Audiolense Convolver vs Hang Loose Convolver

[ShadyHarbinger]

I use both convolvers. Audiolense with .alc format and HLC with .wav format.
I noticed there are huge sound quality differences between them.
When I switch between filters the differences between their quality are much more noticeable in AL convolver han in HLC. Selecting another filter in HLC is if it did not switch.
If Bernt read this would You so kind as to explain if there are any special algorithms is used in .alc format and your convolver?

I use VBAsioBridge for AL and VAC (Virtual Audio Cable) for HLC. VBAsioBridge+AL consumes a lot of resources (30% CPU). HLC+VAC only 1,6%. Something is suspicious here. I am sure VAC is a very well-developed and optimized application unlike VBAsioBridge still...

Edited: It seems there is a bug in HLC causing the selected filter does not load.

 

[ShadyHarbinger]

There are two issues: one is that HLC does not switch the filter and the huge sound quality difference (HLC+VAC sounds better).
Probably the reason for the latter is the much worse quality of VBAsioBridge/HifiCable.

 

[juicehifi]

Both convolvers produce the same sound quality. Convolution is a strictly defined mathematical operation. If you hear differences you need to look elsewhere in your system for the cause..

 

[whoareyou]

Edited: It seems there is a bug in HLC causing the selected filter does not load.

A quick check switching between filters and at least the amount of gain (not a critical listen) changes with each of my filter selections (this is with JRiver and the plugin i.e. not standalone HLC).
You hear no differences?

 

[Mitchco]

@whoareyou Problem solved, user error.

I measured Bernt's convolver some time ago as a Beta tester, but I can't find the measurements as that was 2 computers ago. But I do remember with a Dirac pulse loaded it was flat frequency and phase response, as it should be.
I have also compared HLC to several other convolvers and only EqualizerAPO seemed to have some sort of convolution bug...

Here I compare measurements of HLC with Roon's convolver: https://audiophilestyle.com/forums/...b-headphones-to-the-pinnacle/#comment-1243114
Identical.

Like @juicehifi says, convolution is a well defined mathematical operation, And if implemented correctly, there will be no measurable difference, letalone an audible one. If differences are heard, then it is likely something else which is the issue.

 

[ShadyHarbinger]

@whoareyou Problem solved, user error.

I measured Bernt's convolver some time ago as a Beta tester, but I can't find the measurements as that was 2 computers ago. But I do remember with a Dirac pulse loaded it was flat frequency and phase response, as it should be.
I have also compared HLC to several other convolvers and only EqualizerAPO seemed to have some sort of convolution bug...

Here I compare measurements of HLC with Roon's convolver: https://audiophilestyle.com/forums/...b-headphones-to-the-pinnacle/#comment-1243114
Identical.

Like @juicehifi says, convolution is a well defined mathematical operation, And if implemented correctly, there will be no measurable difference, letalone an audible one. If differences are heard, then it is likely something else which is the issue.

Thank You for confirming Mitch's statement!
I can use AL Convolver with VBAudio only while HLC with LoopBE/VAC.
Now I am sure that the reason is VBAsioBridge and Hifi-Cable. It's probably got a worse implementation than LoopBEAudio or VAC.

By the way Mitch. Could you please provide an option in HLC to enable/disable the zero latency filter switching feature? Would be a great and nice option!
The advantages we would get:
1. The actual latency could be the latency of the selected filter instead of the highest one.
2. CPU resources could be saved because it would process the selected filter only.

 

[ShadyHarbinger]

A quick check switching between filters and at least the amount of gain (not a critical listen) changes with each of my filter selections (this is with JRiver and the plugin i.e. not standalone HLC).
You hear no differences?

No difference but I found out that HLC is buggy and does not handle every configuration variation etc. I have to manually edit Aduilense cfg files and it works afterwards. Annoying because HLC does not tell anything about what happened, or what is wrong. I can do nothing but wait for new versions. Hopefully, it will be more user-friendly sooner or later.

 

[whoareyou]

No difference but I found out that HLC is buggy and does not handle every configuration variation etc. I have to manually edit Aduilense cfg files and it works afterwards. Annoying because HLC does not tell anything about what happened, or what is wrong. I can do nothing but wait for new versions. Hopefully, it will be more user-friendly sooner or later.

Interesting. I've had absolutely zero issues going back several releases (plug-in). Haven't used standalone in a while, but when I did most of my issues were errors outside of HLC's control. Virtual drivers, buffer sizes, sample rates, hardware tha couldn't keep up with number of filter banks, etc..

I try to stay away from the standalone solution because I usually end up with too many moving pieces that are stable for a while, and then all of a sudden break on me. Seems like many people struggle with all of the moving parts since you're basically cobbling stuff together that doesn't have the greatest compatibility record.

 

[ShadyHarbinger]

Interesting. I've had absolutely zero issues going back several releases (plug-in). Haven't used standalone in a while, but when I did most of my issues were errors outside of HLC's control. Virtual drivers, buffer sizes, sample rates, hardware tha couldn't keep up with number of filter banks, etc..

I try to stay away from the standalone solution because I usually end up with too many moving pieces that are stable for a while, and then all of a sudden break on me. Seems like many people struggle with all of the moving parts since you're basically cobbling stuff together that doesn't have the greatest compatibility record.

I am a software developer for 3x years. I exactly know how a program should behave and be tested and what a good user experience should be.
HLC is good at its core. It does its job very well but has not "finished" yet. It needs more feedback on what is wrong if the user does something or a few (in my opinion) important basic features. Then this convolver gets a much wider audience. I use it as a system-wide solution and I know that many more people would use it.
You are right. There are issues with virtual drivers as you said but there is a solution for almost everything so I have no issue with the moving parts. I have no problem with "moving parts". It works flawlessly regardless I play a game, watch a movie, or listen to music.
The "drawbacks" that I am talking about are strictly related to the convolver and have nothing to do with the moving parts. You also mentioned one thing: more filters more resource usage. A great example how HLC could be improved in a way. All filters are processed parallelly but they should not be if the user does not want to test different filters. If they know which filter they prefer and use then that one should be computed only. It can be solved with an option where users can enable/disable this feature.
Both Mitch and Bernt are very helpful and I know they improve their applications and do their best. As far as I know, they both develop these complex apps alone which is reasonable and I respect them for that.
I cannot use WASAPI Exclusive/ASIO with virtual cables (without an audio repeater). I could use the HLC that way with an audio repeater but I do not want more latency so I accept this and live with it.
Still, it provides so much better sound quality with a filter. Totally worth the hassle.

 

[whoareyou]

I am a software developer for 3x years. I exactly know how a program should behave and be tested and what a good user experience should be.
HLC is good at its core. It does its job very well but has not "finished" yet. It needs more feedback on what is wrong if the user does something or a few (in my opinion) important basic features. Then this convolver gets a much wider audience. I use it as a system-wide solution and I know that many more people would use it.
You are right. There are issues with virtual drivers as you said but there is a solution for almost everything so I have no issue with the moving parts. I have no problem with "moving parts". It works flawlessly regardless I play a game, watch a movie, or listen to music.
The "drawbacks" that I am talking about are strictly related to the convolver and have nothing to do with the moving parts. You also mentioned one thing: more filters more resource usage. A great example how HLC could be improved in a way. All filters are processed parallelly but they should not be if the user does not want to test different filters. If they know which filter they prefer and use then that one should be computed only. It can be solved with an option where users can enable/disable this feature.
Both Mitch and Bernt are very helpful and I know they improve their applications and do their best. As far as I know, they both develop these complex apps alone which is reasonable and I respect them for that.
I cannot use WASAPI Exclusive/ASIO with virtual cables (without an audio repeater). I could use the HLC that way with an audio repeater but I do not want more latency so I accept this and live with it.
Still, it provides so much better sound quality with a filter. Totally worth the hassle.

And before retiring I was a software engineer / system architect for 40 years, so I can assure you I understand the difficulties involved and how to build quality applications. I also understand, and appreciate the the difficulties in interpreting and implementing various specifications for cross functional applications.

Not sure what any of this has to do with my experience of finding HLC to be stable and easy to use. As I said, I have much less experience with stand-alone, and where I had issues it was always with the interoperability between drivers.

I don know if you have issues with HLC, and as you indicated, Mitch will be happy to help you out. This is also an area I am familiar with as I've found @Mitchco to be one of the best people to deal with.

 

[Mitchco]

No difference but I found out that HLC is buggy and does not handle every configuration variation etc. I have to manually edit Aduilense cfg files and it works afterwards.

As mentioned to you before, HLC does not automatically load different channel format .cfg's which is in my backlog to be completed. I don't know anyone else manually editing Audiolense.cfg files, as HLC will handle both absolute and relative paths. No-one else has reported this issue. As far as real "bugs" are concerned, there are none ;-)

HLC does have error handling if you try and load a filter that HLC cannot read:

HLC also logs any real errors in a log file, location described in the ops guide.

As it says on the web site, "Hang Loose Convolver (HLC) is a specially designed stereo and multichannel convolver to compare “level matched” convolution filters in real time with “instant switching” between filters."
If you wish to use HLC as a "regular" convolver, simply load one filterbank which will use the minimum amount of resources and will display the latency for that one filter.

Anyway, this is Bernt's Audiolense forum. If you have specific HLC issues, please send me an email.

 

[ShadyHarbinger]

I successfully configured the HLC system-wide using ASIO4All and VAC with minimal latency through the HDMI port. This way I got the best possible sound quality from a multifunctional daily used gaming PC whatever I am doing. Rock solid and works flawlessly.

 

[MNSMike]

I can’t believe nobody is talking about the new VB Audio Matrix yet:

VB-Audio Matrix

VB-Audio Matrix, Real Time Audio Router by V.Burel

vb-audio.com

With this you can finally route audio from any application and I/O on Windows bit perfect. Now I can finally send audio from my media player through the convolver, then out via ASIO. It also allows you to bridge Dante devices to Ravenna devices without the 24/48 pcm limit.

For example I’m sending the audio from my TV’s HDMI eARC port though a Bluesound Node. And out of the Node via SPDIF to a SPDIF to Dante box. Then from there into Matrix, then HL convolver, and out via Ravenna to a Ravenna DAC. All without losing lip sync. And sample rate support up to 24/192.

And if you want to get really crazy there’s also coconut:

VB-Audio Matrix Coconut

VB-Audio Matrix Coconut, Ultimate Real Time Audio Router by V.Burel

vb-audio.com