Something stirs in the Undergrowth

Dedicated to those large boxes at one end of the room
vinylnvalves
Old Hand
Posts: 1074
Joined: Sat Jan 26, 2013 12:08 pm

#2596 Re: Something stirs in the Undergrowth

Post by vinylnvalves »

Sound-wise, the improvement is subtle, but I don't think I'm particularly sensitive to phase shifts. I can say that there's a certain clarity that I like, which I'm not sure was there before. Backing vocals in Fleetwood Mac's Dreams, for instance, are obviously distinct voices, rather than the layer of background mush that lesser systems render them as.
Just done the test .... playing Fleetwood Mac’s Dreams. I can hear all the backing vocals clear and distinct, I don’t have linear phase set up..... would it sound even better if I made it linear... shame I haven’t got enough taps to do a proper crossover. :( I don’t have any sharp phase changes, however I suspect its because only use one driver to cover 300hz to 6000hz.

Btw - that track is awfully mixed, the bass is to hot, and dominates the sound.
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#2597 Re: Something stirs in the Undergrowth

Post by IslandPink »

I'm reading with interest . However I am also getting alternative reports from elsewhere about even well-done FIR not being that good. Will take the ears to decide when we can get some face to face experiments done again.

In the meantime the spark of a new horn project is emerging.
This one is a variant of the FE208Ez-based lower-mid horns, with a concept built around it. Length is 1.9m with expansion of about 50Hz exponential increasing to 60Hz in the last 50cm or so. Mouth 3300cm^2.
It looks good with pretty flat phase to 100Hz, a small amount of ripple.
One wavelength setback at 600Hz is the approach here. I had surprisingly good results on vocals with 2 waves at 600Hz with one version of the folded horn ( Nick was here ) and Yuichi, so this should be significantly better.
I was going to have as an L-horn with the driver above, but when I started looking at the lengths, with the 1-wave setback, it seemed possible to put the driver on the floor and bring the horn upwards, then forwards. This has the advantage of putting the upper frequencies at crossover ( 500Hz to 800Hz region ) coming out from the top of the horn, because they propagate more around the outside.
Back volume of some sort will be used ( to block rear driver sound) but not aggressively loaded, hence it would be big and could form some sort of support ; but the horn mouth would probably have supports reaching down to the ground for stability.

Here's a crude drawing from Paint -
Concept_1p9m.jpg
"Once you find out ... the Circumstances ; then you can go out"
User avatar
Dave the bass
Amstrad Tower of Power
Posts: 12273
Joined: Tue May 22, 2007 4:36 pm
Location: NW Kent, Darn Sarf innit.

#2598 Re: Something stirs in the Undergrowth

Post by Dave the bass »

vinylnvalves wrote: Sat May 23, 2020 3:58 pm
Btw - that track is awfully mixed, the bass is to hot, and dominates the sound.
Oooffff! ^^^ Overruled :lol:

FWIW. I think that track is mixed excellently and 'fits' the instrumental make up of that song, bass guitar and kick drum are locked and mixed forward yes but that's what holds the whole song together. It works in that context (IMO).
"The fat bourgeois and his doppelganger"
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#2599 Re: Something stirs in the Undergrowth

Post by chris661 »

IslandPink wrote: Sat May 23, 2020 7:56 pm I'm reading with interest . However I am also getting alternative reports from elsewhere about even well-done FIR not being that good. Will take the ears to decide when we can get some face to face experiments done again.
I'm curious, Mark, as to where you're reading those reports. The reviews of FIR stuff I've read have been positive as a subtle but definite improvement - the sort of thing that makes music easier to enjoy because it's one less thing for your brain to compensate for.

TIA,
Chris
User avatar
Nick
Site Admin
Posts: 15705
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#2600 Re: Something stirs in the Undergrowth

Post by Nick »

Yep, I am with you on that Dave. Anyway
even well-done FIR not being that good
Just pointing out that FIR is just a category of digital filters, finite impulse response refers to the fact that the response to a impulse can be guaranteed to go to zero in finite time. The other class IIR, has a potentially (but rarely) infinite impulse response. There are many variations of both types. So expecting all FIR to be good and all IIR to be bad is simplistic and almost certainly wrong. Generally FIR filters need more hardware to implement that IIR if we are talking about digital ones. Most analogue filters are IIR, but you could make a analogue FIR with a tapped delay line. Its also possible to make a non linear phase FIR, and an almost linear phase IIR filter.

Andrew L, had a link to a very good (I think) book on DSP that covers the basics. I am sure he could find a link to it. Its something I know very little about other than how to use one to filter in software if needed.

Just saying.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
Greg
Social outcast
Posts: 3198
Joined: Wed May 23, 2007 11:14 am
Location: Bristol, UK

#2601 Re: Something stirs in the Undergrowth

Post by Greg »

Dave the bass wrote: Sat May 23, 2020 8:03 pm
vinylnvalves wrote: Sat May 23, 2020 3:58 pm
Btw - that track is awfully mixed, the bass is to hot, and dominates the sound.
Oooffff! ^^^ Overruled :lol:

FWIW. I think that track is mixed excellently and 'fits' the instrumental make up of that song, bass guitar and kick drum are locked and mixed forward yes but that's what holds the whole song together. It works in that context (IMO).
Completely agree Mr B. This thread caused me to listen this afternoon. Yep, I got all the backing vocals and the bass was distinctly acute both from the drums and the guitar and certainly balanced in the mix. Sounded good to me :D
User avatar
IslandPink
Amstrad Tower of Power
Posts: 10041
Joined: Tue May 29, 2007 7:01 pm
Location: Denbigh, N.Wales

#2602 Re: Something stirs in the Undergrowth

Post by IslandPink »

Nick wrote: Sat May 23, 2020 10:08 pm
Just pointing out that FIR is just a category of digital filters, finite impulse response refers to the fact that the response to a impulse can be guaranteed to go to zero in finite time. The other class IIR, has a potentially (but rarely) infinite impulse response. There are many variations of both types. So expecting all FIR to be good and all IIR to be bad is simplistic and almost certainly wrong. Generally FIR filters need more hardware to implement that IIR if we are talking about digital ones. Most analogue filters are IIR, but you could make a analogue FIR with a tapped delay line. Its also possible to make a non linear phase FIR, and an almost linear phase IIR filter.
OK so I don't know in exactly which way this system was well done.
Have a chat with CV, maybe you can understand the ins & outs of this better than I could. I know he's considering only using digital delay for Bass or HF for a 'long' horn.
"Once you find out ... the Circumstances ; then you can go out"
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#2603 Re: Something stirs in the Undergrowth

Post by ed »

I just had a flashback to the seventies....I remember having a conversation with a chum(he had Mitchel hydraulic reference and quad33/303 and b&w dm2s) he was arguing that amplifiers with tone controls were getting away from what the music should sound like. Like many people at the time he was looking to change his amp for one that had no tone control whatsoever because that was the purist approach.

Me? I really don't know anymore...but try applying that mindset to the DSP approach.

I recently read this on the soundguys website:

"By using a modern DSP, you no longer need to hope that your audio gear will sound good, you can force it to at any time by having the electronics compensate for shortcomings."

bearing in mind that 'sound good' is a personal and subjective metric...what exactly are you chaps trying to achieve with DSP?
There's nowhere you can be that isn't where you're meant to be
User avatar
Nick
Site Admin
Posts: 15705
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#2604 Re: Something stirs in the Undergrowth

Post by Nick »

what exactly are you chaps trying to achieve with DSP?
If I understand correctly, filters with group delay instead of phase delay.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#2605 Re: Something stirs in the Undergrowth

Post by ed »

Ahh ha..much obliged, thanks.

Coming from a strictly user perspective, having used it just to emphasise instruments good points or dismiss their failings, bringing them(the instruments) further forward or further back in the mix.

I will read, much to learn there is!
There's nowhere you can be that isn't where you're meant to be
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#2606 Re: Something stirs in the Undergrowth

Post by chris661 »

ed wrote: Sat May 23, 2020 10:56 pm what exactly are you chaps trying to achieve with DSP?
Pretty much what Nick said.


FIR filtering allows us to exchange a system-wide delay for the ability to arbitrarily alter the phase response.

The simplified idea is that if you implement a delay for the entire system, you then have wiggle room to move some frequencies forwards in time, others backwards, and get it so that, when the sound emits from the speakers, all the time information of the record is maintained.

Speakers delay some frequencies more than others - the group delay curve is one way of showing this, and the phase curve is another. If a speaker has a 90-degree phase lag at 1kHz, then we can use 0.25ms of delay time to correct that. We'd re-align the signals so that 1kHz hits the speaker 1/4 of a cycle early, and everything else arrives afterwards.

The speaker applies its inherent phase shift, which we've effectively unwrapped, and the signal leaving the speaker no longer has that 90-degree lag at 1kHz.

If we wanted to play with larger phase shifts or lower frequencies, then of course a larger delay would be needed.

That's how it's all working in an "analogue" kind of way. The implementation of all that requires some digital signal theory.


FWIW, Nick, the 5-year-old i5 laptop that's running the FIR processing, plus a few IIR filters in the bass, is using 2% of one CPU core to do everything at 192kHz. It doesn't need much processing power, but it does need to have the capability to work in this way.

Chris
User avatar
Nick
Site Admin
Posts: 15705
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#2607 Re: Something stirs in the Undergrowth

Post by Nick »

The simplified idea is that if you implement a delay for the entire system, you then have wiggle room to move some frequencies forwards in time, others backwards, and get it so that, when the sound emits from the speakers, all the time information of the record is maintained.
I don't believe its just that, yes, you can use delay to align the drive units. But I think (or thought) that part of Marks goal was to avoid the phase delay effects of IR filters (including normal analogue xovers). So for example a 100Hz tone with a xover at 200Hz would on the way in have its third harmonic aligned with the original, but on the way out the relationship between them will have altered.

FIR filters will introduce a delay as part of their operation and using a bunch of them will require additional delay to realign them (though I guess if they all used the same number of taps that would not be needed).
is using 2% of one CPU core to do everything at 192kHz.
Well, I guess it all depends on the number of taps used. When I looked at it for a bit last year (and never got any further than a FPGA board) once you get to 10's of thousands tap filters the multiply/sum requirements was not trivial. I use a 32 tap FIR on the control output of that direct drive turntable I brought to Owston last year, and that’s not that complex, but it is only implementing a integer 10Hz smoothing filter and that’s using a 8 bit AVR

I assumed that folk were talking about more complex ones. Again, maybe wrong.
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#2608 Re: Something stirs in the Undergrowth

Post by ed »

chris661 wrote: Sun May 24, 2020 9:19 am

FWIW, Nick, the 5-year-old i5 laptop that's running the FIR processing, plus a few IIR filters in the bass, is using 2% of one CPU core to do everything at 192kHz. It doesn't need much processing power, but it does need to have the capability to work in this way.

Chris
not strictly relevant, but I struggle to correlate that with the past use of the UAD cards on machines prior to the intel core machines that couldn't cope with the processing in digital workstation software. My experience was that latency went through the roof in the pentium era and some time after. The DSP work brought machines to their knees, hence the prolific use of UAD cards.

I need to read more.....perhaps there's more to DSP than just filtering FIR
There's nowhere you can be that isn't where you're meant to be
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#2609 Re: Something stirs in the Undergrowth

Post by chris661 »

Nick wrote: Sun May 24, 2020 10:57 am
The simplified idea is that if you implement a delay for the entire system, you then have wiggle room to move some frequencies forwards in time, others backwards, and get it so that, when the sound emits from the speakers, all the time information of the record is maintained.
I don't believe its just that, yes, you can use delay to align the drive units. But I think (or thought) that part of Marks goal was to avoid the phase delay effects of IR filters (including normal analogue xovers). So for example a 100Hz tone with a xover at 200Hz would on the way in have its third harmonic aligned with the original, but on the way out the relationship between them will have altered.

FIR filters will introduce a delay as part of their operation and using a bunch of them will require additional delay to realign them (though I guess if they all used the same number of taps that would not be needed).
is using 2% of one CPU core to do everything at 192kHz.
Well, I guess it all depends on the number of taps used. When I looked at it for a bit last year (and never got any further than a FPGA board) once you get to 10's of thousands tap filters the multiply/sum requirements was not trivial. I use a 32 tap FIR on the control output of that direct drive turntable I brought to Owston last year, and that’s not that complex, but it is only implementing a integer 10Hz smoothing filter and that’s using a 8 bit AVR

I assumed that folk were talking about more complex ones. Again, maybe wrong.
I was talking about using simple ones, so far. Think I used 1000 taps for the correction in the graphs I posted further up: the stereo gets used for movies etc, so I couldn't mess up the lip sync. A couple of milliseconds seems to be fine.

However, for the fun of it, I tried out a longer filter to see how my laptop would cope. Now, my day-to-day laptop doesn't run the HiFi, and it's a bit more powerful -3rd gen i7. 1000000 (one million) taps = 2.1% CPU usage, and a 5-second delay at 192kHz.

The filter was pretty straightforward, and could probably have been done in fewer taps, but it was a linear-phase +12dB boost at 15Hz, and then a +180 to -180 phase swing from 25Hz to 63Hz, plus a couple of other parametric peaks and dips higher up to make sure it was working.


Ed, DAWs typically need to do a lot of simultaneous processing of signals to get anywhere, so I can understand why the older processors might struggle. FIR seems to be pretty light on CPU usage in comparison, but it is only two audio channels.

Chris
vinylnvalves
Old Hand
Posts: 1074
Joined: Sat Jan 26, 2013 12:08 pm

#2610 Re: Something stirs in the Undergrowth

Post by vinylnvalves »

Oooffff! ^^^ Overruled :lol:

FWIW. I think that track is mixed excellently and 'fits' the instrumental make up of that song, bass guitar and kick drum are locked and mixed forward yes but that's what holds the whole song together. It works in that context (IMO)..
I have a few different versions of fleetwood Mac track. “Dreams”. The observation I made yesterday was for the 2004 remastered version. Listening to the original 1977 version Mr McVie isn’t as pronounced, more balanced, - reissued stuff :x Rory Gallagher’s “Calling Card” is very similar almost unreal on the remastered version I have.

Back to what I use the DSP for.. Alongside the crossover i argument the bottom end. Adding around 12db to level up the sloping off of the drivers response, to get an almost flat response from 30hz upwards from the Supravox drivers. I use a Stonewall filter at 28 Hz .... 4hz above FS to avoid any driver resonance issues.
I have some BMS 4592’s attached to the big horns for experimentation with to see if all the minor issues with these drivers can be removed with a DSP, one day soon maybe. I am limited to 9600 taps at 48khz with the Nadja.. so no FIR capability, except at the top end of the frequency range, where I don’t need it.
As I mentioned the use of an amp with a DSP is one way of getting more from the bass drivers, all the benefits of sealed with the FR of a BR. Obviously my drivers are now 86db eff not 98db eff as the top end is being attenuated in reality. The other way I have seen and heard is to use jRivers where the input signal is modified - so std crossovers can be used, flat response but no benefit to phase. This approach is old hat now a days, processing power is not the limitation now.
Interestingly a friend of mine in the automotive industry a noise and vibration engineer jokes but is serious that the best audio systems are in cars. It’s a controlled environment ( until you open the window) so they can simulate and hardcode the DSP settings into the audio systems for each model.
Post Reply