want to note down your settings and do the actual upmix with drag and drop, with the vst active.

Looks like you did stump me on the Output directory. There is no such option. I can look at adding that to feature list for upcoming versions.

Let me know how it goes for playback, etc.
 
A slight clarification; you CAN specify an output directory, as part of -Foutput-filename.

E.g.:

D:\Glenn\Music>"D:\Google Drive\SpecWeb 2.2vs19\Release\Install\bin\SpecWeb.exe" -F"D:\Glenn\Music\5.1\Love Alive.flac" "D:\Glenn\Music\Love Alive.flac"

However it must be done on the actual command line, vs. just adding -Foutput-filename as an option during drag and drop.
 
I really like what it does but I’m still having the issue with a crackling/distortion in the LCR. If i could find a way to make that disappear this would be 💯 winner

Thanks for bringing the click/pop (you said crackling distortion) to my attention. Please try:

-u1 (which will turn off the default 2x oversample).​

and let me know if it goes away (it will if we are talking about the same thing - sounds like a click/pop from playing vinyl).

I recently stumbled on this after fixing another artifact and already know what the permanent fix is. It will be in the next version.
 
Thanks for bringing the click/pop (you said crackling distortion) to my attention. Please try:

-u1 (which will turn off the default 2x oversample).​

and let me know if it goes away (it will if we are talking about the same thing - sounds like a click/pop from playing vinyl).

I recently stumbled on this after fixing another artifact and already know what the permanent fix is. It will be in the next version.
Thanks, I will try it out later today
 
Thanks Zeerround

I've realised files less than 96kHz play fine. I'll try your flags with some higher res files. My laptop is an i7 with only 2 cores, so not so good.

The surround system I'm playing this through is 6.1. I am aware that SpecWeb only produces 5.1 files, but the AVR is setup so when playing 5.1 it puts sound common to both SR and SL through the rear speaker. Could you suggest some settings I could test to get best results in this circumstance?

Writing output to a distinct subdirectory would help me to select all the files created and move them off to my server. But it's no big deal.
 
Thanks for bringing the click/pop (you said crackling distortion) to my attention. Please try:

-u1 (which will turn off the default 2x oversample).​

and let me know if it goes away (it will if we are talking about the same thing - sounds like a click/pop from playing vinyl).

I recently stumbled on this after fixing another artifact and already know what the permanent fix is. It will be in the next version.
Yes!! It worked
 
Hey Ron,

Glad you are having some good results, at least with drag and drop.

The ability to upmix and play in real time is going to depend on the processing power/speed of your computer. There are some things to try to help mitigate that, at the possible expense of quality of the upmix. However, you could play live. with settings that allow you to hear without problems, then note down the settings and then upmix the song using drag and drop for maximum quality.

You are supposed to be able to save your settings as an ini file for later use, but I recently discovered that that feature isn't working in 2.2. It will be fixed in the next version, which I hope to get out later this month or in January at the latest.

In order to decrease the load on your computer for live upmix/playback, here are some flags to try (one at a time, or in combination):

-u1 will turn off the default 2x oversample, which in turn also affects some of the below FFT default settings.​
-N8192 will set the FFT size to 8192 bins. You could also try 4096,​
-W2 (or or 1) sets the FFT overlap factor. Default is 3. For a given oversample, or FFT size, this would probably have the biggest impact on CPU load, but also the biggest impact of results quality.​

Of course all of those settings have their equivalents in the ini file, should you prefer.

FYI SpecWeb uses threads and parallel tasks so a CPU with more cores will perform better than one with fewer.

The Sonic Anomaly "Unlimited" VST is used (by default) to remaster the upmix before final level adjustment in "Zag". It happens in the background (meaning the GUI doesn't pop open, unless you ask it to) during drag and drop, as the 2nd step. If you feel it is getting in the way of live play you can turn it off with:

-v0 (zero)​
But again in that case you might want to note down your settings and do the actual upmix with drag and drop, with the vst active.

Looks like you did stump me on the Output directory. There is no such option. I can look at adding that to feature list for upcoming versions.

Let me know how it goes for playback, etc.
While you're working on an upgrade 😁 ... I mentioned this in the forum, not sure if you saw it. When using Foobar to do the upmixing, you cannot have an "&" in the file name. It throws an error.
 
While you're working on an upgrade 😁 ... I mentioned this in the forum, not sure if you saw it. When using Foobar to do the upmixing, you cannot have an "&" in the file name. It throws an error.

Another user brought up foobar over there at the same time, and when I tested (foobar2000 1.6.2) it seemed totally broken and I just said so and I would have to look at it later.

What version of foobar2000 do you have?

For me it was trying to invoke SpecWeb separately for each SpecWeb option, as if the option were a file name.
 
That's weird. No I ran a batch of probably 40 files (ripped 4 or 5 albums) and it only crashed with this one. I dropped the one file into SpecWeb and it worked fine. Interestingly, I renamed it and it worked in Foobar but the file sizes ended up being different. Any idea why? I use default settings in both. My Foobar version is: 1.5.3.
 
Thanks Zeerround

I've realized files less than 96kHz play fine. I'll try your flags with some higher res files. My laptop is an i7 with only 2 cores, so not so good.

The surround system I'm playing this through is 6.1. I am aware that SpecWeb only produces 5.1 files, but the AVR is setup so when playing 5.1 it puts sound common to both SR and SL through the rear speaker. Could you suggest some settings I could test to get best results in this circumstance?

Writing output to a distinct subdirectory would help me to select all the files created and move them off to my server. But it's no big deal.

Hmm, I wouldn't think there would be any common sound in LS and RS? But perhaps you can try pushing some sound in there by changing the the "Total Image Width", in arctan, to 360 (290 is the default), or for "slice", turning on "Wrap Wears".

-i360 (arctan total image width)​
-w1 (slice, wrap rears, if using Slice -M0 or ArcTan with Slice Blended Rears -M3)​
I haven't plaid with it in a long time but you could also experiment with Ambient Extraction:

-e1 with -g0 (zero, default) or with -g1. The ini file description of Ambient Extraction has more info.​
 
That's weird. No I ran a batch of probably 40 files (ripped 4 or 5 albums) and it only crashed with this one. I dropped the one file into SpecWeb and it worked fine. Interestingly, I renamed it and it worked in Foobar but the file sizes ended up being different. Any idea why? I use default settings in both. My Foobar version is: 1.5.3.

Yeah I suspected they have made a change (intentional?) in foobar2000 that broke my integration. I'll have to try to figure it out. In the meantime, don't upgrade your foobar version. I seem to remember some DTS or other plugins breaking with upgrades as well (but have since been fixed again).

Re: file size. I haven't looked closely but thought I noticed it being "non-deterministic" as I've been running stuff through alpha code all day every day for a while, but was too deep in the artifact chasing (really SUPER results, I think, there) and haven't done the acid test. Another thing to put on the list.
 
So Zeerround, I tried the -u1 switch and that made all the difference playing hi-res files through the SpecWeb Controls interface. The other suggestions you made had no apparent effect.

However with the same -u1 switch set, I could not play flac files ripped from CD (44100 Hz). I kept getting an Error(14): Can't pause audio. Does that make sense to you?

After converting said CD files, is anything gained keeping them at 24-bit considering they were originally 16-bit?

I've been altering the width controls without really knowing what I'm doing. With image width, the default setting has everything playing through my rear speakers, as I move it left I suddenly hit a point were everything moves to the front. I then tweak the other two controls attempting to get clear central vocals and discrete instruments in the fronts. Is this a reasonable approach?
 
Thanks for bringing the click/pop (you said crackling distortion) to my attention. Please try:

-u1 (which will turn off the default 2x oversample).​

and let me know if it goes away (it will if we are talking about the same thing - sounds like a click/pop from playing vinyl).

I recently stumbled on this after fixing another artifact and already know what the permanent fix is. It will be in the next version.
When will the next version be released?
 
So Zeerround, I tried the -u1 switch and that made all the difference playing hi-res files through the SpecWeb Controls interface. The other suggestions you made had no apparent effect.

However with the same -u1 switch set, I could not play flac files ripped from CD (44100 Hz). I kept getting an Error(14): Can't pause audio. Does that make sense to you?

After converting said CD files, is anything gained keeping them at 24-bit considering they were originally 16-bit?

I've been altering the width controls without really knowing what I'm doing. With image width, the default setting has everything playing through my rear speakers, as I move it left I suddenly hit a point were everything moves to the front. I then tweak the other two controls attempting to get clear central vocals and discrete instruments in the fronts. Is this a reasonable approach?

-u1 for play - Hurray!

Can't Pause Audio. I'll have to check. Hmm, looks like play with with -u1 is not working here either. Another thing for the bug list!

Re 24 vs. 16bit, I would recommend keeping them at 24bit. I take pains to ensure processing at the highest quality and that includes 32bit internal processing and intermediate files, and 24 bit final files. Also 24bits is what is required by DTS and Dolby encoders, should you want to do that.

Lastly, going from 24 bits to 16 bits probably (huge discussions/flame wars on forums) requires dithering (intentionally adding some very low level noise) and I prefer to just avoid that whole topic by staying at 24 bits.

If you are trying to save space, as a side note, I recently became aware of wavpack:

http://www.wavpack.com/
which foobar2000 supports (conversion and playback) so you could see if that makes things any smaller than flac. Also, is space is really an issue, I could look at asking for more compression from flac, at the expense of processing speed.

Re: width controls, Yes that is reasonable. I would add "soloing" the center and front channels to make sure things are placed cleanly within a given channel (or sets of channels), as part of "to get clear central vocals and discrete instruments in the fronts".

This: "With image width, the default setting has everything playing through my rear speaker" is odd, but I have no experience with 6.1 or your particular setup.

If you're happy with the results maybe we shouldn't mess with things but another "knob" to look at would be the angles of your speakers, from the listening position, vs. arctan expecting "ITU speaker placement" for 5.1.

You can tell arctan what your actual speaker angles are, see the ini file or:

-7v,w,x,y,z Speaker angles in degrees CCW from zero (straight ahead). LF,RF,C,LS,RS.
Default is ITU 5.1 angles 30,330,0,110,250.

You put them all in at once, like:

-745,315,0,120,240
 
When will the next version be released?

I'm shooting for this month or next, however I can do some beta's in between. That would the artifacts fixed but probably not the other longs list of minor?;0) bugs. I'll be asking beta users for feedback on a default values for new options.

I have a lot of code merging to do from different branches but can work on a beta ASAP.
 
Yeah I suspected they have made a change (intentional?) in foobar2000 that broke my integration. I'll have to try to figure it out. In the meantime, don't upgrade your foobar version. I seem to remember some DTS or other plugins breaking with upgrades as well (but have since been fixed again).

Re: file size. I haven't looked closely but thought I noticed it being "non-deterministic" as I've been running stuff through alpha code all day every day for a while, but was too deep in the artifact chasing (really SUPER results, I think, there) and haven't done the acid test. Another thing to put on the list.

I know those of us on this thread appreciate all the hard work you are doing here. And I especially appreciate you being so available for everyone's questions. Thanks again!
 
My setup is diagonal! Imagine a square room, I sit in one corner and the centre and front speakers are in the corner diagonally opposite. The front to surround speakers form a rectangle 5ft wide and 10ft long. Not ideal but apart from moving house there's not much I can do about it. I imagine SpecWeb producing a circle of sound around me but in reality it's more like an oval with very little width. I've set speakerangles=10,330,350,100,260; so I also sit a little off centre.

For a while now my surround music has been supplied by the FreeSurround component of Foobar2000. That upmixes all of my tracks on-the-fly and does a very good job of widening the centre speakers. However it has serious issues and is no longer being supported. So I'm looking to replace it.

SpecWeb does sound better, just an overall boast to the fidelity. Also the instruments can be located more precisely in space. But as I say I have that really big gap to fill between the front and rear speakers.

With the prospect of 1500 albums to convert I want to get the default parameters right.
 
  • Wow
Reactions: GOS
I've been trying out SpecWeb tonight for the first time. I have almost ZERO technical knowhow, and liked the sound of drag and drop to create pseudo-surround files. After downloading the software and running a FLAC through it as a test it seemed to work fine. It created an -mch FLAC version of my track on my PC.

My problems started when I tried to stream it to my hifi. My Denon AVR picked up the file via the Media Server option but on playback the sound was thin and low on volume. A quick check showed that only speakers FL, FR, and SW were firing. On doing a bit of searching of the intrawebs, I'm led to believe that my Denon will only decode a 5.1 signal that arrives at it via HDMI (can anyone with a Denon AVRx2200w or similar confirm that?). That seems to rule out direct streaming or using the USB port.

So now I have to find a way to pass those files into my amp via HDMI. I don't have disc burning facilities on my PC (its an HP "all in one" with no CD/DVD slot) so I can't do that. Any ideas?

I'm really keen to give SpecWeb a proper try, and there are several albums in my collection that I'd love to hear in some sort of surround mix. Help!

Lastly, can anyone help me with the Spambot questions on the SpecWeb forum registration page? They don't seem to want to accept any answer that I provide!!
 
Back
Top