Now i wonder if this somehow can be adapted for car stereo use... with all the pressure chamber stuff happening in bass there, the very early reflections and so on. Is there any chance this would work there also, or totally different game?
Interesting question. The shape of a car is a bit too complex to derive a reasonably accurate method but could be done for a sub in the boot I guess. You will need accurate measruements and a convolution engine in the car setup.
@@ocaudiophile Raspberry Pi with CamillaDSP is in the works. But if two channel you could convolve the actual sound files with the final filter by help of ffmpeg.
@@Cathul All can be done in rephase and REW, the method is the problem. Dirac Live ART, if it ever come to cars, does something very similar to this but uses a different speaker to send the counter wave.
@@ocaudiophile yeah... i've done the EQ filters, phase correction and excess phase correction from one of your videos for my car and convolved the sound files with the resulting filter. Imaging and staging as well as impact greatly improved even in the car environment. DSP in the car is a Mosconi DSP that cannot do FIR, but as i said, you can always convolve your soundfiles offline just like AcourateNAS is doing with the help of ffmpeg. ffmpeg just uses the filter as second input and you need to use the afir filter for this. This step is very time consuming though, especially if you need to convolve several thousand soundfiles. I think i leave the VBA out for now though as i tried to align the dips and peaks from the VBA excel sheet with the raw response of the subwoofer and there was no similarity at all unfortunately. So even with guessing all the peaks and dips and trying different sizes it didn't really work. Well, result of filters even without this part is great, so thank you for your effort in providing these informations in the form of videos for us to adapt them. :)
This is guy is a beast. I have seen him in action debating with some of the top minds in the industry on a forum called Audio Science Review. Great place to get real and measureable data to use in making the best decision on what audio component is best for you. Not all that subjective bs you get from most of these "Critics" today. Witnessing players like the guy who created HQ Player, another who wrote rooextension, and the CEO of RME. All of these audio geniuses fiercly debating is something to see. Have to say OCA has no fear and knows what the hell he is talking about by displaying a high level of confidence. By the way OCA, I purchased the rme adi‑2 dac fs, not for phono use! Good buy? Keep the hits coming.
I didn't have an opportunity to listen to it but from the reviews I think it's a bargain at the price given the very high S/N ratio (sound quality) and the very capable PEQ that comes with it.
I owned the RME ADI 2. Really good piece of hardware. Excelente sound, countless features, excelente manual and great price. I sold mine and I dearly miss it.
I successfully made the VBA filter It's hard but rewarding and I'm learning a lot so thank your this you're an amazing person.there is a issue with this VBA : my measurement has first peak 36.5 and highest peak at 131 Hz and between is 1 asymmetrical dip that peaks at 95 Hz, my room modes are co-aligning better with VBA ultra (65,131,196 from your old sheet ) then with VBA revamped (from this video where modes are 13,27,40,54,68,81,95,109,122,136...) when your filter is applied to my measurement starts off exactly at first peak but then boost is not long enough and cuts where my dip is, Can I make asymmetrical LP - HP filter that co-aligns better with my measurement and for my studio monitors or can this filter be tweaked to do that. Another question is is there a difference between applying your filters ( all 7 from digital room and speaker correction workshop) for near-field studio monitoring and applying them to HI-FI/ audiophile speakers i.e. is your method tweaked for hi-fi ?
Thank you very much with this good tutorial. It was the enabler to get creative with REW and Filtering. Mabe there are a few points were i think there can be an improvement: 1. The Lowpassfilter you createt with REW in order to delay it, is not minimal phase. So the VBA is "to late" for frequencys in the transit region. I created the lowpass with RePhase (impulse is on t=0) and got better results. 2. For me an attenuation of the Lowpassfilter of -1.8dB had the best results in correcting the resonant peaks. Everything between -1dB and 0dB pushed the dips to far 3. For my room a lowpassfilter with a higher order worked better, i think it is due to my non-sqarish-room. I would prefer to do corrections for the target curve after the VBA filter (even if this needs more energy). I know that you do an overall phase correction afterwards and the two points will be corrected anyway, but you will need less harsh corrections.
You can also generate min phase version of the Dirac+lpf filter, this will improve group delay. I've also explained a quite different vba approach in the latest video if you wanna check
Nice! Going to try this this week. Would there be any chance you could post the excel file in the show notes? As your old VBA videos have slightly different formulas and tables. Also, when you have multiple subwoofers would you just do this to the L and R signals based on the mains measurements? Or would you do it to the subwoofer output as well based on the combined sub response (as my combined and EQed sub response is pretty flat with no obvious peaks and dips)? Thanks!
It's strange that REW only allow 10ms timing offset. The exceed offset needed to be add on will require segment adding. I didn't realize this and was wondering why the frequency was off so much......
i just realized you have two different methods of producing vba filter this one and using hp lp filter from your other older video, witch one is better?
Hey there, how can apply this to Subwoofers? I have to Big Woofers in my studio and i want to apply VBA to them. Also im actually not shure how to aply the house curve (Harman) to a non Full Range Speaker system. Is it possible to get in touch with you and you can help me? Its my very first project i would also Tip you apropiately!
This version seams to be the best!! But with which other filter from your videos should someone combine it to have the proper result for a stereo pair speaker setup.
Add a crossover phase correction filter created in rephase and a rew auto EQ filter between 20-500Hz based on 1/48 smoothing. Multiply all 3. You'll be golden.
@@ocaudiophile I have managed to complete the VBA filter and its Great!! One question though for the other two that you propose because you have a lot of versions and i have lost it😂. When you say phase correction i will use the mixed phase correction video this is the method u suggest? and i any points for the rew auto filter between 20-500hz i dont seem to find an example. My goal is to achieve detail listening and good translation on my speakers. Thank you very much.
All methods and totorials will get descent results but I keep improving them and REW keeps adding new features hence the multiple number of videos. 20-500hz 1/48 smoothing method is not yet in any of the videos. I am planning to make one this week.
@@ocaudiophile🙏Thank you for your response. Glad to hear that you planning a video for this, it will be very interesting! Yes i agree REW is getting better and it will solve a lot of problems in the future. Have a nice day and as a fun i will follow you new updates!!! Take Care💫🙏
Dear Serkan, Thank you for all the guides you have published on TH-cam. I learned a lot from your effort. I have seen no other person around digging deep as you do and willing to invest time and energy to share your journey with others. Is this way of creating the VBA filter applicable to the speakers response only ? Or it may be applied the same way for the Subwoofer(s) plus speaker response ?
Very interesting!! Will try this. Have a question though (pardon my limited knowledge) - In the acoustic domain, don't we need a second copy of the same signal to create this cancellation? If we apply it on the main signal at the source using a convolution engine (with the phase reversal for modal frequencies), won't the wall reflections that creates standing wave also have the same phase shift?
Think I got it! It is like creating a custom comb filter on the source signal so that even with standing wave nodes/anti-nodes, the energy in those areas becomes equal to other frequencies, right ?
First of, thank you for your great videos! Can I integrate this when trying to calibrate the speakers with audyssey with your other instructional video?
Hi OCA, great channel, thanks for your hard work! What confuses me is why you use t=29.92ms for the impulse peak and not t=59.84ms (which is 2*29.92ms)? Because in a one-dimensional, 29.92 ms is the time the sound wave takes to travel to the back wall (which would be 10.26m away) . If we now send out the inverted response at that time, then we will meet mode 1 (frequency (16.70hz) halfway on its way back through the room and not cancel it... So we only cancel out modes 2,4,6,8, etc... This is also where the dips in the filter are located. But with t=59.84ms, the inverted response is sent out, when the sound wave is back at the speaker, and would cancel out all modes (1,2,3,4,etc)?
@@ocaudiophile Thanks for your answer. I forgot to write that I made a one-dimensional 'simplification' of the room. The idea behind this is, that one can simplify the three-dimensional room to a one-dimensional example with room length 10.26m. This is possible because sound in low-frequencies is near omnidirectional. So the sum of all three room dimensions added is 11.83 meters per your example. If we subtract 15% damping, we receive approximately 10m meters (The theoretical room length). I took for my example 10.26m, because that is the wavelength of your resonance frequency of 16.70725 hz. This wave is back at the speaker in 59.84ms, so that is when the canceling should happen. If you look at the filter, it gets maybe more obvious. At time=12m30s in the video, we see that the first dip is at 33.4hz, but it should be at 16.7hz. Similarly, dips are missing at all uneven mode: mode 3 = 50.1hz, mode 5 = 83.5hz, etc... I hope I could express myself a bit better now and wish you good luck for the future :)
The room modes are produced from a complex combination of all 3 room dimensions. Room length is just an approximation. If you average all 3 room dimensions and deduct for furniture from that about 5%, the room mode dimension to be used in my example is something like 3.72m. This has about 46Hz resonant frequency and first peak at 92Hz and the polarity inverted pulse with about -10.8ms delay (-1000/46Hz) will successfully correct first major dip and peak. You will not need to add any SPL offsets, etc. to that filter either. The example in the video deals with a more local dip and peak but still achieves adequate results.@@commonNightingale
@@ocaudiophile Hi OCA So I plugged the 3.72 m for all three room dimensions into REWs Room Simulator, and it creates a first major peak at 46.1hz (modes 1-0-0, 0-1-0 and 0-0-1), a second major peak at 92.2hz (modes 2-0-0, 0-2-0 and 0-0-2), etc... If I use a time delay of -21.6ms the filter gets dips at 46.1hz, at 92.2hz, etc., so at all multiples of modes 1-0-0, 0-1-0 and 0-0-1. If I use a time delay of -10.8ms the filter gets dips at 92.2hz, at 184.4hz, etc., so at all multiples of modes 2-0-0, 0-2-0 and 0-0-2. I agree with the (-1000/46Hz) that you write in your comment, but that results in -21.6ms, not -10.8ms. But in your excel-file, you use the formula -500/frequency instead of -1000/frequency. Sadly, I cannot upload the screenshot of the simulation and filters here...
I'm now integrate linear phase crossover and DRC into my filter now. Whole DSP XO, thanks this channel so much! It's a bit pity my room shape is too strange and small for VBA to function. (The dips and peak cannot align well in any mode, seems different modes are influencing each other)
Comparing to the previous vba method you described in part 2 with the use of rephase, this one seems to influence the deeps and peaks with a stronger magnitude, but this one creates a nasty deep in the 100Hz area. My room is aroud the same dimentions as yours and after applying the filter using this method I am getting the same deep as it seems on the resulting predicted response on the video. Can you tell what is the reason for the 100Hz behaviour of this filter ? Maybe it is preferable to use the previous method ?
If your room resonant frequency is say 16Hz, the VBA filter will have dips at 2x16, 4x16 and 6x16 =96 Hz. But the effect starts decreasing around 5x16 = 80Hz because of the lpf. The dip at 96Hz is in a very narrow band and is inaudible but you can start the lpf roll off earlier if you don't like its effect on your response.
Hello Mr Serkan, I always follow your excellent works trying to apply them to my system. I use a yamaha avr feeded with JRiver through hdmi using 4 channell straight PCM , L/R and 2 sub ( the second created using the CENTER channel) and so using 4 filters created following your precious indications. Lately i'm having problems trying to synch L/R with SUBS. Till now i used the delay funtion of jriver to put in synch with according to the numbers found in the info tab of REW at the line called "TIME AXIS START", but it seems no longer works. In fact i have 2 filter for subs that are very short (VBA) and the filters for L/R that are quite longer (even if i trimmed them) Could you please show me if there is something more valid than my ears to trust on, trying to adjust? Thank you very much.
Thank you. Measure all four channels with REW with all the filters on, you may need to play a sweep from JRiver to do that. Make a copy of each measurement. Then in ALL SPL tab, with only the copies ticked, "Align IR starts" of all copies to each other. FInally check in Overlays/Impulse tab the distance/time delay (with CTRL+ Right mouse click method) between the impulse peaks of the originals and the copies and adjust delays for each channel accordingly in JRiver DSP. Align IR starts may sometimes get it wrong in the first attempt, you can also try "time align"
Hi OCA! I have a question about this method: how to do it in a room with one side open, roof slopes, and a back wall (behind LP) with the ins and outs? Thanks in advance.
Would you be able to implement this using the FIR tab on a MiniDSP? Or should I just rely on DIRAC in my receiver to get it right? And would it work for subs as well?
It will not be easy. REW cannot export FIR filters in .bin format. rePhase can but you cannot create time delayed filters in rePhase. It should be possible to convert .txt or .wav files to .bin but I've never done it. Dirac Live cannot do such filters. Dirac ART can but there's no MiniDSP that can process Dirac ART (for now!)
Hi OCA, I have made one negative impulse file at the 6th identified impulse frequency. Do we then repeat the procedure for most other room anomalies from say above that frequency till say 600hz? Or would the single filter be enough to negate the other reflective frequencies which would also be present.
Sorry, I am not sure I understood the question. What do you mean by negative impulse? If you mean room mode peaks and dips, a low pass filter starting at the 5th frequency with 18dB/octave slope gives good results in many people's systems.
I guess calculating the volume of the room and using its cubic root as 3 equal room dimensions (say x) in the resonant frequency formula would be a logical approach: f=c/(6x) You can then fine tune it depending on the convolution result.
@Obsessive Compulsive Audiophile Thank you for that. I also have two open doorways in the corners of the front wall opening into another large room. So nothing is perfect. I will have a go following your tutorial. Thanks again!
My room is rather strange, the restroom occupied a portion of the cuboid. What length should I use? Or is it better I use the previous way to look at the peaks and dips in my low frequency?
This video is for stereo hifi systems or for HTPC owners who can use convolution engines in their equipment. Currently, there's no surround receiver that lets user to input FIR filters. You should follow Audyssey videos in this channel for the best possible calibration of your receiver.
Hallo OCA, ich steige bald nicht mehr durch😀 Vielen lieben Dank für all deine Arbeit und das Teilen! Ich hatte erst vor kurzem, nach vorherigen Videos von Dir, eine manuelle Kalibrierung durchgeführt mit einem alten Marantz SR7005. Nun habe ich seid ein paar Tagen einen Yamaha A6a mit YPAO. Nach der automatischen Einmessung klingt es schon nicht schlecht, ich meine sogar besser wie die manuelle vorher. Welche endgültige Vorgehensweise kannst Du mir empfehlen aktuell noch ohne Subwoofer (wird sich noch ändern, denke an einen Canton Sub 500 nach)? Umik 1 ist vorhanden. - Dieses Vorgehen macht das Vorgehen in den Videos Top Tricks 1 und 2 mit Rephase überflüssig? - Wende ich dies auf alle Lautsprecher an in meinem Fall 5.0 und nur für die unteren Frequenzen bis 200 oder 300 Hz? Ich verstehe noch nicht wie ich den erzeugten Filter nun anwenden kann! Muss ich daraus in REW die EQ Filter mit den mir zur Verfügung stehenden Bändern berechnen lassen? Ps.: wo in D lebst du?
Ich wohne in Frankfurt. Leider können die FIR-Filter nicht in Surround-Receivern verwendet werden. Selbst mit einem MiniDSP wird die verfügbare Anzahl an Filterabgriffen nicht ausreichen. Sie können die Methode im Video „Supreme Audyssey Calibration“ anwenden, wenn Sie die MultEQ Editor-App haben. Wenn nicht, können Sie „Manuelle Audyssey-Kalibrierung“-Videos verwenden und grafische Equalizer verwenden. Sie funktionieren alle ohne Sub, da sich der Receiver auf 5,0 einstellt und den Bass zu den Frontlautsprechern leitet. Überspringen Sie einfach die Teile in den Videos über Subwoofer.
@@ocaudiophile Dann mal Grüße ausm Pott! wie gesagt habe ich jetzt YPAO und kein Audyssey! Dann werde ich es wohl so machen müssen wie mit dem Marantz und den genannten Videos. Aber wie werden denn die FIR Filter genutzt wenn sie nirgends in einem Gerät Anwendung finden? Da fällt mir noch etwas merkwürdiges ein: wenn ich den Impuls einstelle habe ich den ein oder anderen Lautsprecher, der trotz richtiger Polung, die 100% in die andere Richtung hat wie der Rest. Manchmal ändert sich dies sogar wenn ich öfters messe und ggf die Messzeit verlängere, so lässt sich dann schwer arbeiten. Dies hatte ich mit dem alten und auch dem neuen AVR. Hast Du da eine Idee wie ich damit umgehen soll?
Mir ist später noch ein Programm für den PC (Equalizer APO) eingefallen, welches ich vor ein paar Wochen entdeckte, habe mich jedoch noch nicht damit befasst. Hiermit wäre die Nutzung der Filter möglich so wie ich sehe. Kann ich denn so auch per HDMI ein 5.2.4 System nutzen?
@@ocaudiophile Ok, irgend einen Haken muss es ja geben... Womit nutzt Du die Filter im Surroundbetrieb wenn es mehr wie 7 Kanäle sind? Hast Du evt. noch eine Idee zu meinem Phänomen welches ich gestern schrieb?
Your videos are quite interesting but, because I am a complete N00B, I have to admit that I feel like I'm trying to drink from a firehose! (Thanks for your playlist "Applied DSP - I'm going through that now, as a starting point.) If you have any other suggestions for understanding the basics, I would greatly appreciate it! That said, I've recently ordered my first dsp product (MiniDSP Flex), and I am really interested in learning to use it along with the recommended software to optimize my 2.1 stereo setup. After watching a couple of your videos, I am concerned that I may not be able to get very accurate with some of what you are doing because my setup is in a room that is in an open(ish) floorplan, with a semi-open kitchen partially behind my listening position on the left, and a dinette area on the left - a half-wall with scalloped openings into a dining room and hallway to other parts of the house on the right of my listening area. I'm thinking it's basically going to be impossible for me to calculate my room's resonant frequency because of the layout. Am I wrong about this? Any particular suggestions as to how to make the best out of this situation?
You don't really need to go beyond basic vector maths to be able to create descent corrections for your system but of course it will help you understand what you're doing and maybe create your own methods. Flex is very popular among the DIY crowd who know what they're doing but I don't have one myself. It really depends on your setup at the end of the day. Are you streaming, playing from a computer or use an analogue source. I understand all irregularities in your room are behind the LP which is similar to my room. As long as you've a relatively symmetric front/side wall organisation, VBA should work. You can use the total volume of the room to find pseudo dimensions.
@@ocaudiophile I was trying to link to a sketch of my room layout to give you an idea of what I'm dealing with, but apparently my comments won't post with a link... My setup currently consists of mostly streaming from a WIIM Mini, and my television, with an occasional session from my laptop via USB, but that is rare. There is also a 3H'X10L' half wall with 2 large scalloped openings above it on the right side of my system. That opens to a 10'X18' dining area and a hallway that goes to the front of my house. I imagine this will have to be accounted for, but I'm not sure how much, considering that the half wall has got to be reflecting a fair amount of the sound. My sub actually faces that half wall currently because that's one of the only positions I can logistically make it work. Anyway, I hope what I'm asking makes sense, and that there's a reasonable path forward in optimizing my system with DSP.
@@ocaudiophile thank you so much as always for your quick response. I'm on a 5.1.4 setup using Marantz Cinema 50. Would you say following the Supreme Audyssey Guide and this video is a good starting point for full manual calibration?
@@Judddernaut Tutorials about FIR filter generation for stereo systems are not compatible with multichannel home theatre receivers. They need a processor (a PC/Mac or to a lesser extent a MiniDSP somewhere in the chain). Btw, a Marantz Cinema 50 is perfectly capable of processing FIR filters and in fact that's how Audyssey/Dirac work but they are not open to users. There's no brand I know of which lets you use your own FIR filters.
@@mikeortiz2139 If you wanna stick with Audyssey, start with "Supreme Audyssey Calibration" video and make sure you also watch "Audyssey for Atmos Music" tutorial. If you want to calibrate your system manually (but you'll lose Dynamic EQ/volume) start with "Dolby Atmos Height Channels... and Dolby Atmos Manual Tunin Complimentary" videos. If you don't want to buy a Umik mic, there's also a link to a calibration file to use with the standard Audyssey microphone in REW in the decription of the "hidden Audyssey mic calibration" video.
I like your idea for VBA very much. However, you can still improve things. First of all, you should never boost the delayed low pass filter waveform. It should correspond with a reflection which will never be stronger then the direct signal. A value of -6dB is good to start with. Further, use an 8th order Bessel low pass filter (LPF) at 100Hz. It will give you an almost perfect and symmetrical impuls respons. And it will correspond nicely with the Harman Kardon curve still. The delay as given in the IR window button will be 4.97ms. Then, no, the total of length, width and heigth is not the correct number. At all. Also, you make a mistake regarding your open room. That would mean the frequency dip would be lower than calculated, not higher. To cut things short, just use twice the room length as reference for the LPF filter delay (with its own delay of 4.97ms subtracted from the needed delay and then put into the time offset of the LPF impuls respons). This gives me an almost perfect result, on the condition that the 8th order Bessel LPF at 100Hz is being used.
Thank you. I will try these suggestions. Recently I am doing it quite differently though (no boost to the lpf and it targets the second peak) but I am reluctant to put anoher VBA video. People don't like when you update methods ;) Here's the mdat if you feel like reverse engineering it: drive.google.com/file/d/1cdY1icKBdYpst1SSs7TdILoLP-cSrdpO/view?usp=sharing
@@jeroendezaire7266 using minimum phase copy of the smoothed vba filter trick sometimes improves the results. Out of all the operations in REW, generating min phase with replicate data outside range option ticked is the only one that'll freeze the smoothing used.
@@ocaudiophile I like the minimum phase copy of the VBA filter, it gives a perfect impuls and step respons with no ringing at all. But I do not understand smoothing, that would ruin the now perfect VBA filter.
Now i wonder if this somehow can be adapted for car stereo use... with all the pressure chamber stuff happening in bass there, the very early reflections and so on.
Is there any chance this would work there also, or totally different game?
Interesting question. The shape of a car is a bit too complex to derive a reasonably accurate method but could be done for a sub in the boot I guess. You will need accurate measruements and a convolution engine in the car setup.
@@ocaudiophile Raspberry Pi with CamillaDSP is in the works.
But if two channel you could convolve the actual sound files with the final filter by help of ffmpeg.
@@Cathul All can be done in rephase and REW, the method is the problem. Dirac Live ART, if it ever come to cars, does something very similar to this but uses a different speaker to send the counter wave.
@@ocaudiophile yeah... i've done the EQ filters, phase correction and excess phase correction from one of your videos for my car and convolved the sound files with the resulting filter. Imaging and staging as well as impact greatly improved even in the car environment. DSP in the car is a Mosconi DSP that cannot do FIR, but as i said, you can always convolve your soundfiles offline just like AcourateNAS is doing with the help of ffmpeg.
ffmpeg just uses the filter as second input and you need to use the afir filter for this. This step is very time consuming though, especially if you need to convolve several thousand soundfiles.
I think i leave the VBA out for now though as i tried to align the dips and peaks from the VBA excel sheet with the raw response of the subwoofer and there was no similarity at all unfortunately.
So even with guessing all the peaks and dips and trying different sizes it didn't really work.
Well, result of filters even without this part is great, so thank you for your effort in providing these informations in the form of videos for us to adapt them. :)
Good to hear that!
Serkan is the is the Dumbeldore of audio calibration ❤ mad respect.
😂
I learned so much because of your videos, what a gem of a channel!
Great to hear!
This is guy is a beast. I have seen him in action debating with some of the top minds in the industry on a forum called Audio Science Review. Great place to get real and measureable data to use in making the best decision on what audio component is best for you. Not all that subjective bs you get from most of these "Critics" today. Witnessing players like the guy who created HQ Player, another who wrote rooextension, and the CEO of RME. All of these audio geniuses fiercly debating is something to see. Have to say OCA has no fear and knows what the hell he is talking about by displaying a high level of confidence. By the way OCA, I purchased the rme adi‑2 dac fs, not for phono use! Good buy? Keep the hits coming.
I didn't have an opportunity to listen to it but from the reviews I think it's a bargain at the price given the very high S/N ratio (sound quality) and the very capable PEQ that comes with it.
🙏
I owned the RME ADI 2. Really good piece of hardware. Excelente sound, countless features, excelente manual and great price. I sold mine and I dearly miss it.
I successfully made the VBA filter It's hard but rewarding and I'm learning a lot so thank your this you're an amazing person.there is a issue with this VBA : my measurement has first peak 36.5 and highest peak at 131 Hz and between is 1 asymmetrical dip that peaks at 95 Hz, my room modes are co-aligning better with VBA ultra (65,131,196 from your old sheet ) then with VBA revamped (from this video where modes are 13,27,40,54,68,81,95,109,122,136...) when your filter is applied to my measurement starts off exactly at first peak but then boost is not long enough and cuts where my dip is, Can I make asymmetrical LP - HP filter that co-aligns better with my measurement and for my studio monitors or can this filter be tweaked to do that. Another question is is there a difference between applying your filters ( all 7 from digital room and speaker correction workshop) for near-field studio monitoring and applying them to HI-FI/ audiophile speakers i.e. is your method tweaked for hi-fi ?
Thank you very much with this good tutorial. It was the enabler to get creative with REW and Filtering. Mabe there are a few points were i think there can be an improvement:
1. The Lowpassfilter you createt with REW in order to delay it, is not minimal phase. So the VBA is "to late" for frequencys in the transit region. I created the lowpass with RePhase (impulse is on t=0) and got better results.
2. For me an attenuation of the Lowpassfilter of -1.8dB had the best results in correcting the resonant peaks. Everything between -1dB and 0dB pushed the dips to far
3. For my room a lowpassfilter with a higher order worked better, i think it is due to my non-sqarish-room. I would prefer to do corrections for the target curve after the VBA filter (even if this needs more energy).
I know that you do an overall phase correction afterwards and the two points will be corrected anyway, but you will need less harsh corrections.
You can also generate min phase version of the Dirac+lpf filter, this will improve group delay.
I've also explained a quite different vba approach in the latest video if you wanna check
Nice! Going to try this this week. Would there be any chance you could post the excel file in the show notes? As your old VBA videos have slightly different formulas and tables.
Also, when you have multiple subwoofers would you just do this to the L and R signals based on the mains measurements? Or would you do it to the subwoofer output as well based on the combined sub response (as my combined and EQed sub response is pretty flat with no obvious peaks and dips)? Thanks!
Done!
It's strange that REW only allow 10ms timing offset. The exceed offset needed to be add on will require segment adding. I didn't realize this and was wondering why the frequency was off so much......
i just realized you have two different methods of producing vba filter this one and using hp lp filter from your other older video, witch one is better?
latest one ;)
Hey there, how can apply this to Subwoofers? I have to Big Woofers in my studio and i want to apply VBA to them.
Also im actually not shure how to aply the house curve (Harman) to a non Full Range Speaker system.
Is it possible to get in touch with you and you can help me? Its my very first project i would also Tip you apropiately!
This version seams to be the best!! But with which other filter from your videos should someone combine it to have the proper result for a stereo pair speaker setup.
Add a crossover phase correction filter created in rephase and a rew auto EQ filter between 20-500Hz based on 1/48 smoothing. Multiply all 3. You'll be golden.
@@ocaudiophile You are Gold, thank you for sharing so valuable information. Huge respect 🙏🙏🙏💯💯💯
@@ocaudiophile I have managed to complete the VBA filter and its Great!!
One question though for the other two that you propose because you have a lot of versions and i have lost it😂.
When you say phase correction i will use the mixed phase correction video this is the method u suggest? and i any points for the rew auto filter between 20-500hz i dont seem to find an example.
My goal is to achieve detail listening and good translation on my speakers.
Thank you very much.
All methods and totorials will get descent results but I keep improving them and REW keeps adding new features hence the multiple number of videos. 20-500hz 1/48 smoothing method is not yet in any of the videos. I am planning to make one this week.
@@ocaudiophile🙏Thank you for your response. Glad to hear that you planning a video for this, it will be very interesting!
Yes i agree REW is getting better and it will solve a lot of problems in the future.
Have a nice day and as a fun i will follow you new updates!!! Take Care💫🙏
Dear Serkan,
Thank you for all the guides you have published on TH-cam. I learned a lot from your effort. I have seen no other person around digging deep as you do and willing to invest time and energy to share your journey with others.
Is this way of creating the VBA filter applicable to the speakers response only ? Or it may be applied the same way for the Subwoofer(s) plus speaker response ?
You can apply VBA to subwoofers just the same.
Very interesting!! Will try this. Have a question though (pardon my limited knowledge) - In the acoustic domain, don't we need a second copy of the same signal to create this cancellation? If we apply it on the main signal at the source using a convolution engine (with the phase reversal for modal frequencies), won't the wall reflections that creates standing wave also have the same phase shift?
Think I got it! It is like creating a custom comb filter on the source signal so that even with standing wave nodes/anti-nodes, the energy in those areas becomes equal to other frequencies, right ?
The polarity inverted cancellation signal is produced with the precise delay.
@@ocaudiophile thank you ! BTW Are you using the paid version of REW ? Impulse response graph looks different from what I have on the free version.
@tkmanutk it's the free "early access version", you can find it with Google search.
First of, thank you for your great videos! Can I integrate this when trying to calibrate the speakers with audyssey with your other instructional video?
Audyssey filters are not capable of such complicated filters but it can be achieved with a higher-end MiniDSP unit connected to subwoofers.
@@ocaudiophile ok Thanks. Good to know!
Hi OCA, great channel, thanks for your hard work!
What confuses me is why you use t=29.92ms for the impulse peak and not t=59.84ms (which is 2*29.92ms)?
Because in a one-dimensional, 29.92 ms is the time the sound wave takes to travel to the back wall (which would be 10.26m away) . If we now send out the inverted response at that time, then we will meet mode 1 (frequency (16.70hz) halfway on its way back through the room and not cancel it... So we only cancel out modes 2,4,6,8, etc... This is also where the dips in the filter are located.
But with t=59.84ms, the inverted response is sent out, when the sound wave is back at the speaker, and would cancel out all modes (1,2,3,4,etc)?
The wall is about 5m away though!
@@ocaudiophile
Thanks for your answer. I forgot to write that I made a one-dimensional 'simplification' of the room.
The idea behind this is, that one can simplify the three-dimensional room to a one-dimensional example with room length 10.26m.
This is possible because sound in low-frequencies is near omnidirectional.
So the sum of all three room dimensions added is 11.83 meters per your example. If we subtract 15% damping, we receive approximately 10m meters (The theoretical room length). I took for my example 10.26m, because that is the wavelength of your resonance frequency of 16.70725 hz.
This wave is back at the speaker in 59.84ms, so that is when the canceling should happen.
If you look at the filter, it gets maybe more obvious. At time=12m30s in the video, we see that the first dip is at 33.4hz, but it should be at 16.7hz. Similarly, dips are missing at all uneven mode: mode 3 = 50.1hz, mode 5 = 83.5hz, etc...
I hope I could express myself a bit better now and wish you good luck for the future :)
The room modes are produced from a complex combination of all 3 room dimensions. Room length is just an approximation. If you average all 3 room dimensions and deduct for furniture from that about 5%, the room mode dimension to be used in my example is something like 3.72m. This has about 46Hz resonant frequency and first peak at 92Hz and the polarity inverted pulse with about -10.8ms delay (-1000/46Hz) will successfully correct first major dip and peak. You will not need to add any SPL offsets, etc. to that filter either. The example in the video deals with a more local dip and peak but still achieves adequate results.@@commonNightingale
@@ocaudiophile
Hi OCA
So I plugged the 3.72 m for all three room dimensions into REWs Room Simulator, and it creates a first major peak at 46.1hz (modes 1-0-0, 0-1-0 and 0-0-1), a second major peak at 92.2hz (modes 2-0-0, 0-2-0 and 0-0-2), etc...
If I use a time delay of -21.6ms the filter gets dips at 46.1hz, at 92.2hz, etc., so at all multiples of modes 1-0-0, 0-1-0 and 0-0-1.
If I use a time delay of -10.8ms the filter gets dips at 92.2hz, at 184.4hz, etc., so at all multiples of modes 2-0-0, 0-2-0 and 0-0-2.
I agree with the (-1000/46Hz) that you write in your comment, but that results in -21.6ms, not -10.8ms. But in your excel-file, you use the formula -500/frequency instead of -1000/frequency. Sadly, I cannot upload the screenshot of the simulation and filters here...
You are right! The Excel formula in the video link is -500/f for some reason. I will update it soon as I can.@@commonNightingale
I'm now integrate linear phase crossover and DRC into my filter now. Whole DSP XO, thanks this channel so much! It's a bit pity my room shape is too strange and small for VBA to function. (The dips and peak cannot align well in any mode, seems different modes are influencing each other)
Great to hear!
Comparing to the previous vba method you described in part 2 with the use of rephase, this one seems to influence the deeps and peaks with a stronger magnitude, but this one creates a nasty deep in the 100Hz area. My room is aroud the same dimentions as yours and after applying the filter using this method I am getting the same deep as it seems on the resulting predicted response on the video. Can you tell what is the reason for the 100Hz behaviour of this filter ? Maybe it is preferable to use the previous method ?
If your room resonant frequency is say 16Hz, the VBA filter will have dips at 2x16, 4x16 and 6x16 =96 Hz. But the effect starts decreasing around 5x16 = 80Hz because of the lpf. The dip at 96Hz is in a very narrow band and is inaudible but you can start the lpf roll off earlier if you don't like its effect on your response.
Hello Mr Serkan,
I always follow your excellent works trying to apply them to my system.
I use a yamaha avr feeded with JRiver through hdmi using 4 channell straight PCM , L/R and 2 sub ( the second created using the CENTER channel)
and so using 4 filters created following your precious indications. Lately i'm having problems trying to synch L/R with SUBS.
Till now i used the delay funtion of jriver to put in synch with according to the numbers found in the info tab of REW at the line called
"TIME AXIS START", but it seems no longer works.
In fact i have 2 filter for subs that are very short (VBA) and the filters for L/R that are quite longer (even if i trimmed them)
Could you please show me if there is something more valid than my ears to trust on, trying to adjust?
Thank you very much.
Thank you.
Measure all four channels with REW with all the filters on, you may need to play a sweep from JRiver to do that. Make a copy of each measurement. Then in ALL SPL tab, with only the copies ticked, "Align IR starts" of all copies to each other. FInally check in Overlays/Impulse tab the distance/time delay (with CTRL+ Right mouse click method) between the impulse peaks of the originals and the copies and adjust delays for each channel accordingly in JRiver DSP. Align IR starts may sometimes get it wrong in the first attempt, you can also try "time align"
Hi OCA! I have a question about this method: how to do it in a room with one side open, roof slopes, and a back wall (behind LP) with the ins and outs? Thanks in advance.
İf you have common first peaks for left and right speaker, use that frequency and ignore the room dimensions.
@@ocaudiophile thanks! I’ll have a look.
Would you be able to implement this using the FIR tab on a MiniDSP? Or should I just rely on DIRAC in my receiver to get it right? And would it work for subs as well?
It will not be easy. REW cannot export FIR filters in .bin format. rePhase can but you cannot create time delayed filters in rePhase. It should be possible to convert .txt or .wav files to .bin but I've never done it. Dirac Live cannot do such filters. Dirac ART can but there's no MiniDSP that can process Dirac ART (for now!)
Hi OCA, I have made one negative impulse file at the 6th identified impulse frequency. Do we then repeat the procedure for most other room anomalies from say above that frequency till say 600hz? Or would the single filter be enough to negate the other reflective frequencies which would also be present.
Sorry, I am not sure I understood the question. What do you mean by negative impulse? If you mean room mode peaks and dips, a low pass filter starting at the 5th frequency with 18dB/octave slope gives good results in many people's systems.
@@ocaudiophile Thanks
Does this filter get placed in a minidsp 2x4 hd or how do input this? I'm thinking it can be used on the input side of the minidsp.
No, 1024 taps is simply not enough for this kind of sophisticated filters.
I have a room with a sloping ceiling,so I have 4 dimensions. What is the best way to approximate the three dimensions you refer to at about 1 min.
I guess calculating the volume of the room and using its cubic root as 3 equal room dimensions (say x) in the resonant frequency formula would be a logical approach: f=c/(6x) You can then fine tune it depending on the convolution result.
@Obsessive Compulsive Audiophile Thank you for that. I also have two open doorways in the corners of the front wall opening into another large room. So nothing is perfect. I will have a go following your tutorial. Thanks again!
Is it a problem if I apply a gain of more than 4 db to the lpf filter to align it well with my curve???
It's dangerous?
Thank you!!
No it's fine
@@ocaudiophile 6-8 db??? It’s fine??
Such differences are not common but if that's what is calculated, it'll not have any harm on the speaker.
@@ocaudiophile i love you ❤️
My room is rather strange, the restroom occupied a portion of the cuboid. What length should I use? Or is it better I use the previous way to look at the peaks and dips in my low frequency?
Ro dimensions will only give you a ballpark figure anyway. Look for the largest and first peaks common in both speakers and all mic locations.
After all REW editting and correction for Subwoofer and Speakers how to import or make it effective it in to AVR? Still unable to understand.
This video is for stereo hifi systems or for HTPC owners who can use convolution engines in their equipment. Currently, there's no surround receiver that lets user to input FIR filters. You should follow Audyssey videos in this channel for the best possible calibration of your receiver.
Is there any way to calibrating speakers and Subwoofer with REW and to import settings or that calibration in my denon x4700h
@@ocaudiophilethanks for reply
@@ocaudiophile so can you make a .wav file and use EAPO?
@@l.s.1709 sure you can
Hallo OCA,
ich steige bald nicht mehr durch😀 Vielen lieben Dank für all deine Arbeit und das Teilen!
Ich hatte erst vor kurzem, nach vorherigen Videos von Dir, eine manuelle Kalibrierung durchgeführt mit einem alten Marantz SR7005.
Nun habe ich seid ein paar Tagen einen Yamaha A6a mit YPAO. Nach der automatischen Einmessung klingt es schon nicht schlecht, ich meine sogar besser wie die manuelle vorher.
Welche endgültige Vorgehensweise kannst Du mir empfehlen aktuell noch ohne Subwoofer (wird sich noch ändern, denke an einen Canton Sub 500 nach)? Umik 1 ist vorhanden.
- Dieses Vorgehen macht das Vorgehen in den Videos Top Tricks 1 und 2 mit Rephase überflüssig?
- Wende ich dies auf alle Lautsprecher an in meinem Fall 5.0 und nur für die unteren Frequenzen bis 200 oder 300 Hz?
Ich verstehe noch nicht wie ich den erzeugten Filter nun anwenden kann!
Muss ich daraus in REW die EQ Filter mit den mir zur Verfügung stehenden Bändern berechnen lassen?
Ps.: wo in D lebst du?
Ich wohne in Frankfurt. Leider können die FIR-Filter nicht in Surround-Receivern verwendet werden. Selbst mit einem MiniDSP wird die verfügbare Anzahl an Filterabgriffen nicht ausreichen. Sie können die Methode im Video „Supreme Audyssey Calibration“ anwenden, wenn Sie die MultEQ Editor-App haben. Wenn nicht, können Sie „Manuelle Audyssey-Kalibrierung“-Videos verwenden und grafische Equalizer verwenden. Sie funktionieren alle ohne Sub, da sich der Receiver auf 5,0 einstellt und den Bass zu den Frontlautsprechern leitet. Überspringen Sie einfach die Teile in den Videos über Subwoofer.
@@ocaudiophile Dann mal Grüße ausm Pott!
wie gesagt habe ich jetzt YPAO und kein Audyssey!
Dann werde ich es wohl so machen müssen wie mit dem Marantz und den genannten Videos.
Aber wie werden denn die FIR Filter genutzt wenn sie nirgends in einem Gerät Anwendung finden?
Da fällt mir noch etwas merkwürdiges ein:
wenn ich den Impuls einstelle habe ich den ein oder anderen Lautsprecher, der trotz richtiger Polung, die 100% in die andere Richtung hat wie der Rest. Manchmal ändert sich dies sogar wenn ich öfters messe und ggf die Messzeit verlängere, so lässt sich dann schwer arbeiten. Dies hatte ich mit dem alten und auch dem neuen AVR.
Hast Du da eine Idee wie ich damit umgehen soll?
Mir ist später noch ein Programm für den PC (Equalizer APO) eingefallen, welches ich vor ein paar Wochen entdeckte, habe mich jedoch noch nicht damit befasst.
Hiermit wäre die Nutzung der Filter möglich so wie ich sehe. Kann ich denn so auch per HDMI ein 5.2.4 System nutzen?
APO kann perfekte FIR-Abgriffe ausführen, ist jedoch auf 7.1-Kanäle beschränkt. Ich glaube nicht, dass es für Atmos-Kanäle geeignet ist
@@ocaudiophile Ok, irgend einen Haken muss es ja geben...
Womit nutzt Du die Filter im Surroundbetrieb wenn es mehr wie 7 Kanäle sind?
Hast Du evt. noch eine Idee zu meinem Phänomen welches ich gestern schrieb?
any update on the ‘RTA’ tutorial? Thanks
The high frequency response is too low with RTA, I wasn't able to obtain a useful measurement :(
@@ocaudiophile I see, I was afraid of that
Your videos are quite interesting but, because I am a complete N00B, I have to admit that I feel like I'm trying to drink from a firehose! (Thanks for your playlist "Applied DSP - I'm going through that now, as a starting point.) If you have any other suggestions for understanding the basics, I would greatly appreciate it!
That said, I've recently ordered my first dsp product (MiniDSP Flex), and I am really interested in learning to use it along with the recommended software to optimize my 2.1 stereo setup.
After watching a couple of your videos, I am concerned that I may not be able to get very accurate with some of what you are doing because my setup is in a room that is in an open(ish) floorplan, with a semi-open kitchen partially behind my listening position on the left, and a dinette area on the left - a half-wall with scalloped openings into a dining room and hallway to other parts of the house on the right of my listening area. I'm thinking it's basically going to be impossible for me to calculate my room's resonant frequency because of the layout. Am I wrong about this? Any particular suggestions as to how to make the best out of this situation?
You don't really need to go beyond basic vector maths to be able to create descent corrections for your system but of course it will help you understand what you're doing and maybe create your own methods. Flex is very popular among the DIY crowd who know what they're doing but I don't have one myself. It really depends on your setup at the end of the day. Are you streaming, playing from a computer or use an analogue source.
I understand all irregularities in your room are behind the LP which is similar to my room. As long as you've a relatively symmetric front/side wall organisation, VBA should work. You can use the total volume of the room to find pseudo dimensions.
@@ocaudiophile Hmm - my last comment got deleted for some reason - I'll try again.
@@ocaudiophile I was trying to link to a sketch of my room layout to give you an idea of what I'm dealing with, but apparently my comments won't post with a link...
My setup currently consists of mostly streaming from a WIIM Mini, and my television, with an occasional session from my laptop via USB, but that is rare.
There is also a 3H'X10L' half wall with 2 large scalloped openings above it on the right side of my system. That opens to a 10'X18' dining area and a hallway that goes to the front of my house. I imagine this will have to be accounted for, but I'm not sure how much, considering that the half wall has got to be reflecting a fair amount of the sound. My sub actually faces that half wall currently because that's one of the only positions I can logistically make it work.
Anyway, I hope what I'm asking makes sense, and that there's a reasonable path forward in optimizing my system with DSP.
You can only post google drive links in youtube comments
@@ocaudiophile -
Asking in laymen terms, but is the takeaway of this video that acoustic room treatment is "unnecessary" after following this video?
Acoustic treatment is always better but for these very low frequencies the depth of the traps required will be too large to be physically viable.
@@ocaudiophile thank you so much as always for your quick response. I'm on a 5.1.4 setup using Marantz Cinema 50. Would you say following the Supreme Audyssey Guide and this video is a good starting point for full manual calibration?
If you'll use Audyssey and dynamic EQ then yes but if you want to use graphic equalizers there are two videos for full manual atmos calibration.
Wondering after all these steps in REW, how do you apply the result in the receiver?
@@Judddernaut Tutorials about FIR filter generation for stereo systems are not compatible with multichannel home theatre receivers. They need a processor (a PC/Mac or to a lesser extent a MiniDSP somewhere in the chain).
Btw, a Marantz Cinema 50 is perfectly capable of processing FIR filters and in fact that's how Audyssey/Dirac work but they are not open to users. There's no brand I know of which lets you use your own FIR filters.
In the live chat I replied Surround. I guess it didn't go through.
I hope you could read my answer!
@@ocaudiophile I could not. I'm curious in the order of how I could use your methods to get the best surround sound systems calibrated.
@@mikeortiz2139 If you wanna stick with Audyssey, start with "Supreme Audyssey Calibration" video and make sure you also watch "Audyssey for Atmos Music" tutorial. If you want to calibrate your system manually (but you'll lose Dynamic EQ/volume) start with "Dolby Atmos Height Channels... and Dolby Atmos Manual Tunin Complimentary" videos. If you don't want to buy a Umik mic, there's also a link to a calibration file to use with the standard Audyssey microphone in REW in the decription of the "hidden Audyssey mic calibration" video.
@@ocaudiophile I have a UMIK I use for my subs. Super Thanks. I'll do that.
I like your idea for VBA very much. However, you can still improve things. First of all, you should never boost the delayed low pass filter waveform. It should correspond with a reflection which will never be stronger then the direct signal. A value of -6dB is good to start with. Further, use an 8th order Bessel low pass filter (LPF) at 100Hz. It will give you an almost perfect and symmetrical impuls respons. And it will correspond nicely with the Harman Kardon curve still. The delay as given in the IR window button will be 4.97ms. Then, no, the total of length, width and heigth is not the correct number. At all. Also, you make a mistake regarding your open room. That would mean the frequency dip would be lower than calculated, not higher. To cut things short, just use twice the room length as reference for the LPF filter delay (with its own delay of 4.97ms subtracted from the needed delay and then put into the time offset of the LPF impuls respons). This gives me an almost perfect result, on the condition that the 8th order Bessel LPF at 100Hz is being used.
Thank you. I will try these suggestions. Recently I am doing it quite differently though (no boost to the lpf and it targets the second peak) but I am reluctant to put anoher VBA video. People don't like when you update methods ;)
Here's the mdat if you feel like reverse engineering it:
drive.google.com/file/d/1cdY1icKBdYpst1SSs7TdILoLP-cSrdpO/view?usp=sharing
I downloaded your file and tried my method. I think it worked fine. The attenuation of the LPF needed to be 3dB, 6dB was too much.
@@jeroendezaire7266 using minimum phase copy of the smoothed vba filter trick sometimes improves the results. Out of all the operations in REW, generating min phase with replicate data outside range option ticked is the only one that'll freeze the smoothing used.
@@ocaudiophile I like the minimum phase copy of the VBA filter, it gives a perfect impuls and step respons with no ringing at all. But I do not understand smoothing, that would ruin the now perfect VBA filter.