Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1
  • 2

TOPIC: Re: Feature requests and suggestions

Feature requests and suggestions 7 years 5 months ago #8155

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
This topic is meant to be a mini "suggestion box" for rePhase.

If you feel your suggestion will need discussion, then please open a new topic in this section (after all we now have an entire dedicated section here :silly: )

Feel free to express any idea you might have!
https//wavetracing.com | rephase.org
The administrator has disabled public write access.
The following user(s) said Thank You: john.reekie

Re: Feature requests and suggestions 7 years 5 months ago #8156

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Here is a small todo list of features that will already hopefully make it in a coming rePhase release (as of rePhase 0.9.3) :
  • Measurement import, to be able to see the effect of a correction on a real measurement in realtime, instead of having to use HOLM C=A*B manipulation
  • Range selection in graph for frequency and phase
  • Biquad section, to import biquads from ACD or REW, in minimal or linear phase
  • Math section, to enter equations for amplitude or phase corrections and filters
  • Multiple banks in paragraphic EQ sections, to get rid of the existing "EQ Bank" sections
  • Make rePhase remember its settings and corrections from one run to another (like HOLM does)
  • Curves memory, to be able to "freeze" a curve and use it as a guide for subsequent corrections
  • Customizable optimization (a lot to be said on that subject...)
https//wavetracing.com | rephase.org
Last Edit: 7 years 5 months ago by pos.
The administrator has disabled public write access.

Re: Feature requests and suggestions 7 years 5 months ago #8187

  • Richard
  • Richard's Avatar
  • Offline
  • Expert Boarder
  • Posts: 132
  • Thank you received: 71
  • Karma: 13
Hi Pos,

Rephase is already excellent(!) but these are my suggestions for useful features....

- Default view is "Large"

- Double-click on sliders to reset to 0dB (outputs, paragraphic EQ, etc.)

- Filters Linearization, currently only works for textbook Linkwitz-Riley type crossover (HPF and LPF slopes both at same freq) but would be nice to be able to handle overlapping filters where the HPF and LPF points are at different frequencies, and also ability to choose other targets shapes than Linkwitz-Riley, such as Butterworth and Bessel. So ideally, allow settings for the LPF (type, slope, freq) and the HPF (type, slope, freq) separately to allow all possibilities. Maybe a tick-box to automatically "pair" settings of LPF to corresponding HPF (greyed out), for textbook crossover alignments, but untick the box for independent LPF and HPF settings.

- Linear filters, other shapes in preset list. eg. Butterworth, Bessel. I know you could produce these with existing control over Q, but easier if a named preset exists.

- Linear filters, add option to also produce standard IIR filters ie. WITH their normal phase shift.

- Import user measurement and display it on same graph, but in different colour, eg. green.

- Option to show group delay plot curve on graph, to see actual delay (in milliseconds) over the frequency range. Should be easy to calculate mathmatically from the phase curve.
Last Edit: 7 years 5 months ago by Richard.
The administrator has disabled public write access.
The following user(s) said Thank You: pos

Re: Feature requests and suggestions 7 years 5 months ago #8268

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hi Richard
- Default view is "Large"
Some screen resolutions will not be large enough to support the "Large" view (small notebooks and the like). Some do not even support the "Normal" mode (hence the "compact one).
Automatically choosing the largest possible view based on screen resolution upon start would be a solution, but would go against the principle of least surprise...

In a next version I will try to make rePhase remember its settings from one run to another, and that should solve the problem: if you prefer the "Large" view then set it the first time you use the software, and it should remain identical upon runs.
- Double-click on sliders to reset to 0dB (outputs, paragraphic EQ, etc.)
good idea!
- Filters Linearization, currently only works for textbook Linkwitz-Riley type crossover (HPF and LPF slopes both at same freq) but would be nice to be able to handle overlapping filters where the HPF and LPF points are at different frequencies, and also ability to choose other targets shapes than Linkwitz-Riley, such as Butterworth and Bessel. So ideally, allow settings for the LPF (type, slope, freq) and the HPF (type, slope, freq) separately to allow all possibilities. Maybe a tick-box to automatically "pair" settings of LPF to corresponding HPF (greyed out), for textbook crossover alignments, but untick the box for independent LPF and HPF settings.

- Linear filters, other shapes in preset list. eg. Butterworth, Bessel. I know you could produce these with existing control over Q, but easier if a named preset exists.
I will open a separate topic to explain my view on the subject, and why I do not plan to implement these (although it would be quite straightforward to do...)
- Linear filters, add option to also produce standard IIR filters ie. WITH their normal phase shift.
I will try to implement that in a next version, although that will only be possible with LR and simple 1st/2nd order filters. Simulating the phase shift for a LR of arbitrary slope requires some special tricks though...
- Import user measurement and display it on same graph, but in different colour, eg. green.

- Option to show group delay plot curve on graph, to see actual delay (in milliseconds) over the frequency range. Should be easy to calculate mathmatically from the phase curve.
All good ideas!

Thanks for your suggestions!
https//wavetracing.com | rephase.org
The administrator has disabled public write access.
The following user(s) said Thank You: Richard

Feature requests and suggestions 7 years 2 months ago #9321

Hi everybody,

I would enjoy to have be able to do :

1 - A custom windowing (start/end index, window type, etc.) on a measurement impulse
2- Be able to do a mathematical A/B to be able to get the measurement inversion, with compensation of a measuring filter protection (A = filter measure, B = measure ==> C = measure windowed inverted and compensated with measurement filter).
3- Be able to get, using a custom target impulse, the convolution of the previous impulse with the target impulse response.

I would be eventually greatly interesting in participating to software development to implement these things if needed (even if my DSP knowledge is quite bad ^^, the multiplication and division in frequency domain seems easy to do....).

I actulally made all thoses steps with HolmImpulse to correct my system and I have a great correction, but HolmImpulses often crash and is quite annoying to do thoses simples things.... You can watch my video if you understand "french" ;) :
Last Edit: 7 years 2 months ago by LeChacal619.
The administrator has disabled public write access.

Feature requests and suggestions 7 years 2 months ago #9362

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hi LeChacal619

Nice method you are showing here :)
Such a feature would indeed be possible to integrate in rephase, but I would like it to stay in the "manual" way of doing correction, based on several measurements ans careful corrections.
Doing measurement inversion for automated correction is an attractive solution solution indeed, but there is a lot of pitfalls along the way as it heavily rely on a (single) measurement. Even if you use windowing you will still integrate lots of specially local artifacts such as diffraction, and in the other hand you will likely loose too much precision in the low to do any useful correction under a few hundreds Hz.
I don't say it cannot be done successfully, but it requires a lot of care and in the long run I find manual corrections to me easier, safer, and more reliable.
A careful and reliable automated correction will generally result of lowish Q corrections anyway, so it will be quite easy to operate such corrections manually.
Aggressive corrections that give an artificially flat result at one point are of course not a viable solution anyway, as any microphone move will add its own high Q differences to those of the correction...

Still, if you want to do automated correction may I suggest you to try DRC-FIR, which uses a lot of special tricks to circumvent some of the problems mentioned above (frequency dependent windowing for example) and is also able to integrate custom target curves I think (not sure about that one, but you can still generate them using rephase and use SoX to convolve different corrections together). PORC also springs to mind as an elegant solution, based on Bank's method.
Better still: you can use Align2 as an interface to both, and the ability to average several measurement for a more reliable correction:
www.minidsp.com/forum/python-open-room-c...ction/8674-re-align2
https//wavetracing.com | rephase.org
The administrator has disabled public write access.

Feature requests and suggestions 7 years 2 months ago #9382

Hi Pos,

I used DRC first 2 or 3 years ago to correct a full range speaker. I didn't got any good results (but my knowledge about windowing etc. was really bad so I made no one custom configuration : just used the moderate correction). I measured and corrected for sweetspot and as I said for full range only..... DRC correction resulted for me in very audible artifacts for transients sounds....

What I'm currently doing is quite different : I measure and correct each way by way in close range recordings (20cm or less). I choosed speakers that have a quite flat response, so the windowing size for lowest frequency of each way is greatly sufficient but provides problems effectively for low frequencies (even if I think for a close range measurement the reflections are quite negligibles...). Furthermore, DRC needed I think the measurement wave, the recording sweep etc., and I think it's really easier to just give the measured impulse direclty, but perhaps it's possible to do, I should read more about DRC another time I guess !

I chosed before reading your post to make my own program in C#, but perhaps a solution would be to make just a GUI for DRC, as Align2 ? I don't really know....
The administrator has disabled public write access.

Feature requests and suggestions 7 years 1 week ago #9787

  • john.reekie
  • john.reekie's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 3772
  • Thank you received: 1584
  • Karma: 141
Hi Thomas, thanks for the great work on rePhase. Couple of suggestions:

1. Add shelving filters to the minimum phase gain EQ. (Would actually be helpful for phase EQ as well, although that is a minor thought)
2. Allow arbitrary (or more flexible) vertical limits on the graph display. For example when doing amplitude EQ, the lower limit is 0 dB, but you want to "zoom in" on the display a bit more. The workaround I used is to rescale the plot in REW first before importing, so it's not the end of the world just a small suggestion :)

Thanks again!
I am not miniDSP support.

"You must ask the right questions." - Dr. Alfred Lanning's hologram.
-> Have you read the User Manual??
-> Have you drawn and posted a diagram?
-> Have you posted a screenshot?
-> Have you posted your config file?
The administrator has disabled public write access.

rePhase: feature requests 7 years 1 week ago #9821

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hi John

Thanks for your kind words.

Regarding Shelving EQs, that would indeed be interesting, and several people already expressed some interest for such a feature.
One problem is that Shelving definitions are even less standardized than EQ, with different interpretation of the positioning of the corner frequency...
Will have to see if I add this in the (minimum-phase) filter or in the paragraphic EQ tab...

Regarding arbitrary amplitude ranges, this is a good idea and should get implemented in the next release. There will still be a drop menu with common values, but the user will be able to edit the input manually when clicking on the field itself. This is already implemented in the current development version for the sampling rate (to implement downsampling to 3kHz for example), and will be generalized to the frequency , phase and amplitude ranges as per your proposal!

As of now, instead of normalizing before exporting, what you can use the "gain offset" setting in the "Measurement" tab to adjust the level when importing your measurement.

Thank you for your suggestions!
https//wavetracing.com | rephase.org
The administrator has disabled public write access.

rePhase: feature requests 7 years 2 days ago #9903

  • devteam
  • devteam's Avatar
  • Online
  • Administrator
  • Posts: 8750
  • Thank you received: 1125
  • Karma: 162
@ Pos,

Thanks again for your great comments. Did you notice that a new application note that John put together for rePhase?
www.minidsp.com/applications/advanced-tools/rephase-fir-tool

He did spend quite a lot of time to put together your great work a bit more info for the community. Send us your comments if any!

DevTeam
miniDSP, building a DSP community one board at a time.
The administrator has disabled public write access.

rePhase: feature requests 6 years 11 months ago #9971

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hi Devteam

Wow! This is a very comprehensive and easy to follow tutorial on how to use rePhase!
Kudos to John (john.reekie?) and devteam for putting this together.

Now I don't feel the urge to write a user manual for rePhase anymore :P

A couple of notes:
- When doing phase correction only, or amplitude correction less than a few dB, rectangular windowing (ie no windowing algorithm per se) is often the best option.
- When using multiple impulses on different drivers (either in a stereo correction setup or a full multiway crossover), it is indeed a good idea to keep the centering on "middle" to keep things simple (but then you also need to keep the same number of taps on each channel), but to make the best use of available taps the "energy" centering is still better, and the delay differences can be compensated for in the crossover settings of the device (following the offset given by rephase). In any case, the "float" option should always be used as it will really reduce ripples near the Nyquists frequency, and the time difference it will imply is tiny enough to not need any compensation (it could not be compensated anyway, as it is a fractional sample...)

Thanks again for this great work, I will certainly refer to it :D
https//wavetracing.com | rephase.org
The administrator has disabled public write access.

Feature requests and suggestions 6 years 11 months ago #9974

  • John Ashman
  • John Ashman's Avatar
  • Offline
  • Gold Boarder
  • Posts: 226
  • Thank you received: 23
  • Karma: 5
Just a general question......what actually causes the pre-ringing? My NHT Xds had this and never noticed any detriment because the sound was so F'ing good otherwise, but is it from trying to fix the phase? Or from FIR? I'm really not concerned so much about phase if the speakers have a reasonably flat response.
The administrator has disabled public write access.

Feature requests and suggestions 6 years 11 months ago #9975

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Preringing occurs because linear phase filters (or phase-linearized minimal-phase filters, which end up being exactly the same thing) have a symmetrical impulse response: they are not causal.
But if you put a linear phase lowpass together with a linear phase highpass with the same parameters (complementary amplitude slopes and time aligned, in phase summation) the preringing (centered on the cuttoff frequency) of each part of the crossover will cancel each other (they are out of phase when you look at the impulses). The problem is that it gets "perfect" in-phase summation to cancel that preringing, so it *can* become apparent (measurable at least) off axis when the polar patterns of the two crossed-over drivers are not perfectly matched, or on the vertical plan. Of course the sharper the slope, the most apparent it can become off axis (because any polar pattern difference will be emphasized).

The corollary of this is that any linear phase filter (low or high pass) with no counterpart (high or low pass) will have preringing.
This typically occur with some DAC (antialins filter) at Nyquist, and it will also occur if you correct the low end response of your system (bassreflex phase compensation for example), resulting in a linear-phase high pass response...

So preringing is inherent to linear-phase filtering (and phase-linearized minimal-phase filters), but not to FIR as you can do minimal-phase filters in FIR.
A linear phase crossover (low and high pass with good summation on and off axis and "reasonable" slope) should not have any preringing problem, and preringing that could occur at the ends of the audio range are probably not audible, so...

As a side note, minimal-phase filters do not suffer from preringing, but their post-ringing is typically twice as long for the same slope. All in all they have the same amount of ringing (which can also be canceled when properly summed with a counterpart in a crossover), but it all occurs after the impulse peak.
https//wavetracing.com | rephase.org
The administrator has disabled public write access.
The following user(s) said Thank You: John Ashman, Richard

Feature requests and suggestions 6 years 11 months ago #9976

  • John Ashman
  • John Ashman's Avatar
  • Offline
  • Gold Boarder
  • Posts: 226
  • Thank you received: 23
  • Karma: 5
Thanks,that fills in the gaps in my knowledge on that. I knew that they pretty much cancelled, but thought that the pre-ringing was a new artifact, not one that was just "moved forward" in essence. From what I've heard thus far, it's practically a non-issue compared to the advantages of the steeper slopes and digital correction. I've sold some of the best conventional speakers in the world and none of those sounded as lifelike or as natural as a properly digitized speaker system. IMO, w'ere taking away huge distortions and replacing them with very small ones.
The administrator has disabled public write access.

Feature requests and suggestions 4 years 10 months ago #21217

  • planetti
  • planetti's Avatar
  • Offline
  • Senior Boarder
  • Posts: 45
  • Thank you received: 7
  • Karma: -1
Hi Thomas,

your rePhase tool is really excellent. Thank you very much. One can do sophisticated and unique things which we all wanted to try since decades B)

There is a topic which I dream of: can you consider to import external filter data into rephase in order to further manipulate the data?
Some very good automated room correction software do not allow tuning manually or splitting by linear phase XO. Is there a chance of having his function included?

Regards
The administrator has disabled public write access.
  • Page:
  • 1
  • 2
Moderators: devteam