Question for the DSP Experts - how often do you DSP?

[Pareto Pragmatic]

In a Word: NEVER. And this is regardless if I'm playing an LP record, a CD or High Rez digital files.

I get what you are saying here, and elsewhere in the thread.

I enjoy sound in my room without DSP... up to about 68dB/1m levels. By 80dB/1m levels it sounds horrible. And since I like very much to hit that level, I really don't have much of a choice. At least no choice that would not run into 5 figures of remodeling the house, or to build a dedicated addition. A mic and REW, much more cost effective.

I think some of your reasoning is flawed on technical aspects, but I generally agree with your philosophy. Keep things as simple as possible, and if you like what you have there is no reason to add anything else.

 

[boxerfan88]

If you're just coming to DSP as anything less than an audio engineer, mathematician or physicist, the learning curve seems pretty steep. To be clear, I'm not talking about plug-and-play solutions, like Audessy. I refer to REW, MSO, phase alignment, multiple sub setup, yada yada. Beyond grasping the theory there's a helluva lot of experimentation and experience that comes into play. There's software to learn, and it's complicated. REW and MSO aren't what I'd call simple or intuitive.

How very true!!

So, on to the question: How often do you some kind of DSPing on your system(s)? I myself have been pretty well sucked in, finding myself doing more DSP fiddling than actually listening. No bueno. What's your experience? Thanks and cheers,

My journey took 6 months. The desire to migrate to foobar 64b forced me to go deep into convolver dsp & designing my own FIR filter. In fb64 there is no crossover component, i had to implement it via convolver dsp. Initially I had 2 convolvers, one for room correction, one for crossover. Running two caused issues with foo_record. Took me some time to learn how to merge them into one. Then there is that learning rabbit hole about phase correction. Also the other learning rabbit hole about time aligning subwoofer using delay. Then there was this other rabbit hole of incorporating ToTape8 tape head bump EQ (coloration) that i liked. I’ve arrived at my “end game” DSP convolver setting (for now). I had to reduce the house sample rate from 176k to 88k because the higher sample rate wasn’t giving me sufficient resolution in the low frequency region for room correction. Its a trade off between resolution & latency. Final step was to trim down the taps to the minimum required, i settled on 90ktaps. I haven’t made any changes for close to 2 months now. That’s ages compared to the days I was tweaking dsp, measuring , listening weekly.

I learned a lot from my dsp journey. I’m sure you will too.

 

[Kiko1974]

I get what you are saying here, and elsewhere in the thread.

I enjoy sound in my room without DSP... up to about 68dB/1m levels. By 80dB/1m levels it sounds horrible. And since I like very much to hit that level, I really don't have much of a choice. At least no choice that would not run into 5 figures of remodeling the house, or to build a dedicated addition. A mic and REW, much more cost effective.

I think some of your reasoning is flawed on technical aspects, but I generally agree with your philosophy. Keep things as simple as possible, and if you like what you have there is no reason to add anything else.

I see you got my point, keep things as simple as possible.
Sometimes less is more.

 

[boxerfan88]

I have several problems with DSP:
1- If I'm playing an LP record, I want a full Analogue chain. I don't see the point of playing an analogue medium and digitize It for playing, unless for a certain album, the LP record IS the best sounding available version.
2- DSD can't be processed natively on the digital domain, It needs to be converted to PCM first. I want my SACD's and DSD files to be played natively, with no intermediate PCM convertion.
3- Many A/V Receivers that feature DSP/Room Correction, are unable to play High Rez PCM, either at 96/24 or 192/24 , when any kind of DSP/Room Correction is on.
Somewhere well hidden in the User's Manual states that if you want to play High Res PCM at its full native resolution, change to the "Direct" or "Puré Direct" (depending on the brand the name changes) mode.

Try something like this https://www.accuphase.com/model/dg-68.html in your analog chain, it might just surprise you how close to end game system it can take you compared to no DSP.

 

[Kiko1974]

Try something like this https://www.accuphase.com/model/dg-68.html in your analog chain, it might just surprise you how close to end game system it can take you compared to no DSP.

Thanks but no thanks. My simple and humble Hi Fi set up IS fine as It now.

 

[voodooless]

I have several problems with DSP:
1- If I'm playing an LP record, I want a full Analogue chain. I don't see the point of playing an analogue medium and digitize It for playing, unless for a certain album, the LP record IS the best sounding available version.
2- DSD can't be processed natively on the digital domain, It needs to be converted to PCM first. I want my SACD's and DSD files to be played natively, with no intermediate PCM convertion.
3- Many A/V Receivers that feature DSP/Room Correction, are unable to play High Rez PCM, either at 96/24 or 192/24 , when any kind of DSP/Room Correction is on.
Somewhere well hidden in the User's Manual states that if you want to play High Res PCM at its full native resolution, change to the "Direct" or "Puré Direct" (depending on the brand the name changes) mode.

None of these are real problems…

1. Digitizing LPs is useful if you want to apply EQ and room corrections. The end result will be vastly superior to a purely analog chain
2. Same as above. There is zero advantage to native playback
3. There is zero advantage to high-res audio either, so again, see point 1

Having the means to actively correct your speakers and account for room effects will always trump any other so called improvement in the signal chain by many orders of magnitude.

 

[Kiko1974]

None of these are real problems…

1. Digitizing LPs is useful if you want to apply EQ and room corrections. The end result will be vastly superior to a purely analog chain
2. Same as above. There is zero advantage to native playback
3. There is zero advantage to high-res audio either, so again, see point 1

Having the means to actively correct your speakers and account for room effects will always trump any other so called improvement in the signal chain by many orders of magnitude.

DSP is of no interest to me. Is a solution for a problem I don't see/hear on my humble Hi Fi system. As this system is for my use alone (as I live alone) and I like the sound as It is, I don't see the need of using DSP.
I've used DSP in the past as I owned a Home Theater set up. The core of this system was a Pioneer SC LX-76 from 2012, second best on that year European range. I used its automatic set Up and Room correction for years.
I went back to plain stereo three years ago and I couldn't be happier.

 

[DWPress]

I've been doing the DSP/DRC thing for years trying to perfect a 3-way active XO + 2 subs. Started with miniDSP and now full chain on the computer. I even run my turntable into the system for correction.

Now I don't change things or re-measure unless something drastic has changed in the room. Every once in a while I'll do a MMM sweep just to be sure levels are still good and maybe a regular sweep in REW to check time domain stuffs. It's been a year since my last deep dive.

 

[theREALdotnet]

So, on to the question: How often do you some kind of DSPing on your system(s)?

Ideally, whenever something in the room changes (speaker positioning or toe-in, furniture, treatment) or when equipment changes. Also, when I learn something new or want to try something new.

In reality, I didn't re-do my DSP as often I should (as above) while I was doing it all manually – measurement with REW, vector arithmetic, rePhase, back to REW for some more vector arithmetic, calculating filters, exporting filters, typing them into a PEQ plugin. It was just a lot of effort.

Nowadays with Dirac I do it as often as I should

Proper measurement is still a bit of an effort, as is playing around to find the best cross-over point, but the overall time spent and the potential for mistakes are drastically reduced.

 

[Berwhale]

Not often enough. I redecorated and rebuilt my study last year; my desk and seating position have rotated by 90 dregrees and my monitors (iLoud MTMs) are mounted in a completely different way. I haven't got around to re-measuring my setup in REW yet, although I have run the auto calibration on the MTMs.

Hmmm, now I have a motorized standing desk, I should measure the MTMs in both seating and standing positions and then work out some way of automatically applying appropriate correction based on desk height.

 

[IEales]

After 8 years, I may be done on the tri-amped with dual powered subs system.
The thing I liked most about it was the ability to easily what-if. 9 years ago I was going to design an analogue crossover but quickly put that idea to rest.
Frequency response is how I like it and I'm able to rotate it easily to compensate for bass or treble shy recordings.
Room nodes are adequately tamed.

More importantly, time correction is exact.

 

[Keith_W]

DSP is of no interest to me. Is a solution for a problem I don't see/hear on my humble Hi Fi system. As this system is for my use alone (as I live alone) and I like the sound as It is, I don't see the need of using DSP.
I've used DSP in the past as I owned a Home Theater set up. The core of this system was a Pioneer SC LX-76 from 2012, second best on that year European range. I used its automatic set Up and Room correction for years.
I went back to plain stereo three years ago and I couldn't be happier.

I cringe a little whenever somebody says they are turned off DSP because of negative experiences like that. I'll put it this way - the influence of DSP is profound. It can be profoundly good, or profoundly bad. If DSP does something wrong, it will be profoundly bad. So I do not blame average punters who get an AVR, use those hockey puck microphones, follow the instructions, and the result is bad sound. The awful sound prejudices them against DSP forever.

AVR's have NO HOPE of getting good sound with automation and hockey puck microphones. At best, AVR's should avoid degrading the sound through inappropriate DSP.

For DSP to work, you need two things - (1) proper measurements and (2) correct decisions. You need a calibrated condenser mic, either XLR or USB, and a microphone tripod. Even if you have a proper microphone, you still need to know how to take measurements. Otherwise the measurements will be unrepresentative of reality and the corrections will be inappropriate. If software is to be automated, it has to be good software - programmers can not possibly anticipate the thousands of permutations of systems and listening rooms, so they should allow a fair degree of customisation. And finally, the user needs to know what to correct and what to leave alone.

So if you take bad measurements (the inevitable result of the cheap mics that come with AVR's) and use bad software - bad corrections will be the result. If this was your experience, I can understand your position.

 

[IEales]

So I do not blame average punters who get an AVR, use those hockey puck microphones, follow the instructions, and the result is bad sound

Methinks thou are a tad harsh on the AVR and microphone. While yes, I've heard some truly awful setups, but they were done by people with zero knowledge of what they are doing, often with grossly inadequate rooms, speakers and placement.

IF good sound is a priority - after it's still radio with pictures - and they are willing to take advice, the same AVR and mic in the hands of an experienced driver may create an excellent sound experience.

 

[Kiko1974]

I cringe a little whenever somebody says they are turned off DSP because of negative experiences like that. I'll put it this way - the influence of DSP is profound. It can be profoundly good, or profoundly bad. If DSP does something wrong, it will be profoundly bad. So I do not blame average punters who get an AVR, use those hockey puck microphones, follow the instructions, and the result is bad sound. The awful sound prejudices them against DSP forever.

AVR's have NO HOPE of getting good sound with automation and hockey puck microphones. At best, AVR's should avoid degrading the sound through inappropriate DSP.

For DSP to work, you need two things - (1) proper measurements and (2) correct decisions. You need a calibrated condenser mic, either XLR or USB, and a microphone tripod. Even if you have a proper microphone, you still need to know how to take measurements. Otherwise the measurements will be unrepresentative of reality and the corrections will be inappropriate. If software is to be automated, it has to be good software - programmers can not possibly anticipate the thousands of permutations of systems and listening rooms, so they should allow a fair degree of customisation. And finally, the user needs to know what to correct and what to leave alone.

So if you take bad measurements (the inevitable result of the cheap mics that come with AVR's) and use bad software - bad corrections will be the result. If this was your experience, I can understand your position.

I used a Sennheiser microphone, not the dirty cheap one included on the Pioneer SC LX-76. And I did a final tweaking myself with a proper sound level preassure meter.
After a coupled of years of mixed results with the Pioneer SC LX-76 (which costs around 1900/2000 Euros on early 2.012) built in self adjust feature and Room correction (I don't remember its name), I ended Up doing my home Theater set Up "the old way". Measuring with a measuring tape and the help of a friend, the distance from the different speakers to the listening point, and adjusting the different Channel levels with the sound level preassure meter.
I used a bit of equalization, mainly on the center channel and the rears, and that was It.
In my opinion, that was using DSP but not relaing on an algorithm.

 

[klettermann]

How very true!!

My journey took 6 months. The desire to migrate to foobar 64b forced me to go deep into convolver dsp & designing my own FIR filter. In fb64 there is no crossover component, i had to implement it via convolver dsp. Initially I had 2 convolvers, one for room correction, one for crossover. Running two caused issues with foo_record. Took me some time to learn how to merge them into one. Then
(1) there is that learning rabbit hole about phase correction. Also the (2) other learning rabbit hole about time aligning subwoofer using delay. Then there was this other rabbit hole of incorporating ToTape8 tape head bump EQ (coloration) that i liked. I’ve arrived at my “end game” DSP convolver setting (for now). I had to reduce the house sample rate from 176k to 88k because the higher sample rate wasn’t giving me sufficient resolution in the low frequency region for room correction. Its a trade off between resolution & latency. Final step was to trim down the taps to the minimum required, i settled on 90ktaps. I haven’t made any changes for close to 2 months now. That’s ages compared to the days I was tweaking dsp, measuring , listening weekly.

I learned a lot from my dsp journey. I’m sure you will too.

Encouraging and interesting, thanks. Over the past few days I think I may be over the hump with 1 and 2 above AND facile enough with the work flow/processes to [measure/listen => tweak => remeasure/relisten => repeat fast enough to make it painless. Below is a kind of representative example entailing time alignment of subs (there are 2), mains, setting XOs and EQing with 1/ smoothing. Phase plots unsmoothed. This was just REW since I'm really only interested in MLP. Room is untreated apart from carpet and 4 ASC bass traps in corners. Perfect? Certainly not, but way better. Most important it sounds great. Very clean, defined with super soundstage/imaging. Maybe I'll tweak more, especially with Dirac, but for the time being my energies will return to listening. Cheers,

1. Uncorrected L+R+subs 2. Aligned, XO'd and EQ'd L+R+subs.

3. Subs, no EQ, no XO, no nothing 4. Subs+L+R, Aligned, EQ'd, XO'd

 

[digitalfrost]

I rarely change my DSP settings, but after a couple of months, I just might like something different and then change the settings. However, the meme is quite accurate. Everything runs through my PC.

The turntable is connected to the PC, so I can have DSP while listening to vinyl, and I do not mind the AD DA, heck I have 60-70dB noisefloor anyway at best. And I also use DSP for the headphones. I also use DSP on my cellphone. And the microphone that is connected to my PC? Yup, you guessed it.

There is literally not a single thing that records or outputs audio that I use that does not have at least the capability to apply DSP. DSP is just fantastic.

Not only does it improve the sound quality massively for comparatively little cost, but it allows me to enjoy a different sound signatures if I grow tired of something without the need to buy new gear.

 

[dedobot]

No dsp or eq . Eventually sometimes simple tone controls bass/treble. Why ? Cause when I setup a correction, in the beginning I'm satisfied, a hour or less after something starts to gnaw at me. Knowing my modest knowledge of music and sound, I start to build suspect that I've messed things and finally when the suspicion go massive - I m back to the starting point. I've been through this cycle hundreds of times and I've long since given up on corrections.

 

[izeek]

If you're just coming to DSP as anything less than an audio engineer, mathematician or physicist, the learning curve seems pretty steep. To be clear, I'm not talking about plug-and-play solutions, like Audessy. I refer to REW, MSO, phase alignment, multiple sub setup, yada yada. Beyond grasping the theory there's a helluva lot of experimentation and experience that comes into play. There's software to learn, and it's complicated. REW and MSO aren't what I'd call simple or intuitive.

So, on to the question: How often do you some kind of DSPing on your system(s)? I myself have been pretty well sucked in, finding myself doing more DSP fiddling than actually listening. No bueno. What's your experience? Thanks and cheers,

REW and MSO aren't what I'd call simple or intuitive

aint that the truth. i finally gave up. havent used my minidsp that i bought like 2-3yrs ago, yet.

 

[gberchin]

Signal processing, whether digital or analog or somewhere in-between, makes heavy use of complex (as in real and imaginary parts) arithmetic and time/frequency relationships. Nobody should expect it to be simple, because it isn't.

 

[Waxx]

Most people want a set and forget tool, not something with endless tweaking. Here on this forum are mainly the nerds, but the general public is not interested when it gets complicated. That is where dsp need to go a long way i'm affraid, and where AI could help probally. And it's probally also the main reason why it's not that mainstream yet to use dsp. Passive speakers are set and forget, and that advantage is way more important for the mainstream public than accuracy.