The red line is without any EQ. It's just the sum of the nine measurements without any FDW.
Regards,
Dan
Yes I realise that, the red line is the base measurement to EQ. With that measurement go into REW's auto EQ and regenerate the correction if you are aiming for a flat response. If you set the reference level at about 68dB REW will most likely produce a fairly flat response. If not move the level up a down and see what it gives you. The EQ you were convolving before did not seem to be designed to produce a flat response.
You can see in the REW EQ window what the predicted response will look like without the need for convolving the two together, when you are happy with the prediction export it to rephase.
To follow the tutorial apply the FDW and smoothing before performing the auto EQ and turn it off before convolving the responses together then back on again to see the result.
You can see in the REW EQ window what the predicted response will look like without the need for convolving the two together, when you are happy with the prediction export it to rephase.
To follow the tutorial apply the FDW and smoothing before performing the auto EQ and turn it off before convolving the responses together then back on again to see the result.
First and foremost, thank you for taking the time to help me figure where I’m getting off track here.
Despite being a very well written tutorial the learning curve here can be a challenge for those of us new to audio measurement. Let me take you step by step on how I got where I am right.
I took the nine measurements for each channel with the acoustic timing reference enabled and all seemed to go well.
When it came to averaging them I did have FDW enabled and as is outlined on page 2 of the tutorial. I first time aligned the nine measurements. I then smoothed them at 1/6 before vector averaging for the final result for each channel.
Following the instructions below I created the EQ filters for each channel. My goal was NOT a flat response. My goal was to flatten the response using a modest amount of filters. So now I saved the EQ filters for each channel in .XML format to be imported into Rephase. Up to this point things seemed to go very well. I easily imported the averaged response from each channel into Rephase along with the matching EQ filters. Perfect!
Next, following the instructions I didn’t seem to have any issues creating the corresponding correction impulse for each channel. Keeping in mind the correction impulse was created with the averaged measurement for each channel having FDW enabled.
Give me another day or two to ponder your advice and try to put it into practice.
Once again, many thanks for your input.
Regards,
Dan
dantwomey
Would you mind pasting the links your'e refering to (Swissbear rephase etc..) ?
Would be great.
Jesper.
They're all listed on the Rephase website at rePhase - Official Site - Free FIR filtering tool
Regards,
Dan
No problem, if this is your first measurement attempt you picked a steep curve to learn on
The vector average needs to be performed first before any other form of smoothing or windowing. The timing reference should have aligned everything. Check the impulses of the individual measurements in the overlay screen to see if they are aligned correctly, if not move them or use the time align function before vector averaging.
If you smooth or window the individual measurements before averaging it will give an incorrect average. Don't throw the data away at this point.
I don't really understand the reason for that but ultimately it's your choice. As long as you are EQ'ing an actual problem that will respond well to EQ there is no reason to not use as much as you need. EQ below 500Hz is safe in most situations as the response is dominated by the room. Above 500Hz be careful because the power response of a speaker with a crossover between horizontal drivers is going to be problematic, you could do more harm than good.
No problem, if this is your first measurement attempt you picked a steep curve to learn on
The vector average needs to be performed first before any other form of smoothing or windowing. The timing reference should have aligned everything. Check the impulses of the individual measurements in the overlay screen to see if they are aligned correctly, if not move them or use the time align function before vector averaging.
If you smooth or window the individual measurements before averaging it will give an incorrect average. Don't throw the data away at this point.
I don't really understand the reason for that but ultimately it's your choice. As long as you are EQ'ing an actual problem that will respond well to EQ there is no reason to not use as much as you need. EQ below 500Hz is safe in most situations as the response is dominated by the room. Above 500Hz be careful because the power response of a speaker with a crossover between horizontal drivers is going to be problematic, you could do more harm than good.
My first goal is to master creating a minimum phase system. After that I'll tinker with amplitude.
Regards,
Dan
It's not a problem of importance, rather of ergonomy, if you target to have
a minimum phase system, first cancel the excess phase, then easier try
different EQ to your taste without wondering any longer about phase.
I accept the point that you are making but I disagree about it's importance. If a crossover is well designed and has good phase tracking through the crossover region and there is no magnitude issue caused by it, in most cases it will be very difficult to hear the difference between that and the same crossover made phase linear, if a difference can be heard it will be even more difficult to say if one if preferable than the other. This is why I say it is less important.
I think this is one of those areas where it is more of a problem for the eyes than the ears.
The graphs posted above show that without a good understanding of what you are trying to achieve and why, it is easy to make things worse not better.
In my experience where targeting a minimum phase bandpass response is most audible is below 200Hz and this can result in quite audible changes and they are easier to say which are preferable. This does take some trial and error with measurement and correction as it is quite easy to make things sound worse here too
I think this is one of those areas where it is more of a problem for the eyes than the ears.
The graphs posted above show that without a good understanding of what you are trying to achieve and why, it is easy to make things worse not better.
In my experience where targeting a minimum phase bandpass response is most audible is below 200Hz and this can result in quite audible changes and they are easier to say which are preferable. This does take some trial and error with measurement and correction as it is quite easy to make things sound worse here too
In my experience where targeting a minimum phase bandpass response is most audible is below 200Hz and this can result in quite audible changes and they are easier to say which are preferable. This does take some trial and error with measurement and correction as it is quite easy to make things sound worse here too
In my experience, canceling the excess phase gives a better focus on the soundstage, holography, width and depth,
even if it is barely perceptible for 90% of the recordings, a bit like MP3 where the feeling of something lost confirms itself in the long term, a kind of difference between the real sound and a "hi-fi" sound.