Mini-DSP quality

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2014
Paid Member
I tend to side with Benchmark on this. It's very rare, only really happens with MP3 and 3.5dB of digital attenuation kills it if you are paranoid. It's clear it keeps you awake at night. It bothers me not a jot and seems most audiophiles don't spot it as it is so rare in actual playback.

https://benchmarkmedia.com/blogs/application_notes/13545433-audio-that-goes-to-11

I have no Apple products so not interest in 'mastered for itunes'.
 
Strawman arguments are pretty.....sad. I suggest you re-read my posts.
The only opinion I gave was to not listen if you don't like the sound. (An opinion I suspect you agree with.)

And, I simply remarked that many of the LX/Orion users are utilizing different solutions.....both ASP and DSP (and different brands/types of DSP as it happens.) I offered no opinion on any of those solutions.

Regardless, this seems to be a thread about the subjective sound "quality" of miniDSP (which ones I'm not sure of) units. Your usage of an analog-based unit....and advocating for such....is off-topic for this thread.

Dave.
Thanks Dave, this is exactly what I'm thinking about. Is the minidsp a good option or maybe ASP is a better option.....


Sent from my Moto G (4) using Tapatalk
 
Even solutions with overhead bits may have intersample distortion while resampling. We have seen it in a SHARK DSP.
So the proof is in the pudding. Run a signal with intersample values higher than 0dBFS through it and measure or listen. Often also artifacts is clearly audible.
The wave files in my linked thread produces clicks in my resampler (mac mini)
No click if same samplerate is used
 

TNT

Member
Joined 2003
Paid Member
Well it's not rearer than that they attunate 3.5 dB
In fact it happens on most CDs
SACD attunate 6 dB
As you say the solution is to attunate before the resampling. A dB is enough in most cases ref Apple.
But bitperfect is gone. (Who cares:))

You can do -6dB (/2) by a shift left. Then you preserve all MSBs but the LSB which you need to insert (as dither?). I would call this almost bit perfect and I would not loose any sleep of it.

//
 
Account Closed
Joined 2001
Thanks Dave, this is exactly what I'm thinking about. Is the minidsp a good option or maybe ASP is a better option.....

Sent from my Moto G (4) using Tapatalk

I'm sorry these other fellas have hi-jacked your thread, but I can't make a subjective evaluation for you. :)

I tend to think the best way to approach this type of premise is to configure the miniDSP unit (or whatever ADC/DSP/DAC DUT you have in mind) for no EQ/xover/etc, and insert it into a known system. You can then subjectively evaluate any inherent "sonic signature" the unit might be adding. Obviously, you should operate within a nominal signal level window so you avoid clipping and rise above the noise floor sufficiently.

Cheers,

Dave.
 
Last edited:
The problem with minidsp asrc is that it is done before one can attunate the signal in the unit. So connecting a CD player to spdif and resample to 96khz will distort way more than the -120 dB spec for +0dBFS signals.
Hence my love for analog in/out. Then the behaviour is according to spec and no artifacts except the on/off thumph
 
I'm sorry these other fellas have hi-jacked your thread, but I can't make a subjective evaluation for you. :)

I tend to think the best way to approach this type of premise is to configure the miniDSP unit (or whatever ADC/DSP/DAC DUT you have in mind) for no EQ/xover/etc, and insert it into a known system. You can then subjectively evaluate any inherent "sonic signature" the unit might be adding. Obviously, you should operate within a nominal signal level window so you avoid clipping and rise above the noise floor sufficiently.

Cheers,

Dave.
This is what started the thread. Post one..... I know there's a difference, just wanted to get others thoughts.

Sent from my Moto G (4) using Tapatalk
 
Even solutions with overhead bits may have intersample distortion while resampling. We have seen it in a SHARK DSP.
So the proof is in the pudding. Run a signal with intersample values higher than 0dBFS through it and measure or listen. Often also artifacts is clearly audible.
The wave files in my linked thread produces clicks in my resampler (mac mini)
No click if same samplerate is used


a audible click is evidence of broken resampler software math, only overflow/wrapping would give that, not just clipping Gibbs ringing overs

most modern DSP should have overflow saturation as a mode/setting

any resampler software plugin code should have the headroom or saturate properly


and you should be able to see/show us this in the digital output
 
Account Closed
Joined 2001
This is what started the thread. Post one..... I know there's a difference, just wanted to get others thoughts.

Sent from my Moto G (4) using Tapatalk

Yours is the only opinion that matters. (Others thoughts are irrelevant.)
However, if you feel the unit was not working properly and/or you weren't operating it correctly, then maybe there's some discussion to be had.

Cheers,

Dave.
 
I think this is relevant since a lot of MINIDSP products uses SHARK DSP with the inherent DSP samplerate conversion. I have discussed this with another manufacturer that uses SHARK DSP:
http://www.diyaudio.com/forums/digi...amp-crossover-dac-project-43.html#post4487781
He decided to attunate the signal before the SHARK DSP.

The reason we dont hear the artifacts i everyday use is that the mastering machines and mastering engineers are clever and mask it. (So please don't loose any sleep)
 
Member
Joined 2002
Paid Member
The reason we dont hear the artifacts i everyday use is that the mastering machines and mastering engineers are clever and mask it. (So please don't loose any sleep)

I was curious to see if the “intersample peaks” is an issue with properly recorded real music when down sampled or up sampled.
I used a high resolution (352.8KHz/24bit) download which has no flat peaks at –0.82dB FS (L) and –0.62dB FS (R).This recording I down-converted up to 44.1KHz/16bits.
http://www.diyaudio.com/forums/everything-else/169484-what-wrong-op-amps-476.html#post4955946

Then I used the low down-converted file and up-converted it.
Conversions were done in Steinberg Wavelab
In both down and up conversions, there isn’t any significant peak change, only at the second decimal point of a dB.

George
 

Attachments

  • conversion peak monitor.PNG
    conversion peak monitor.PNG
    23 KB · Views: 490
Yes it it only achiveable when you manipulate in the digital domain.
The digital mastering machines push the samples towards 1 to make i loud.

So thats why analog in protects against it, even with samplerate conversion inside.
Provided you don't use digital amplification before the ASRC, of course.
(You can still ofcourse have clipping on the input with analog in)
 
I was curious to see if the “intersample peaks” is an issue with properly recorded real music when down sampled or up sampled.
I used a high resolution (352.8KHz/24bit) download which has no flat peaks at –0.82dB FS (L) and –0.62dB FS (R).This recording I down-converted up to 44.1KHz/16bits.
http://www.diyaudio.com/forums/everything-else/169484-what-wrong-op-amps-476.html#post4955946

Then I used the low down-converted file and up-converted it.
Conversions were done in Steinberg Wavelab
In both down and up conversions, there isn’t any significant peak change, only at the second decimal point of a dB.

George

1 Most mastering software (all that I know of) can be set to inter sample peak limiting. So clipping should not be a problem. Of cause if you then use perceptual coding, you change the waveshape and clipping can occur.

2 If you design the analog part of the converter to have a bit of headroom, ISP clipping doesn't need to be a problem. In any case a tiny bit of clipping is not audible under normal conditions.
 
Account Closed
Joined 2001
I think this is relevant since a lot of MINIDSP products uses SHARK DSP with the inherent DSP samplerate conversion.

Most of them don't use a SHARC DSP.....and even the ones that don't still use sample-rate conversion.

You've already linked to your thread on this particular topic where you advocate your point of view.
Why do you continue to threadjack this topic??

Dave.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.