Member
Joined 2009
I need some clarity on the topic of digital audio resampling.
I have my mind set on active multi-way speakers with a DSP crossover. I will use a computer as the source. What will give me better sound quality?
A: PC -> low-jitter USB -> DAC -> analog -> DSP
B: PC -> low-jitter USB -> digital -> DSP
At first glance option B looks great - keep the signal path in lossless digital. The issue (or non-issue) is that all my music is in 44.1kHz and DSPs tend to run at 48 or 96kHz. DSPs with digital audio inputs must digitally resample the signal to their internal clock rate. Resampling to a rate that is not an integer multiple of the original always introduces some error.
With option A there will inevitably be some signal degradation done by the DAC, the wire and the subsequent Analog to Digital Conversion. The DSP however does the ADC and processing on it's own clock so there will be no timing issues and those tend to be more offending.
The question is which option is more likely to introduce audible problems? I should probably audition both and pick one but I was wondering what you guys have to say about this.
Are there DSP solutions that adapt their clock to the incoming digital signal?
I have my mind set on active multi-way speakers with a DSP crossover. I will use a computer as the source. What will give me better sound quality?
A: PC -> low-jitter USB -> DAC -> analog -> DSP
B: PC -> low-jitter USB -> digital -> DSP
At first glance option B looks great - keep the signal path in lossless digital. The issue (or non-issue) is that all my music is in 44.1kHz and DSPs tend to run at 48 or 96kHz. DSPs with digital audio inputs must digitally resample the signal to their internal clock rate. Resampling to a rate that is not an integer multiple of the original always introduces some error.
With option A there will inevitably be some signal degradation done by the DAC, the wire and the subsequent Analog to Digital Conversion. The DSP however does the ADC and processing on it's own clock so there will be no timing issues and those tend to be more offending.
The question is which option is more likely to introduce audible problems? I should probably audition both and pick one but I was wondering what you guys have to say about this.
Are there DSP solutions that adapt their clock to the incoming digital signal?
Last edited:
It doesn't answer your question, but if the computer is your only signal source - why not use the computer as your dsp?
Less D/A the better!?
What platform are you planning on using, I'm using miniDSP,
Mac - Digit FP - MiniDSP 2x8 -Amp.
I plan to add some CurryMan DACS for each channel used,
What platform are you planning on using, I'm using miniDSP,
Mac - Digit FP - MiniDSP 2x8 -Amp.
I plan to add some CurryMan DACS for each channel used,
Member
Joined 2009
Mainly because I fear something going wrong with the PC and sending low frequencies into the tweeters. I imagine standalone DSP solutions would offer better protection.
I'm open to the idea of a small, fanless computer dedicated exclusively to music playback. But a PC solution still implies either software resampling (which could potentially be of acceptable quality) or a software DSP solution that detects and adapts to the source sampling rate.
I'm open to the idea of a small, fanless computer dedicated exclusively to music playback. But a PC solution still implies either software resampling (which could potentially be of acceptable quality) or a software DSP solution that detects and adapts to the source sampling rate.
Member
Joined 2009
I currently use miniDSP 2x4 (analog) but I ultimately need 8 channels. Honestly, I'm happy with it so far but I would like to explore my options. I would probably go with a miniDSP 4x10 HD, miniDIGI 2x8 (digital), Najda DSP or maybe a Lunix PC based DSP.
Keep your pc on a ups, it is very simple solution.....or
Checkout the www.bodziosoftware.com.au/ there is an excellent consortium between Delta/Lynx audio, Minidsp and Bodzio. I think this solution is quite elegant and all you need is a wireless USB volume control and sit back and enjoy the music.
The mini dsp amp packs look very nice at $275 and working with the Bodzio Ultimate Equalizer the signal is digital until the ICE amplifiers. If the computer crashes the digital signal connection will not take out your speakers
I've have designed a 3 way system with a passive crossover between the tweeter and a dome mid, if you select your drivers correctly this is a simple and ez crossover to implement..... the Ultimate Equalizer can do its job and have a powerful crossover where it really matters.
Member
Joined 2009
The Bodzio software does look like a nice package and I don't doubt that it provides great sound quality.
But the point I'm making is that DSPs usually resample the original data which degrades the sound quality to some degree. Granted there are very good software resampling algorithms out there that a PC should handle without problem. An interesting experiment would be to record the same analog material once at 44.1kHz and then at 96kHz. Listen to both files through your Bodzio crossover and see if you can detect a difference. They should sound identical.
The miniAMP kits are digital until the amp output. The plate amps with the ICE Power modules convert the SPDIF to analog before amplifying it.
But the point I'm making is that DSPs usually resample the original data which degrades the sound quality to some degree. Granted there are very good software resampling algorithms out there that a PC should handle without problem. An interesting experiment would be to record the same analog material once at 44.1kHz and then at 96kHz. Listen to both files through your Bodzio crossover and see if you can detect a difference. They should sound identical.
The miniAMP kits are digital until the amp output. The plate amps with the ICE Power modules convert the SPDIF to analog before amplifying it.
I use an all PC software solution DSP for digital XO, driver time alignment/linearization and room correction. There is no resampling and the music playback software hosting the digital filters automatically switches the filters based on source sampling rate.
I wrote a long and boring article on it here: Computer Audiophile - Advanced Acourate Digital XO Time Alignment Driver Linearization Walkthrough Maybe you will find it meets your requirements...
Uh, no. MiniDSP typically resamples to 48k in their implementations but any DSC or DSP will work on pretty much any sample rate you might want. That also holds for DAWs and VST hosts in PC implementations. It's a question of whether the convenience of a MiniDSP is worth the time domain degradation from the ASRC applied. In my experience resampling 44.1 to 48 is pretty horrid but I'd suggest ABXing with SoX to make your own assessment.
A good hardware ASRC with the proper polyphase filtering will be better than doing DAC to ADC. Best if there's a reasonable upsampling ratio (e.g. 44.1 to 96) though.
Better yet, clocking an ASRC off your master clock that's running your DSP and DACs will eat any remaining jitter coming out of the USB or SPDIF input.
Better yet, clocking an ASRC off your master clock that's running your DSP and DACs will eat any remaining jitter coming out of the USB or SPDIF input.
Mmm, no and no, most likely. All hardware ASRCs I'm aware of are linear phase brick wall, so DAC -> ADC can fairly easily have less impulse response ringing that's more masked psychoacoustically if minimum phase, slow antialiasing rolloff signal conversion's chosen. For asynchronous USB audio or ASIO the clock is local to the audio hardware, so there's no "jitter" due to sample stuffing or need for ASRC as only one clock domain is present.
In the SPDIF case, yes, clock domain transition is required. The elastic buffer in Wolfson's receivers offers something of an optimal tradeoff between source and local clock tracking. In most cases an ASRC does more violence to the audio data, though a exceptions exist. So it depends somewhat on specifics not specified here. 😉
In the SPDIF case, yes, clock domain transition is required. The elastic buffer in Wolfson's receivers offers something of an optimal tradeoff between source and local clock tracking. In most cases an ASRC does more violence to the audio data, though a exceptions exist. So it depends somewhat on specifics not specified here. 😉
Yes exactly forgive my grammar.
The ICE amplifiers are analog so the mini dsp has to do the DA before running to the ICE amps.
"Re-sampling.
In order to keep the whole system in time alignment, there must be a single clock source in the system. So, if I play .wav file at 44.1kHz, I could simply use 44.1kHz as the clock frequency – without re-sampling, if this is your preference. Then the whole playback chain works with the same sampling frequency.
Having googled the internet, I got the impression, that jitter claims are exaggerated, and contemporary hardware uses PLL clock locking/filtering techniques, which reduce jitter by 50-100 times. I am not an expert here, so I could be wrong."
That is a quote from another thread from Bohdon of Bodzio Software
http://www.diyaudio.com/forums/vendors-bazaar/240273-ultimate-equalized-v5-has-been-released.html
Not to sell in the forum but I am selling my 5 leftover Digmoda DP-130 crossovers in deference to the PC solution. I have them listed in the swap section. They may suit your purposes better, I built a 7.1 system and I liked them very much....
Member
Joined 2009
I'm probably obsessing over this more than it's worth. Admittedly, I have very limited experience and as I stated earlier the best approach would be for me to audition both options and pick one.
I base my view on the fact that I can't reliably detect if the miniDSP2x4(analog in) is a part of my signal chain or not. Whatever signal degradation the ADC and DAC on the miniDSP2x4 are doing sounds transparent to me (I'm using a linear regulated power supply).
However Windows 7 software resampling is always audible when compared to bit-perfect playback. I'm curious to audition other software resampling algorithms as I'm sure there are better ones out there.
The argument often given by hardware designers is that we shouldn't worry about the sound quality of the resampling hardware because it's THD measures lower than that of the DAC portion. But I suspect that resampling errors could be more offensive than errors introduced by the digital/analog conversion even if they measure more favorably. In essence the digital-only path with DSP will run on 2 clocks and the points of digital/analog conversion eliminate the clock problem.
A DSP that syncs it's clock rate to the digital input signal is obviously the ideal solution.
I appreciate the more informed opinions on the matter.
I base my view on the fact that I can't reliably detect if the miniDSP2x4(analog in) is a part of my signal chain or not. Whatever signal degradation the ADC and DAC on the miniDSP2x4 are doing sounds transparent to me (I'm using a linear regulated power supply).
However Windows 7 software resampling is always audible when compared to bit-perfect playback. I'm curious to audition other software resampling algorithms as I'm sure there are better ones out there.
The argument often given by hardware designers is that we shouldn't worry about the sound quality of the resampling hardware because it's THD measures lower than that of the DAC portion. But I suspect that resampling errors could be more offensive than errors introduced by the digital/analog conversion even if they measure more favorably. In essence the digital-only path with DSP will run on 2 clocks and the points of digital/analog conversion eliminate the clock problem.
A DSP that syncs it's clock rate to the digital input signal is obviously the ideal solution.
I appreciate the more informed opinions on the matter.
If it hadn't been for other limiting factors (loopback/input limitations) I would have continued using a PC as DSP. I've yet to find proprietary DSP solutions offering the same amount of power and flexibility with regards to signal processing. Consider Acourate in combination with JRiver as a strong candidate.
JRiver have the ability of leaving the playback at original sample rate, but off course, the internal Maths is executed at a (much) higher rate and bit Depth.
I've never experienced that the computer software goes bananas and unintentionally delivers full bandwidth / full power to the tweeters - other than for user errors.
I have however experienced incidents with an 8ch USB DAC With digital volume Control where the volume controller suddenly lived its own life turning the speaker volume to full blast. Also I have experienced nasty incidents using standalone DSP platform with filter preset options, where the empty presets uses full bandwidt as default - not very funny when you push the wrong button on the remote.
So indenpendently of computer or DSP, my safest option is to av a high quality manual analog volume controller to fiddle with, at least for the setup/tuning purpose.
The hobby is obsessive but you're right about having to experience them.
I have been building DSP speaker systems for more than 15 years, and you have to prioritize what is most important and avoid trying to satisfy every fad, fetish and issue. And make a list of your own priorities.
All of these dsp's can provide great sound, but not all are equal under certain conditions. What are your conditions? what amps are you going to use, what drivers, etc, etc,etc,
I can tell you for me getting away from the extra cables and boxes was a big deal and I believe improves total system performance. I've build 15-20 projects with outboard speaker controllers using like Behringer, DBX, BSS/London, XTA, and Lake tech....but the digmoda/B&O ICE system I like very much for the fact I could move speakers easily and put them in other systems, so for me integration was critical, the computer is a bit of step backwards but the new system will have a permanent home I am not trying to sell this system, like the last one.
This is a rhetorical question....but how good are you at building speakers? Are all these tiny defects you're trying to eliminate really going to matter for your project or are you simply achieving paralysis through analysis?
I would think about the form you want your speakers to take and that will narrow your possibilities which I think will be helpful. If only to provide provide a more focused question to help you address these smaller issues.
When you choose a technology or form just follow the rules the system requires to function and not ask it to do more.
BTW I have not purchased the Bodzio Software yet but functionally this suits my project the best.
Hope that was helpful
Member
Joined 2009
I'm using the Jameco wallwart power supply that was suggested in this thread.
DDU120050F0981: JAMECO VALUEPRO: Power Supplies & Wall Adapters
I'm not sure if it's optimal but like I mentioned it sounds good enough to me. I guess you can't use that in the UK. If you are currious what the best possible sound quality with the mniDSP2x4 is you can try a 9V battery.
DDU120050F0981: JAMECO VALUEPRO: Power Supplies & Wall Adapters
I'm not sure if it's optimal but like I mentioned it sounds good enough to me. I guess you can't use that in the UK. If you are currious what the best possible sound quality with the mniDSP2x4 is you can try a 9V battery.
- Status
- Not open for further replies.