Hello there,
I think I have lost a few years somewhere, and as a result, I am somewhat out-of-touch with DSP processing and its cost.
I thought it might be good for some members to learn more about it.
It would probably be good for 'posting' to start in a simple & basic way, then leading to more technical aspects.
I, myself look forward to some reading & thoughts. A question I have is:
Is DSP the future of audio alignment and 'perfection'?
I think I have lost a few years somewhere, and as a result, I am somewhat out-of-touch with DSP processing and its cost.
I thought it might be good for some members to learn more about it.
It would probably be good for 'posting' to start in a simple & basic way, then leading to more technical aspects.
I, myself look forward to some reading & thoughts. A question I have is:
Is DSP the future of audio alignment and 'perfection'?
"You and me both brother" - as Harry Bosch would say.
I've been using the excellent Najda 4 in, 8 out.
10 years have passed!
It still does the business and I'm still finding different ways to time align.
I've usually always aligned 1st peak positive with the mic at the listening position.
After a chat here, it was suggested to align the mouths, if not physically (impossible to do), then in DSP.
I did this and the difference in milliseconds is pretty small.
But you do hear the difference, quite significantly!
I'm evaluating that.
As I've evolved my system over the years, changing drivers, horn types etc , DSP X/O room correction on the tapped horn subs has been great.
I'm perhaps lucky but only need a few cuts in dB here and there below 100Hz to tame the room. 30Hz and small one at 85 or so.
Being able to trial X/O slopes and points so easily is essential in system dev, and when changes are made.. Even moving house and settling into a new room.
Can't imagine doing it without such a solution.
I like Najda so much I have a spare..
I've given it linear power, used analogue input with Turntable.
Coax spdif and I2S routes in with digital.
Very sad that the developer is no longer with us.
CamillaDSP looks interesting.
I'm following that thread.
Other suggestions were to do partial or full FIR in Jriver - I've not grasped that nettle so far.
I've not really looked at miniDSP.
I mostly stream Qobuz. Once in the digital domain, the tech evolution is limited only me tech, time and money.
I see DSP evolving, being more powerful and even an AI element. "GPT-4, make my hifi sound better"😂
I've been using the excellent Najda 4 in, 8 out.
10 years have passed!
It still does the business and I'm still finding different ways to time align.
I've usually always aligned 1st peak positive with the mic at the listening position.
After a chat here, it was suggested to align the mouths, if not physically (impossible to do), then in DSP.
I did this and the difference in milliseconds is pretty small.
But you do hear the difference, quite significantly!
I'm evaluating that.
As I've evolved my system over the years, changing drivers, horn types etc , DSP X/O room correction on the tapped horn subs has been great.
I'm perhaps lucky but only need a few cuts in dB here and there below 100Hz to tame the room. 30Hz and small one at 85 or so.
Being able to trial X/O slopes and points so easily is essential in system dev, and when changes are made.. Even moving house and settling into a new room.
Can't imagine doing it without such a solution.
I like Najda so much I have a spare..
I've given it linear power, used analogue input with Turntable.
Coax spdif and I2S routes in with digital.
Very sad that the developer is no longer with us.
CamillaDSP looks interesting.
I'm following that thread.
Other suggestions were to do partial or full FIR in Jriver - I've not grasped that nettle so far.
I've not really looked at miniDSP.
I mostly stream Qobuz. Once in the digital domain, the tech evolution is limited only me tech, time and money.
I see DSP evolving, being more powerful and even an AI element. "GPT-4, make my hifi sound better"😂
Last edited:
Hi,
Yeah its the future, and present
You can get amplifier channel per few dollars from china, or "reputable" modules like IcePower for some tens. I mean, multiway speaker needs the DSP and some amplifiers and its not that expensive anymore. I built 8 channels with cost around 1500e total including cabling and all. Thats 8 channels power amplification with full processing in 2 rack unit size less than 1€ per (optimistic)watt. Could be a lot cheaper, could be a lot more expensive, depends what you want and need.
Now, why DSP? It enables all sorts of things in speaker design, if you start designing a speaker system for your application you'll soon find out its all about compromises. If you grant yourself to assume that any crossover and processing is possible it gives a lot more of freedom on all kinds of compromises you can now explore. And today it really is, check out VituixCAD software. Its "mere trivial task" to make perfect crossover for any object you've built, there is no black magic to it with modern software tools and knowledge. Main task is to figure out what you want and what kind of physical structure delivers, with "ideal crossover and filters", which basically includes DSP and assumes VituixCAD. In the end its perfectly possible to end up with a system that works fine with passive crossover and you could sell the DSP stuff.
Even if you have speakers already, any speakers, you could still benefit from DSP to manage some bass peaks at listening position. For this reason DSP is a no brainer for any system. Still, physics are the same as before so you might want to have passive parts in mind for possibility to manipulate impedance with the drivers for distortion reduction, or tailoring system Q for example.
Only reason not to use DSP is to prefer less complexity, but trade-off is reduced ceiling for possibly achievable top audio quality Well, another reason not to use DSP is if you are more into nostalgia and aesthetics than "absolute audio quality". After all its a hobby and fun time so what ever rocks your boat. Mind you, the audio quality is not automatic with DSP but you have to do some work for it, like with any tools. For this reason, you probably get best results with what ever system that is interesting to you for what ever reason and keeps you motivated. DSP is like long ladders, you could climb higher for better scenery but there is no point if you are fine with the level you are already at.
Have fun!
Yeah its the future, and present
You can get amplifier channel per few dollars from china, or "reputable" modules like IcePower for some tens. I mean, multiway speaker needs the DSP and some amplifiers and its not that expensive anymore. I built 8 channels with cost around 1500e total including cabling and all. Thats 8 channels power amplification with full processing in 2 rack unit size less than 1€ per (optimistic)watt. Could be a lot cheaper, could be a lot more expensive, depends what you want and need.Now, why DSP? It enables all sorts of things in speaker design, if you start designing a speaker system for your application you'll soon find out its all about compromises. If you grant yourself to assume that any crossover and processing is possible it gives a lot more of freedom on all kinds of compromises you can now explore. And today it really is, check out VituixCAD software. Its "mere trivial task" to make perfect crossover for any object you've built, there is no black magic to it with modern software tools and knowledge. Main task is to figure out what you want and what kind of physical structure delivers, with "ideal crossover and filters", which basically includes DSP and assumes VituixCAD. In the end its perfectly possible to end up with a system that works fine with passive crossover and you could sell the DSP stuff.
Even if you have speakers already, any speakers, you could still benefit from DSP to manage some bass peaks at listening position. For this reason DSP is a no brainer for any system. Still, physics are the same as before so you might want to have passive parts in mind for possibility to manipulate impedance with the drivers for distortion reduction, or tailoring system Q for example.
Only reason not to use DSP is to prefer less complexity, but trade-off is reduced ceiling for possibly achievable top audio quality
Well, another reason not to use DSP is if you are more into nostalgia and aesthetics than "absolute audio quality". After all its a hobby and fun time so what ever rocks your boat. Mind you, the audio quality is not automatic with DSP but you have to do some work for it, like with any tools. For this reason, you probably get best results with what ever system that is interesting to you for what ever reason and keeps you motivated. DSP is like long ladders, you could climb higher for better scenery but there is no point if you are fine with the level you are already at.Have fun!
Last edited:
Not necessarily so for an analog phono system. Why make analog sound like digital if you don't have to?
A claim to the effect that DSP and class-D can produce 'absolute audio quality' is necessarily over-stated.
Last edited:
Yeah that is for everyone to decide, if it is important to have all analog chain then obviously DSP will not fit.
I hear no digital sound in DSP, never have, but certainly hear when frequency response is off so I'm not sure why people are afraid of it. I understand appeal for all analog, and it probably sounds best if that is the assumption.
My target is best possible sound, regardless of technology. I'd like the technology to disappear, same for speakers, I don't want to watch them, I just want to enjoy good sound It might be so that analog is better at the end, but currently with the little experience I have doesn't support it in my application at least, where I have to do room EQ due to practical stuff dominating positioning.
I hear no digital sound in DSP, never have, but certainly hear when frequency response is off so I'm not sure why people are afraid of it. I understand appeal for all analog, and it probably sounds best if that is the assumption.
My target is best possible sound, regardless of technology. I'd like the technology to disappear, same for speakers, I don't want to watch them, I just want to enjoy good sound
It might be so that analog is better at the end, but currently with the little experience I have doesn't support it in my application at least, where I have to do room EQ due to practical stuff dominating positioning.
Last edited:
To save on stylus (MC) and disc wearing, for practicity ( no need to clean stylus/disc at each face play), to 'solve' some artefacts ( clics and surface noise), etc,etc,...
People freak out about digital in analog chain (TT). I do limit AD/DA too in my analog chain but... vinyl released today are most probably mastered through DAW in digital before being sent to a DA driving cuting lathe... even from digital 16/44,1 master as original tapes are often lost for 'classic' recording!
Dsp open possibility. Hybrid system sounds good too.
Time alignement is easy... about it you can use 'wavelet method' to time and phase align. Works great:
https://www.diyaudio.com/community/...ind-delay-phase-and-polarity-at-xover.370287/
Last edited:
We know Mobile Fidelity has been using DSD256 for some years. I use DSD256 too. Also IIUC once mastered to vinyl, the vinyl still sounds 'better' than the DSD256 source.
OTOH, it is possible to hear the digital artifacts on the old Telarc vinyl records originally recorded to digital (PCM, again IIUC). Doubtful anyone here is using a DSD256 AD/DA system at home equivalent in quality to what Mobile Fidelity has. More likely there is a lot of consumer grade PCM.
Also, there is very little in the way of clicks and noise here. Before ever being played once, records get cleaned with a Nessie system using a custom process. After that static and dust pickup is not a problem.
Anyway, with optical phono, Marantz MA9-S2 mono blocks (Benchmark AHB2 also available), large panel electrostatic speakers (time-aligned by design), a treated room, and some custom analog electronics, we have never heard digital that sounds as good as good vinyl. The dacs here are pretty good too, better sounding than any Topping I have heard, and better sounding than the old Benchmark DAC-3 (as well as better sounding several others we have auditioned). So, I'm not convinced there is a net benefit to going digital for phono. Rather, the evidence here is more to the opposite.
OTOH, it is possible to hear the digital artifacts on the old Telarc vinyl records originally recorded to digital (PCM, again IIUC). Doubtful anyone here is using a DSD256 AD/DA system at home equivalent in quality to what Mobile Fidelity has. More likely there is a lot of consumer grade PCM.
Also, there is very little in the way of clicks and noise here. Before ever being played once, records get cleaned with a Nessie system using a custom process. After that static and dust pickup is not a problem.
Anyway, with optical phono, Marantz MA9-S2 mono blocks (Benchmark AHB2 also available), large panel electrostatic speakers (time-aligned by design), a treated room, and some custom analog electronics, we have never heard digital that sounds as good as good vinyl. The dacs here are pretty good too, better sounding than any Topping I have heard, and better sounding than the old Benchmark DAC-3 (as well as better sounding several others we have auditioned). So, I'm not convinced there is a net benefit to going digital for phono. Rather, the evidence here is more to the opposite.
Last edited:
Most DSP functions do not use a microphone. On the other hand, if you have so many different options only a DSP can give you, it is highly recommended to measure your system and room.
Calibrated microphones can be bought under 100 US$. Calibrated ones give you some certainty to have a device useable for measuring, if you know what you want, even an uncalibrated mike will do.
For basic measurements a USB microphone seems simpler. If you are an advanced user and into building speakers too, the combination of a “normal” microphone and an USB audio interface is more universal.
A measuring microphone is just as useful in a fully analog system as it is in a DSP'd one. There is no advantage in not knowing what the problems of your audio system may be.
Calibrated microphones can be bought under 100 US$. Calibrated ones give you some certainty to have a device useable for measuring, if you know what you want, even an uncalibrated mike will do.
For basic measurements a USB microphone seems simpler. If you are an advanced user and into building speakers too, the combination of a “normal” microphone and an USB audio interface is more universal.
A measuring microphone is just as useful in a fully analog system as it is in a DSP'd one. There is no advantage in not knowing what the problems of your audio system may be.
As the feature sizes of IC processes are reduced, digital circuits become faster and smaller, the costs of digital processing go down, and the IC processes get less and less suitable for analogue electronics, mainly because the supply voltages that the tiny transistors can handle also get smaller and smaller. So generally there is a trend to do more and more in digital.
If you go with the chips used for smartphones and MP3 players the cost of DSP is a few dollars total, although the spec's might be rather modest for ADCs and DACs (though DACs aren't used much these days in such small devices, since direct class D outputs avoid the analog domain). Its certainly the case that such miniaturized signal handling would be prohibitively large for a smartphone if done analog - the volume of one audio decoupling capacitor is more than the entire digital chain(!)
These days you can get DSP-capable microcontrollers with GFLOP performance if you want to have fun designing compact standalone hardware (for instance Teensy 4 which I've used). Couple with high-spec ADCs and DACs and you've got fantastic performance.
And course a good USB soundcard turns any computer into a powerful DSP platform.
DSP has been the professional way for >30 years, its definitely not the future, its the past and present. Once they figured out low-latency fast FFT convolution (non-uniform partitioned) it became possible to use 100's of thousands to millions of taps in a FIR filter economically for reverb effects, there was no need for any analog processing steps for professional/real-time audio any more!
These days you can get DSP-capable microcontrollers with GFLOP performance if you want to have fun designing compact standalone hardware (for instance Teensy 4 which I've used). Couple with high-spec ADCs and DACs and you've got fantastic performance.
And course a good USB soundcard turns any computer into a powerful DSP platform.
DSP has been the professional way for >30 years, its definitely not the future, its the past and present. Once they figured out low-latency fast FFT convolution (non-uniform partitioned) it became possible to use 100's of thousands to millions of taps in a FIR filter economically for reverb effects, there was no need for any analog processing steps for professional/real-time audio any more!
Although really a technicality, it is not the operating voltages that are getting smaller with size reduction -
it is the operating currents that are getting smaller. Smaller size AND heat.
Although I didn't really make it clear at the start -
I do already understand about 'sound effects' and DAW editing.
What I am really interested in is the role of DSP in 'audio system' analysis & tuning,
really with a microphone as a prerequisite for measuring.
What microphones are various people using - and what are their prices?
I do already understand about 'sound effects' and DAW editing.
What I am really interested in is the role of DSP in 'audio system' analysis & tuning,
really with a microphone as a prerequisite for measuring.
What microphones are various people using - and what are their prices?
Back in the old days, there were CMOS and BiCMOS processes with 600 nm minimum MOSFET channel length that could easily handle 5 V +/- 10 %. Nowadays the CMOS processes have core and I/O transistors, the core transistors are typically meant for voltages below 1 V and the I/O transistors for voltages below 2 V. You can use longer than minimum length transistors, but it doesn't help much, because the gate oxides are too thin to handle higher voltages. Low supply voltages don't leave much room for stacking things in an analogue design.
Microphone is mandatory ( but it's true for non dsp loudspeakers too), yes.
What i have used range from DPA/B&K, Earthworks ( if you are wealthy) to Behringer ECM8000, Dayton Emmc ( cheaper ones).
There is differences between this two extreme example but imho the cheaper solution are usable for home purpose as long as they are calibrated ( they don't withstand higher spl as well as the best and you woould not use them for serious recording duty (if you are into that) as an Earthwork or Dpa).
Just stay away from usb mic, they have limitation if you want to time align. I use a Dayton atm, costed me 100euros and it does the job.
Dsp in analysis won't help ( maybe in replacing a soundcards for some hardware units -loudspeaker management system but it would be a serious limitation compared to a 150euros 2 chanel soundcard- time alignement too here ).
In tuning it's so dependent of the unit ( if hardware) your question is too vague without being more accurate about a reference.
If using a computer/ soundcard even more vague as it can do almost everything you could think of... from eq/delay/ xover ( IIR or FIR)/ DRC/...
Here is an example of what could be achieved with 'recent' software:
https://audiophilestyle.com/ca/ca-a...nd-room-correction-software-walkthrough-r682/
I would not use a distortion generator( this is what exciter are and Aphex are exciters) plug in in my listening chain but ymmv.
Last edited:
I use a Behringer that has been calibrated, phantom power/amp and Holmimpulse PC SW.
I don't think I need more🙂
I measure but also listen and change PEQ values according to the bass effect I want.
X/O slopes I measure and listen my way forward.
Time alignment is all measurement.
I don't think I need more🙂
I measure but also listen and change PEQ values according to the bass effect I want.
X/O slopes I measure and listen my way forward.
Time alignment is all measurement.
Last edited: