I've completed this and the preceding workshop video, the result was amazing! ⭐ I'm using Roon to implement the filters, passive stereo full-range speakers. Just out of curiosity I downloaded the free trial for Dirac Live Room Correction Suite Stereo & Bass Control, they do not even come close to the level of clarity the filters from this workshop produces. Nevertheless, the Dirac software does correct boomy bass, but does little with the mids and overall soundstage for my setup compared to this.
@@ocaudiophile I recently updated and repositioned my speakers, which essentially means I have to repeat Workshop I and II. Prior to revisiting this process, is it advisable to integrate the following into both each workshop: 1. Should I craft custom microphone filters using "Boost your mic precision with REW Pro"? (Note: I own a UMIK-2). 2. Would it be beneficial to apply the "Phase Match Your Speakers with AllPass Filters" technique? 3. Are there any upcoming developments related to passive stereo full-range speakers that I might consider waiting for? I enjoyed the outcome from the two workshops. Now, I'm looking to identify the most effective way to proceed, given that these involve considerable time investments and are designed for long-term use.
1. Umik-2 is already highly accurate right out of the box with its custom calibration file so that's not necessary. 2. Cohesion between the left and right speakers are very important for overall sound quality and imaging and can be harmed by almost any asymmetry in the room. Allpass filters are used to try and fix these but this is only possible to a limited extent as filters required for major differences will cause pre-echo effects. If you have a fairly symmetric set up in a fairly rectangular room, you can just leave phase correction at "uniform" XO and box correction of your speakers. But since you are watching this channel, I am pretty certain you will want to explore both options ;) 3. Optimal sound reproduction is a never ending journey and that's one of the reasons I am so into it. I continuously work on new techniques and there are frequently minor findings here and there. I will share them when they accumulate to the extent that I decide it's worth making a new video. @@esmolan
Hey man just wanted to say that your videos have been extremely helpful in helping me optimize my system. While I didnt have great results from the speaker crossover/box phase correction and vba filter, the frequency response correction and excess phase inversion did wonders. I originally started with the inversion filter video you did which was already a massive improvement, but this updated approach overall provided a more balanced sound which i havent been able to stop listening to when i get the opportunity. Thanks for being obsessive compulsive as without it we wouldn't get these extremely insightful and knowledge-filled tutorials. Consider posting a link so we can donate a beer or two for your troubles. Cheers!
Since I am also obsessive compulsive when it comes to audio, I would really like to thank you for all your videos and help. I was avoiding for a long time to get UMIK-1, since I had no time to learn REW, but you changed that. Those front speakers, surround speakers, all is now placed as it should be, listener in the middle (exact middle) from left and right speaker. And the best thing is that your methods work flowless.
Thank you again for once more providing such helpful information! One of your latest videos came with an additional PDF as a step by step manual. Could you provide such a step by step transcript also for your latest 2 videos? That would be really great and helpful! All the best!
Man, you make some great videos with very good info. I just wish you would slow down a little. I'm guessing these videos are made to teach people how to use REW better, but it is difficult to keep up with the speed
Excellent tutorial as usual. By far the best and most informative on youtube. Keep up the good work! I am eagerly looking forward to the optimal speaker placement video. That is the thing I have the most difficulties with.
@@ocaudiophile Thanks! I will check the papers linked in that video. Btw, what are the changes, if any to be applied to this method if there are subwoofers in the picture, crossed over using a miniDSP?
@@paulsoumya same filters apply to a sub+speaker combination but the sub should be properly integrated to the speakers (correct xo freq and correct delay)
@@ocaudiophile Thanks for the reply. I have dual subs which are placed symmetrically next to the FL and FR speakers. I use them as stereo subs, crossing over the FL to the left sub and the FR to the right sub using a miniDSP 2x4. I hope that's the way it should be(?) I remember having seen in one of your videos on how to set up the proper crossover between a speaker and a sub. I will try to look that up again. Thanks!
The new trim function is essential to lower the latency. Should be in the stable REW release rather than having to go and fetch the Beta version to have this feature.
Wowww. This is this a LOT of steps, but I love seeing someone take REW this far! This is no doubt the most detailed use of REW that I'm aware of. What if I don't know the crossover frequency for my bi-amped studio monitors (an old pair of KRKs that they never posted the crossover data for)? How could I test this? How important is that to know? I would love to see measurement data from after you correct your monitors! That would be fantastic Wonderful videos. Thanks again!
İ assume your speakers are passively bi-amped or else you wouldn't need xo phase correction. With a two-way speaker the xo freq could be extracted easily from a near field measurement. With 3-way speaker it could be harder. Even visually guessed linearization will improve things somewhat though. Measurement data will be identical to REW calculations and anyone can get a full flat response even by ear within minutes just by using a PEQ. It's about avoiding the boxy/throttled sound while flattening the response curve.
I have completed creating all the specific filters but as Im using a minidsp 2x4HD, how do I covert the final set of filter files to Bi-Quads to install in my digital processor? I asked the question of REW and they say it cant be done!
I posted a tutorial about how to do it with biquads: czcams.com/video/ydMkpPKYuaA/video.html Howeve newer Minidsp models weren't able to process any time reversed filters. You can copy EQ filters from REW (also in biquad format if you wish) and add a 1024 tap FIR filter for phase correction created in rephase to your 2x4HD. VBA and excess phase inversion will need to be skipped.
@@ocaudiophile quick question, should you perform SBIR from each channels own measurement, use a single measurement for both, or the average of both channels?
You don't need measurements for front wall (the wall behind speakers and in front of you) SBIR, it's simply moving the speakers far enough from the wall so that the lowest frequency cancellation notch is outside the frequency range that your speakers reproduce i.e with 1.5 m distance, you will avoid a dip if your speakers can't go below 60Hz [(343/(4*1.5))=57Hz]. But this is very difficult to achieve in real life for long distances required and low speakers bass roll off frequencies. A good and easier to implement formula that works great in rectangular rooms is placing speakers 1/5 of the room length and placing the LP 1/5 of the room length from the rear wall and achieve an equilateral triangle.
Is there any way that taking this and importing it to an audussey curve will make a difference? or is this just not possible due to restrictions on audussey? If it is possible should I do this on all my speakers then include these in the correction files? Thanks! Love your videos!
Audyssey applies 1024 taps (XT32) FIR filters to each speaker and they are linear phase meaning no phase change to the original response. Hence, you're limited to only frequency domain corrections. Phase corrections are called time domain corrections and placing speakers as symmetrically as possible adjusting speaker distances accurately is the most you can do with Audyssey.
@@ocaudiophile damn that's annoying. So the most audyssey allows is an eq? What AVRs are there that are not massively expensive that allow for all these fantastic tweaks that you're teaching?
@ocaudiophile - what Your movie/method is best for calibrating 2.1 desktop system ? For now I use "convolution with inversion" method and result is great. Computer ->usb cable (could be spdif)->dac->Y rca cables to front speakers and to sub. Greetings
These two would help a lot: czcams.com/video/qoFZPlXrTeM/video.html Generate minimum phase versions of the final filters if you hear any ringing with the method filters. And align vector average of your speakers with the sub as if they are two subs with the method here: czcams.com/video/DofUiHz7Fu4/video.html
Hi OCA, your videos are really great. I just need time to digest them as they're packed with info. Not sure if you've been asked before, but I was wondering if you have come across the 'Ambiophonics' idea. This creates a 180 degrees soundstage (or 360 degree with 4 speakers) having speakers close together and using a Recursive Crosstalk Elimination filter (called RACE where A is for Ambiophonics). I never managed to find a way to listen to it even if some say it's easy to replicate with a convolver. Some say that this is also the basis for the BACCH Labs audio application. Would be great to have your take on it. Thanks
Thank you. I haven't heard of Ambiophonics but did a little reading just now. It's interesting. They claim eliminating crosstalk completely between the right and left channels gives the listener a greater dimensionalized impression of the overall sound. A polarity inverted copy of the right channel signal could be added to the left channel with a very precise delay (as in the time it takes sound to travel the distance between the ears) and vice versa. That would in theory avoid right channel signal to be heard from the left ear. However the convolution filter should be actively replacing itself continously. I guess this is where the word recursive (RACE) is coming from. I will start thinking and researching about it.
@@ocaudiophile oh that is great. Glad to have brought this to your attention. While reading I saw that the algorithm is public domain. I will try and track it down and give you the link. Thanks and best regards
Hi OCA! Thanks for the latest workshop videos. One question.... did you use only the fast sweep length measurements for all these filters or did you take longer length sweeps when doing the EQ filters? I thought you said to use fast sweeps for phase filters and long sweeps for EQ filters in the past.
I used 128K sweeps for the phase filter and continued with it for the tutorial but I would normally switch to long measurements right after finding out my optimal XO & box correction filter. Also, I'd probably take 5 measurements at the same mic spot.
Amazing video, I like the updated methods. How exactly should I go about correcting a system in which there are more than just the L and R speakers? I have 2x Genelec 8030c, and 2x Stereo Integrity 18" subs, about 1.5 feet behind each Genelec speaker. 120hz seems to be the ideal xover frequency for the space. Do I measure the L as the left speaker + left subwoofer, perhaps? Thank you!!!
Thanks. First phase align the subs with each other using REW's "Alignment tool". That will give you the most powerful bass response. Then time align the "aligned sum" with the speakers. Then decide on the best XO frequency. Ideally, that should be where your speakers start rolling off -3dB. Once everything is aligned and crossedover, you can take measurements for left and right speakers with the subs active and apply correction on these responses just as you would without the subs. Speaker crossover phase linearization will be the same and according to speaker specs. Box/port phase correction will need to be replaced by the subs' design.
Hi, Many thanks for your excellent tutorials. I have just finished DRC part 2 using my copy of REW (5.30 Beta 1) and have stumbled at the (nearly) last hurdle. I cannot find the DIRAC operand in TA used to merge with the final A*B result. Is this because it is no longer included, or do I need the Pro version, or is there an alternative? Regards
Thanks a lot OCA for this videos. One question folowing the steps on these Workshop do I still need to align L&R speaker as you showed on your convolution with inversion filter ?
Well, precisely central LP position is one of the prerequisites mentioned in the workshop as well however millimetric differences within your head movement space can be ignored.
Hello! Thanks for these videos it seems very interesting and I would love to try your method. Have you got any idea on how to implement it for a proffessional studio environment? MiniDSP would be an affordable option but I wonder if it exists a plugin that could mount the convolution files so that the correction can be dialed in inside a DAW like Protools. Thanks again!
I am unfamiliar with the pro studio/DAW environment but convolution reverb plugins should be able to process FIR filters AFAIK. A MiniDSP 2x4HD can process up to a total of 4096 taps of FIR filters at 4 channels. This is not very generous for phase correction especially in the low bass region but will still do the job with some windowing and optimization tricks.
Thanks OCA! I tried this method on my studio monitors and it turned out great! However, I find myself in this weird predicament: I have a sub and my audio interface only supports 2 channels. So I'm using the crossover inside the sub (which supports 80Hz and 120Hz) at 80Hz and connecting PC -> interface -> sub -> monitors. What do I need to read and/or understand, so that I can maximise performance in this setup? Even if I get an interface that supports more output channels, allowing a digital crossover, I still don't know how to integrate a subwoofer into the main speakers 😕.
Can you not use the PC's own soundcard? That way you can use the subwoofer and left right speaker outputs separately and with Equaliser APO you can obtain a full digital interface.
Sorry I don't know much about Macs. I believe it should be possible to use multiple channels through audio midi setup but have never done myself. Connecting speakers directly to the subwoofer and selecting the crossover on the sub will also work but you need to move the sub equal distance from LP as the speakers otherwise there'll be timing issues.
Thank you so much for all your work! Is it possible to make a guide about the MultEQ-X software? I tried everything to get a Harman Curve there but nothing works... Would be absolutly awsome :)
@@ocaudiophile That sounds great! Then I hope for a quick fix. May I ask why you don't turn off DEQ? That's how I run it at the moment. Sorry for the questions. Your videos are so full of informations. Its hard to find the right things for a newbie :/
@@philipph119 I use 2 presets, in one DEQ is turned off and it sounds better than the other at the usual level I listen to movies. The other preset has DEQ on and I use it for low volume regular TV watching. Even the DEQ preset is calibrated such that at 75dB speaker responses prefectly replicate the Harman curve. SUpreme Audyssey video has instructions on how to do that.
Hi, thanks as always for these insights. Regarding measurement process do you think we can take an MMM using RTA for EQ ?early reflections would be cancelled with averaging, literature showed that you measure the speaker above schroeder frequency with this method. What do you think? Just want to be sure, i would take 2 or 3 positions for sweeps inclusing MLP, and save a LOT of time.
In theory yes but with MMM you don't only lose the reflections, you lose a great deal from the HF information and can only read a sharp roll off. Besides you lose phase data completely. As Audyssey deosn't correct phase, I had thought of MMM for Audyssey measurements but the HF response was nowhere near workable. Right window resizing is the best method I can think of to remove reflections and limit HF equalization.
My preamp has no dedicated sub output, which means I don't have a seperate channel for the sub. Normally I measure each speaker individually but should I include the sub with the individual L&R measurements or continue to do them separately? If seperate, how would I handle filters for the sub? I'm using roon and I use a hpf at 80hz on each of the mains to let the sub handle the bass. Also, thanks for the videos. I've learned a lot and am in the process of creating my own filters based on my preferred target curve.
If I understand correctly, your subs get the full frequency spectrum. Unless there's a lowpass filter applied somewhere, this would make the overal sound muddy. If you have variable phase adjustments on the subs, you should align them. You can perform RTA measuremnts in REW to optimize phases. If that's not an option, it's possible to cretae filters which will delay frequencies below 80Hz a certain amount but I haven't tested this personally.
@ocaudiophile Thanks for the reply. When I first got the sub it was exactly as you said. I literally spent weeks dialing in the settings because I had to do everything manually. What I'd like to know is to achieve the best possible sound should I eq each speaker individually or as mains plus sub? Then if it is individually, how would I apply those settings to the sub without a seperate subwoofer channel?
Each of the filters can be skipped and excluded in the combination filter. You need to create XO filter from the uncorrected, short length measurement so it will have to be the first one created. VBA filter should precede the EQ filters. EQ filters need to be done in a row. Excess phase filter has to be the last one.
Hi OCA, the frequency choosed for merge 100hz-1000hz is 1413. But how was it calculated? Did you use a precise formula? I ask you because I would like to merge on 100hz-3000 hz. I tried with 1413 but after 2000hz does not align the phase well. What value should I use? Thank you for everything you do, you taught me how to use REW and REPHASE!
Just trial and error, merge uses SPL and phase levels at merge point so it's not predictable. Try 2000 and go from there until merged response seems flat and around 0
@@ocaudiophile I ask you another question, I would like to do the procedure of the 4 filters by inserting 4 subwoofers. I was thinking of doing 4 individual subwoofer measurements (0-80hz), aligning the pulses and doing the EQ for each sub. Then join the 4 equalizations together. Then I do the two front measurements (80-24kh), doing EQ and finally I combine the EQ sub and the EQ front. In general then I will follow the tutorial but with the addition of subs. Can it be okay?
@@BarileTixxoFilms You should always measure 0-24000Hz because filters are going to be 48000Hz and trace arithmetic operations may generate noise otherwise. EVen if you EQ each sub, the combined sub response will be very different and will need equalization. The best way is to first align delays between all subs such that they produce the deepesyt bass and have a relatively flat total response and then apply one correction for this summed response. Use REW's alignment tool for sub alignment. You can align two responses each time, vector averaging aligned subs and aligning this average with the third and so on.
I think I am missing something obvious - I created some phase correction filters, and using CamillaDSP (on Windows) applied the filters. I'm getting delay in audio output (e.g. audio is delayed watching this video), seems equal in length to about half the "impulse length" of the filter. With such minimum-phase filter I thought I wouldn't have this issue?
@@ocaudiophile Yes I saw that approach in this video - but even so, your "very fast" filter is still going to be a few hundred ms, so unusable for anything with video. But I thought it was a moot point since minimum-phase filter shouldn't introduce any delay associated with its length? (I posted in the CamillaDSP and rePhase threads on diyAudio forum). Thanks again for these videos - I'm already getting predicted huge improvements in phase response. Hopefully I can sort out my delay issue to properly enjoy them :)
@MikeyAntonakakis your video player should be able to apply a delay to video to sync it with the audio. Most players have an audio delay / lip sync function
@@ocaudiophile I figured it out with help from someone on diyAudio - it was a setting in rePhase. I was able to create a 131k taps filter with no perceptible delay. In the "impulse settings" area, instead of using "middle" centering, typing "0" or "1" and "use closest perfect impulse" seems to have done the trick. The problem was that with the "middle" setting, it moves them to the middle of the file, so especially for a simple filter there is a ton of leading zeros in the wav file. Setting to zero instead fixed the delay, and it's reflected in the "impulse delay" calculation after generating the filter too. Might mean you end up with less manual trimming in REW :)
132k taps at 48kHZ sampling rate will generate 1.4 seconds long filters. Most filters in REW can be trimmed down to 32768 taps which is fast enough for any audio application. To speed up and decrease the number of taps in rePhase you need to use rectangular windows and 99% or 100% centering with moderate optimization. There should be detailed instructions in my MiniDSP video where I create filters with just 1022 taps (that's "5 ms" at 96kHZ sampling rate of MiniDSP)
Hello! as for EQ filters.... instead of making 3 different filters and then joining them I do VAR SMOOTHING that applies you 1/48 on low frequencies and psychoacoustic on high frequencies. I don't need to have boosts in any frequency and drop only the peaks. Do you think it could be okay?
Var smoothing is closest to the technique and will work but I have obtained highest clarity with the exact same settings in the tutorial including 5db boost allowance up to 200hz
@@ocaudiophile I have another question, when creating the XO filter with rephase I have to drop the gain by 3 db (in rephase) otherwise it gives me clipping.. It's normal? Can I make the filter at -3db or is it not good?
@@ocaudiophile Last question! In the last filter (phase inversion with rew), could it be advantageous to make a range of 80hz-2000hz? With these frequencies I can still use 3 fwd cycles. Should I always use the 1413hz frequency in that case? Thank you for everything you do...in my opinion you are a fucking genius!!!!
1413 was chosen for 100-1000 range, not sure if it's optimal for 80-1000, you need to try and see. Correcting phase below 100Hz has potential to cause pre-echo.
I've completed this and the preceding workshop video, the result was amazing! ⭐
I'm using Roon to implement the filters, passive stereo full-range speakers.
Just out of curiosity I downloaded the free trial for Dirac Live Room Correction Suite Stereo & Bass Control, they do not even come close to the level of clarity the filters from this workshop produces. Nevertheless, the Dirac software does correct boomy bass, but does little with the mids and overall soundstage for my setup compared to this.
Great, thanks.
Dirac PC trial version also tends to mess with Windows ASIO drivers quite badly, good luck fixing it ;)
@@ocaudiophile
I recently updated and repositioned my speakers, which essentially means I have to repeat Workshop I and II.
Prior to revisiting this process, is it advisable to integrate the following into both each workshop:
1. Should I craft custom microphone filters using "Boost your mic precision with REW Pro"? (Note: I own a UMIK-2).
2. Would it be beneficial to apply the "Phase Match Your Speakers with AllPass Filters" technique?
3. Are there any upcoming developments related to passive stereo full-range speakers that I might consider waiting for?
I enjoyed the outcome from the two workshops. Now, I'm looking to identify the most effective way to proceed, given that these involve considerable time investments and are designed for long-term use.
1. Umik-2 is already highly accurate right out of the box with its custom calibration file so that's not necessary.
2. Cohesion between the left and right speakers are very important for overall sound quality and imaging and can be harmed by almost any asymmetry in the room. Allpass filters are used to try and fix these but this is only possible to a limited extent as filters required for major differences will cause pre-echo effects. If you have a fairly symmetric set up in a fairly rectangular room, you can just leave phase correction at "uniform" XO and box correction of your speakers. But since you are watching this channel, I am pretty certain you will want to explore both options ;)
3. Optimal sound reproduction is a never ending journey and that's one of the reasons I am so into it. I continuously work on new techniques and there are frequently minor findings here and there. I will share them when they accumulate to the extent that I decide it's worth making a new video. @@esmolan
Hey man just wanted to say that your videos have been extremely helpful in helping me optimize my system. While I didnt have great results from the speaker crossover/box phase correction and vba filter, the frequency response correction and excess phase inversion did wonders. I originally started with the inversion filter video you did which was already a massive improvement, but this updated approach overall provided a more balanced sound which i havent been able to stop listening to when i get the opportunity. Thanks for being obsessive compulsive as without it we wouldn't get these extremely insightful and knowledge-filled tutorials. Consider posting a link so we can donate a beer or two for your troubles. Cheers!
Thank you.
Since I am also obsessive compulsive when it comes to audio, I would really like to thank you for all your videos and help.
I was avoiding for a long time to get UMIK-1, since I had no time to learn REW, but you changed that.
Those front speakers, surround speakers, all is now placed as it should be, listener in the middle (exact middle) from left and right speaker.
And the best thing is that your methods work flowless.
Thanks ;)
Thank you again for once more providing such helpful information! One of your latest videos came with an additional PDF as a step by step manual. Could you provide such a step by step transcript also for your latest 2 videos? That would be really great and helpful! All the best!
Noted
Man, you make some great videos with very good info. I just wish you would slow down a little. I'm guessing these videos are made to teach people how to use REW better, but it is difficult to keep up with the speed
Noted!
Excellent tutorial as usual. By far the best and most informative on youtube. Keep up the good work! I am eagerly looking forward to the optimal speaker placement video. That is the thing I have the most difficulties with.
Thank you for the kind words. The preceeding video to this one (Part I) has quite a bit of info and links to papers about speaker placement!
@@ocaudiophile Thanks! I will check the papers linked in that video. Btw, what are the changes, if any to be applied to this method if there are subwoofers in the picture, crossed over using a miniDSP?
@@paulsoumya same filters apply to a sub+speaker combination but the sub should be properly integrated to the speakers (correct xo freq and correct delay)
@@ocaudiophile Thanks for the reply. I have dual subs which are placed symmetrically next to the FL and FR speakers. I use them as stereo subs, crossing over the FL to the left sub and the FR to the right sub using a miniDSP 2x4. I hope that's the way it should be(?) I remember having seen in one of your videos on how to set up the proper crossover between a speaker and a sub. I will try to look that up again. Thanks!
@@paulsoumya Yes a MiniDSP 2x4 is good enough to apply the required crossovers, delays and some PEQ.
The new trim function is essential to lower the latency.
Should be in the stable REW release rather than having to go and fetch the Beta version to have this feature.
Wowww. This is this a LOT of steps, but I love seeing someone take REW this far! This is no doubt the most detailed use of REW that I'm aware of. What if I don't know the crossover frequency for my bi-amped studio monitors (an old pair of KRKs that they never posted the crossover data for)? How could I test this? How important is that to know? I would love to see measurement data from after you correct your monitors! That would be fantastic Wonderful videos. Thanks again!
İ assume your speakers are passively bi-amped or else you wouldn't need xo phase correction. With a two-way speaker the xo freq could be extracted easily from a near field measurement. With 3-way speaker it could be harder.
Even visually guessed linearization will improve things somewhat though.
Measurement data will be identical to REW calculations and anyone can get a full flat response even by ear within minutes just by using a PEQ. It's about avoiding the boxy/throttled sound while flattening the response curve.
I have completed creating all the specific filters but as Im using a minidsp 2x4HD, how do I covert the final set of filter files to Bi-Quads to install in my digital processor?
I asked the question of REW and they say it cant be done!
I posted a tutorial about how to do it with biquads:
czcams.com/video/ydMkpPKYuaA/video.html
Howeve newer Minidsp models weren't able to process any time reversed filters. You can copy EQ filters from REW (also in biquad format if you wish) and add a 1024 tap FIR filter for phase correction created in rephase to your 2x4HD. VBA and excess phase inversion will need to be skipped.
Thanks!
Extremely kind of you, thanks!
@@ocaudiophile quick question, should you perform SBIR from each channels own measurement, use a single measurement for both, or the average of both channels?
You don't need measurements for front wall (the wall behind speakers and in front of you) SBIR, it's simply moving the speakers far enough from the wall so that the lowest frequency cancellation notch is outside the frequency range that your speakers reproduce i.e with 1.5 m distance, you will avoid a dip if your speakers can't go below 60Hz [(343/(4*1.5))=57Hz]. But this is very difficult to achieve in real life for long distances required and low speakers bass roll off frequencies. A good and easier to implement formula that works great in rectangular rooms is placing speakers 1/5 of the room length and placing the LP 1/5 of the room length from the rear wall and achieve an equilateral triangle.
are these new filters superior to the methods shown in the previous videos?
Yes
Is there any way that taking this and importing it to an audussey curve will make a difference? or is this just not possible due to restrictions on audussey? If it is possible should I do this on all my speakers then include these in the correction files?
Thanks! Love your videos!
Audyssey applies 1024 taps (XT32) FIR filters to each speaker and they are linear phase meaning no phase change to the original response. Hence, you're limited to only frequency domain corrections. Phase corrections are called time domain corrections and placing speakers as symmetrically as possible adjusting speaker distances accurately is the most you can do with Audyssey.
@@ocaudiophile damn that's annoying. So the most audyssey allows is an eq? What AVRs are there that are not massively expensive that allow for all these fantastic tweaks that you're teaching?
@@olitre3 There are some but with over the moon prices like Trinnov, Lyngdorf, Stormaudio
@ocaudiophile - what Your movie/method is best for calibrating 2.1 desktop system ? For now I use "convolution with inversion" method and result is great. Computer ->usb cable (could be spdif)->dac->Y rca cables to front speakers and to sub. Greetings
These two would help a lot:
czcams.com/video/qoFZPlXrTeM/video.html
Generate minimum phase versions of the final filters if you hear any ringing with the method filters. And align vector average of your speakers with the sub as if they are two subs with the method here:
czcams.com/video/DofUiHz7Fu4/video.html
Hi OCA, your videos are really great. I just need time to digest them as they're packed with info.
Not sure if you've been asked before, but I was wondering if you have come across the 'Ambiophonics' idea. This creates a 180 degrees soundstage (or 360 degree with 4 speakers) having speakers close together and using a Recursive Crosstalk Elimination filter (called RACE where A is for Ambiophonics). I never managed to find a way to listen to it even if some say it's easy to replicate with a convolver. Some say that this is also the basis for the BACCH Labs audio application.
Would be great to have your take on it. Thanks
Thank you. I haven't heard of Ambiophonics but did a little reading just now. It's interesting. They claim eliminating crosstalk completely between the right and left channels gives the listener a greater dimensionalized impression of the overall sound. A polarity inverted copy of the right channel signal could be added to the left channel with a very precise delay (as in the time it takes sound to travel the distance between the ears) and vice versa. That would in theory avoid right channel signal to be heard from the left ear. However the convolution filter should be actively replacing itself continously. I guess this is where the word recursive (RACE) is coming from. I will start thinking and researching about it.
@@ocaudiophile oh that is great. Glad to have brought this to your attention. While reading I saw that the algorithm is public domain. I will try and track it down and give you the link.
Thanks and best regards
Hi OCA! Thanks for the latest workshop videos. One question.... did you use only the fast sweep length measurements for all these filters or did you take longer length sweeps when doing the EQ filters? I thought you said to use fast sweeps for phase filters and long sweeps for EQ filters in the past.
I used 128K sweeps for the phase filter and continued with it for the tutorial but I would normally switch to long measurements right after finding out my optimal XO & box correction filter. Also, I'd probably take 5 measurements at the same mic spot.
@@ocaudiophilehello. What if my measurement system supports only 4, 16 and 65k Chirp measurements(Clio Pocket). Is it good enough for room correction?
John has added 64k to REW so I believe that's acceptable but I wouldn't go below that.
Amazing video, I like the updated methods. How exactly should I go about correcting a system in which there are more than just the L and R speakers? I have 2x Genelec 8030c, and 2x Stereo Integrity 18" subs, about 1.5 feet behind each Genelec speaker. 120hz seems to be the ideal xover frequency for the space. Do I measure the L as the left speaker + left subwoofer, perhaps? Thank you!!!
Thanks. First phase align the subs with each other using REW's "Alignment tool". That will give you the most powerful bass response. Then time align the "aligned sum" with the speakers. Then decide on the best XO frequency. Ideally, that should be where your speakers start rolling off -3dB. Once everything is aligned and crossedover, you can take measurements for left and right speakers with the subs active and apply correction on these responses just as you would without the subs. Speaker crossover phase linearization will be the same and according to speaker specs. Box/port phase correction will need to be replaced by the subs' design.
@@ocaudiophile thank you!!! Will try this.
Hi,
Many thanks for your excellent tutorials. I have just finished DRC part 2 using my copy of REW (5.30 Beta 1) and have stumbled at the (nearly) last hurdle.
I cannot find the DIRAC operand in TA used to merge with the final A*B result. Is this because it is no longer included, or do I need the Pro version, or is there an alternative?
Regards
You can generate a Dirac perfect impulse response with "generate measurement from filters" in EQ window when there are no filters.
Thanks a lot OCA for this videos. One question folowing the steps on these Workshop do I still need to align L&R speaker as you showed on your convolution with inversion filter ?
Well, precisely central LP position is one of the prerequisites mentioned in the workshop as well however millimetric differences within your head movement space can be ignored.
Hello! Thanks for these videos it seems very interesting and I would love to try your method. Have you got any idea on how to implement it for a proffessional studio environment? MiniDSP would be an affordable option but I wonder if it exists a plugin that could mount the convolution files so that the correction can be dialed in inside a DAW like Protools. Thanks again!
I am unfamiliar with the pro studio/DAW environment but convolution reverb plugins should be able to process FIR filters AFAIK.
A MiniDSP 2x4HD can process up to a total of 4096 taps of FIR filters at 4 channels. This is not very generous for phase correction especially in the low bass region but will still do the job with some windowing and optimization tricks.
Thanks OCA! I tried this method on my studio monitors and it turned out great!
However, I find myself in this weird predicament: I have a sub and my audio interface only supports 2 channels. So I'm using the crossover inside the sub (which supports 80Hz and 120Hz) at 80Hz and connecting PC -> interface -> sub -> monitors.
What do I need to read and/or understand, so that I can maximise performance in this setup?
Even if I get an interface that supports more output channels, allowing a digital crossover, I still don't know how to integrate a subwoofer into the main speakers 😕.
Can you not use the PC's own soundcard? That way you can use the subwoofer and left right speaker outputs separately and with Equaliser APO you can obtain a full digital interface.
@@ocaudiophile The "PC" in-question is a Macbook pro. Are you talking about the 3.5mm output on that?
Sorry I don't know much about Macs. I believe it should be possible to use multiple channels through audio midi setup but have never done myself. Connecting speakers directly to the subwoofer and selecting the crossover on the sub will also work but you need to move the sub equal distance from LP as the speakers otherwise there'll be timing issues.
Thank you so much for all your work! Is it possible to make a guide about the MultEQ-X software? I tried everything to get a Harman Curve there but nothing works... Would be absolutly awsome :)
I don't use X but I'll buy it if they add option to stop surround boost with DEQ. Then a tutorial will be imperative ;)
@@ocaudiophile That sounds great! Then I hope for a quick fix. May I ask why you don't turn off DEQ? That's how I run it at the moment. Sorry for the questions. Your videos are so full of informations. Its hard to find the right things for a newbie :/
@@philipph119 I use 2 presets, in one DEQ is turned off and it sounds better than the other at the usual level I listen to movies. The other preset has DEQ on and I use it for low volume regular TV watching. Even the DEQ preset is calibrated such that at 75dB speaker responses prefectly replicate the Harman curve. SUpreme Audyssey video has instructions on how to do that.
Hi, thanks as always for these insights. Regarding measurement process do you think we can take an MMM using RTA for EQ ?early reflections would be cancelled with averaging, literature showed that you measure the speaker above schroeder frequency with this method. What do you think? Just want to be sure, i would take 2 or 3 positions for sweeps inclusing MLP, and save a LOT of time.
In theory yes but with MMM you don't only lose the reflections, you lose a great deal from the HF information and can only read a sharp roll off. Besides you lose phase data completely. As Audyssey deosn't correct phase, I had thought of MMM for Audyssey measurements but the HF response was nowhere near workable. Right window resizing is the best method I can think of to remove reflections and limit HF equalization.
@@ocaudiophile got it thanks!
My preamp has no dedicated sub output, which means I don't have a seperate channel for the sub. Normally I measure each speaker individually but should I include the sub with the individual L&R measurements or continue to do them separately? If seperate, how would I handle filters for the sub? I'm using roon and I use a hpf at 80hz on each of the mains to let the sub handle the bass. Also, thanks for the videos. I've learned a lot and am in the process of creating my own filters based on my preferred target curve.
If I understand correctly, your subs get the full frequency spectrum. Unless there's a lowpass filter applied somewhere, this would make the overal sound muddy.
If you have variable phase adjustments on the subs, you should align them. You can perform RTA measuremnts in REW to optimize phases.
If that's not an option, it's possible to cretae filters which will delay frequencies below 80Hz a certain amount but I haven't tested this personally.
@ocaudiophile Thanks for the reply. When I first got the sub it was exactly as you said. I literally spent weeks dialing in the settings because I had to do everything manually. What I'd like to know is to achieve the best possible sound should I eq each speaker individually or as mains plus sub? Then if it is individually, how would I apply those settings to the sub without a seperate subwoofer channel?
You can apply all these filters to the combined sub+speaker response.
hi, SHOULD FILTERS BE CREATED IN THIS PRECISE ORDER OR IS IT IRRELEVANT?
Each of the filters can be skipped and excluded in the combination filter. You need to create XO filter from the uncorrected, short length measurement so it will have to be the first one created. VBA filter should precede the EQ filters. EQ filters need to be done in a row. Excess phase filter has to be the last one.
Hi OCA, the frequency choosed for merge 100hz-1000hz is 1413.
But how was it calculated? Did you use a precise formula? I ask you because I would like to merge on 100hz-3000 hz. I tried with 1413 but after 2000hz does not align the phase well. What value should I use? Thank you for everything you do, you taught me how to use REW and REPHASE!
Just trial and error, merge uses SPL and phase levels at merge point so it's not predictable. Try 2000 and go from there until merged response seems flat and around 0
@@ocaudiophile Thank you!!!
@@ocaudiophile I ask you another question, I would like to do the procedure of the 4 filters by inserting 4 subwoofers. I was thinking of doing 4 individual subwoofer measurements (0-80hz), aligning the pulses and doing the EQ for each sub. Then join the 4 equalizations together. Then I do the two front measurements (80-24kh), doing EQ and finally I combine the EQ sub and the EQ front. In general then I will follow the tutorial but with the addition of subs. Can it be okay?
@@BarileTixxoFilms You should always measure 0-24000Hz because filters are going to be 48000Hz and trace arithmetic operations may generate noise otherwise. EVen if you EQ each sub, the combined sub response will be very different and will need equalization. The best way is to first align delays between all subs such that they produce the deepesyt bass and have a relatively flat total response and then apply one correction for this summed response. Use REW's alignment tool for sub alignment. You can align two responses each time, vector averaging aligned subs and aligning this average with the third and so on.
I think I am missing something obvious - I created some phase correction filters, and using CamillaDSP (on Windows) applied the filters. I'm getting delay in audio output (e.g. audio is delayed watching this video), seems equal in length to about half the "impulse length" of the filter. With such minimum-phase filter I thought I wouldn't have this issue?
You can decrease the number of taps of the filter wav file before exporting with "trim IR" function.
@@ocaudiophile Yes I saw that approach in this video - but even so, your "very fast" filter is still going to be a few hundred ms, so unusable for anything with video. But I thought it was a moot point since minimum-phase filter shouldn't introduce any delay associated with its length? (I posted in the CamillaDSP and rePhase threads on diyAudio forum). Thanks again for these videos - I'm already getting predicted huge improvements in phase response. Hopefully I can sort out my delay issue to properly enjoy them :)
@MikeyAntonakakis your video player should be able to apply a delay to video to sync it with the audio. Most players have an audio delay / lip sync function
@@ocaudiophile I figured it out with help from someone on diyAudio - it was a setting in rePhase. I was able to create a 131k taps filter with no perceptible delay. In the "impulse settings" area, instead of using "middle" centering, typing "0" or "1" and "use closest perfect impulse" seems to have done the trick. The problem was that with the "middle" setting, it moves them to the middle of the file, so especially for a simple filter there is a ton of leading zeros in the wav file. Setting to zero instead fixed the delay, and it's reflected in the "impulse delay" calculation after generating the filter too. Might mean you end up with less manual trimming in REW :)
132k taps at 48kHZ sampling rate will generate 1.4 seconds long filters. Most filters in REW can be trimmed down to 32768 taps which is fast enough for any audio application. To speed up and decrease the number of taps in rePhase you need to use rectangular windows and 99% or 100% centering with moderate optimization. There should be detailed instructions in my MiniDSP video where I create filters with just 1022 taps (that's "5 ms" at 96kHZ sampling rate of MiniDSP)
Hello! as for EQ filters.... instead of making 3 different filters and then joining them I do VAR SMOOTHING that applies you 1/48 on low frequencies and psychoacoustic on high frequencies. I don't need to have boosts in any frequency and drop only the peaks. Do you think it could be okay?
Var smoothing is closest to the technique and will work but I have obtained highest clarity with the exact same settings in the tutorial including 5db boost allowance up to 200hz
@@ocaudiophile I have another question, when creating the XO filter with rephase I have to drop the gain by 3 db (in rephase) otherwise it gives me clipping..
It's normal?
Can I make the filter at -3db or is it not good?
@@BarileTixxoFilms It's normal with phase correction. You can comfortably apply a -3dB normalization in rePhase before creating the filter.
@@ocaudiophile Last question!
In the last filter (phase inversion with rew), could it be advantageous to make a range of 80hz-2000hz?
With these frequencies I can still use 3 fwd cycles.
Should I always use the 1413hz frequency in that case?
Thank you for everything you do...in my opinion you are a fucking genius!!!!
1413 was chosen for 100-1000 range, not sure if it's optimal for 80-1000, you need to try and see. Correcting phase below 100Hz has potential to cause pre-echo.