Depending on the filter, this might result in clipping (with multiplicator 1) unless you have the reduced multiplicator in FIR2 or use the digital volume control.
But e.g. with the NQ filters it is safe as the highest coeficient was 1/8
The Izotope settings were:
Steepness: 7
Filter Max Length: 1,500,000 samples
Cutoff: 1.04 x Nyquist
Anti-aliasing: 200
Preringing: 0.65
No longer posting any details of new filters as a direct result of the carry on over in the main thread. You give the key perps a pat on the back for the great efforts. I was very, very close to not posting anything new at all.
Thank you very much for . I understand that F2 is 8 instead of 6.8333.
The new filters have less distortion than previous. In armonic distortion C128 Better than C960. Both saturated slightly in treble (0 dB) Is -80 db distorsion in trebles only in 0 dB and perfect -1 dB. . I prefer F2 filter 8 like 6.833.
The other measures is better C960 filter: in phase, amplitude, impulse, this filter is technically perfect in every way.
I put measures -1 db. Red C128 Blue C960
In sound are an improvement, more information. I listening clear difference in piano sound. The new c128 gives feeling of being heavier piano than the previous filter.
Both are nice and not bother especially C960. I had little time, it was too late and could not put volume necessary.
I regret and understand your decision. In my opinion to have common reference: Bal output. Independent of the impedance of the following equipment, available at the DAM no need to add nothing. Is important to have common opinions.
Anyway anything you post will be well received. And if you finish to publish, I appreciate what you has been done. Not only the filters. I learned a lot.
The new filters have less distortion than previous. In armonic distortion C128 Better than C960. Both saturated slightly in treble (0 dB) Is -80 db distorsion in trebles only in 0 dB and perfect -1 dB. . I prefer F2 filter 8 like 6.833.
The other measures is better C960 filter: in phase, amplitude, impulse, this filter is technically perfect in every way.
I put measures -1 db. Red C128 Blue C960
In sound are an improvement, more information. I listening clear difference in piano sound. The new c128 gives feeling of being heavier piano than the previous filter.
Both are nice and not bother especially C960. I had little time, it was too late and could not put volume necessary.
I regret and understand your decision. In my opinion to have common reference: Bal output. Independent of the impedance of the following equipment, available at the DAM no need to add nothing. Is important to have common opinions.
Anyway anything you post will be well received. And if you finish to publish, I appreciate what you has been done. Not only the filters. I learned a lot.
Paul, please understand you inspire more people here than the few members needing to voice their lack of understanding how important actually knowing the sound in your own test rig is.
It’s like those people talking about “I like to come as close to the sound they mixed in the studio”. Failing to understanding some of the hardest work done in the studio was to make it sound good on low quality speakers like NS10 and Auratone for the average listener to enjoy the end result on their crappy DAB radios. On a side-note, that’s the most valid reason for hi res files as I see it. If they all had new masters optimized for high quality reproduction gear rather than today’s radio masters.
Most of us don’t have the time and skills to keep up the pace you have moved the sound quality of this DAC forward. I find myself jumping pages just to keep up. Sadly without details on new filter there isn’t much for us to discuss and learn from other than for people to roll they’re dice and rate and judge new filters.
I hope you find it in yourself to forgive the poor souls giving you a hard time back then and remember your ‘fans’ are still here to droll all over your amazing work.
It’s like those people talking about “I like to come as close to the sound they mixed in the studio”. Failing to understanding some of the hardest work done in the studio was to make it sound good on low quality speakers like NS10 and Auratone for the average listener to enjoy the end result on their crappy DAB radios. On a side-note, that’s the most valid reason for hi res files as I see it. If they all had new masters optimized for high quality reproduction gear rather than today’s radio masters.
Most of us don’t have the time and skills to keep up the pace you have moved the sound quality of this DAC forward. I find myself jumping pages just to keep up. Sadly without details on new filter there isn’t much for us to discuss and learn from other than for people to roll they’re dice and rate and judge new filters.
I hope you find it in yourself to forgive the poor souls giving you a hard time back then and remember your ‘fans’ are still here to droll all over your amazing work.
Hi Paul, I too had reason to voice 'my opinion' on said thread. My decision was easy, just leave the critics behind and stay here where people want to improve the listening experience, learn about filter design, and care not about personal preference or which output to use. Hifi is subjective! Friends ask me what to buy? Simple answer, whatever appeals to you. Please do keep the filters coming and sharing your enthusisum for making 'our dac' one of the very best out there. Thanks for your thoughts, much appreciated!!!
I put measures -1 db. Red C128 Blue C960
Hi oneoclock. There is a problem with the images you uploaded. Can you please fix it? Thank you
@Paul, I think everybody here appreciates your work with the filters, for three different reasons: a) for marginally improving the sound of the dac b) for documenting your quest and thoughts in the filter world and the discussions with other members and c) for providing the "source code" of the filters. I really wish that you continue your work on this track if possible.
Furthermore, I would suggest doing the opposite of silencing; probably you can provide the doc that you mentioned in an old post of yours, where you documented your steps and efforts. This doc could be a perfect guide for a nice HOWTO on filtering and could probably engage other people on this track create an organized and crowdsourced library of filters, each filter with its documentation, reviews, etc. At least I would be interested on this aspect.
Thank you very much for . I understand that F2 is 8 instead of 6.8333.
The new filters have less distortion than previous. In armonic distortion C128 Better than C960. Both saturated slightly in treble (0 dB) Is -80 db distorsion in trebles only in 0 dB and perfect -1 dB. . I prefer F2 filter 8 like 6.833.
The other measures is better C960 filter: in phase, amplitude, impulse, this filter is technically perfect in every way.
I put measures -1 db. Red C128 Blue C960
In sound are an improvement, more information. I listening clear difference in piano sound. The new c128 gives feeling of being heavier piano than the previous filter.
Both are nice and not bother especially C960. I had little time, it was too late and could not put volume necessary.
I regret and understand your decision. In my opinion to have common reference: Bal output. Independent of the impedance of the following equipment, available at the DAM no need to add nothing. Is important to have common opinions.
Anyway anything you post will be well received. And if you finish to publish, I appreciate what you has been done. Not only the filters. I learned a lot.
+1. Thanks much Paul
Hi Paul,
Just want to express my appreciation for all the wonderful work you have done and all the time you spent that benefited so many of us.
I think it is pretty safe to say that one can never make everyone happy no matter what he does. So it is probably better to take a stand and simply ignore those you think are unreasonable.
Anyway, thanks again!
Just want to express my appreciation for all the wonderful work you have done and all the time you spent that benefited so many of us.
I think it is pretty safe to say that one can never make everyone happy no matter what he does. So it is probably better to take a stand and simply ignore those you think are unreasonable.
Anyway, thanks again!
Old c128-100 green. New c128 red (1021filtNQ_C128_100bp.skr).
Measure 0 dB output.
That really shows why the filters were sounding "off" in the mid range and HF.
Much of that is purely the nature of the R2R architecture. MSB and TotalDAC display the same characteristics.
That said the filters will contribute to the distortion characteristics but won't effect the harmonics of the main test tone.
What oneoclock's plots don't show is the peak in distortion around 11-12kHz that all the filters except NOS display.
Attached shows:
NQB3:red, C960_140dp:green, C128_100dp:blue, Izotope steep rolloff defaults: orange, stock 44.1: purple
Attachments
Last edited:
This is analysis of the fir coefficent behaviour, not DAC behaviour so don't jump to conclusions about what is going on.
If Izotope is displaying this behaviour, and is regarded as "the best" it's either not especially unusual, it's an artefact of SoX/Izotope upsampling or fuzzmeasure is telling porkies.
The peak of the Izotope trace appears to be very close to 11025, so fs/4 of the original impulse file.
Last edited: