I played also a lot with lots of method in REW alone or with rePhase. I managed to fully iron out EQ and phase, or just use simple EQ, and tried everything in between, including room mode dips with clever phase filtering. Neumann KH 310 with dedicated studio converter… these are very good and powerful Class AB amp, which is important as they have naturally really good flat phase response despite being 3-way speakers, plus EQ correction does not push them to distort even with 6dB boost that requires double effort from amp to perform. At the end of the day all failed and were much more useless then the powerful algorythm of my Sonarworks SoundID, but likely Dirac and IK ARC are safer too. Note: I still measured the SoundID correction with REW and applied further fine tuning as it is not flawless, but I got much tighter response, and coherent response without manual phase correction. Because the problem is always IR caused delay and pre-ringing, or sometimes post-ringing side effect. :(( This hasn’t been mentioned or presented in your vid. Listening test is always the most important beyone freq response or phase response smoothing. Michael Jackson’s - Beat it… the kick part right after the intro will always reveal if you went too far with phase linearization, and always always check distortion ratio and group delay, which is much more important in our brain’s process when we are listening to music.
EQ is by far the greatest thing in audio. Free, highly effective and allows for speaker/headphone designer to not compromise as much between FR and technical performance.
What’s funny is I’ve tried those Oratory settings before on my headphones and everytime I just like the headphones as they are stock and turn them off. Same goes for pad rolling, never once found a pad that beats out what the headphone shipped with.
@@mikeg2491 The Oratory presets are usually personalized for taste, but still my experience is the opposite in that essentially every headphone is better with than without. Same experience with pads though with IEMs some tips have been way better than the stock ones.
It's always going to be personal and I'm starting to think it depends on each listener's priorities. Those who prioritise tone above all else will gravitate towards EQ. I'm in the other camp though where the slight timing, imaging and resolution smearing created by EQ is enough to detract from the listening experience. To be clear, I'm not promoting one side over the other - what matters is that people enjoy the music.
Room correction is essential in unprepared living rooms. I currently use Roon with 2 convolution filters in succession. First is for two channel correction. Second for full response including subwoofer.
I know nothing about this except that the final result is obviously a better sound tailored to my taste and room. I am wondering if there are any introduction type tutorials along with what you teach here? Sorry but it is all new to me. Thanks for posting this.
Lachlan, thanks for this video. I've made a first pass effort using the rePhase method you describe and have gotten good results, but I need to work on it some more. A suggestion for you is to run REW with and without the convolution filter engaged to see the results for IR, phase and EQ and then explain what one should look for and how to interpret the results. I've done this with the filters I've created, and I see the changes, but I have only a superficial understanding of acoustics. Again, very nice video! 👍
Glad you've started trying it out. Finding the right room curve / target curve can take some fiddling. I actually thought about remeasuring with the filters running on the DMP-A8 just to see the phase shifts created. I can't do it with the rePhase version because that has to go via Roon and I can't measure via Roon when using REW.
Another nice one! Saved to watch properly another day. Am looking for a streamer. Eversolo was top on the list cos it works with Apple Music but I don’t like the DAC inside it.
Brillant video! It works even with the DMP A6 . If i understand well, REW EQ filters recommend me 6.3 db headroom for my 10 filters. Does it mean i have to reduce the EQ Gain by 6.3db in the DMP A6, config 1 menu, EQ gain as well? Fortunately, my 300W has enough juice for it but i must listen at 64% instead of 45% volume for the same 85db output.
The deeper dips in the response are *probably* there because of a cancellation. These are often from a floor reflection (or ceiling) that is a path that is about half the wavelength longer than the direct path (from the driver to your ears). And these dips are from the specific mic position. OR it is from phase cancellation between the drivers because they are at different positions on the speaker. In the crossover region Which are going to happen differently at different listening positions - so compensating for the room/set up is problematic, in my opinion. I would measure the speakers at 1 meter on the listening axis - this removes the floor/ceiling cancellations. You need to use the listening axis that is midway between the center of the tweeter and midrange/bass driver; so the distance is the same to the two drivers. And I would average the left and right speakers. Then you are not trying to compensated for "physical" things that are "outside" the speakers themselves. These "physical" things are going to vary for all the other possible listening positions - so the changes are *not* correcting the sound for any other listening position. This process can only go so far (as you mention) and so it is best to use it to "fix" the frequency response deviations of the speakers. Speaker placement and room treatment is the best way to get their response to match.
Thanks for the info, Neil. I can't recall now which set of measurements I used for the examples in this video so I'm not sure where the mic was placed or if I used a real time measurement across a small range of movement with the mic.
Hello and thanks for the very useful video. A question: why not focus on frequencies whose length does not depend on the movements on the most common sofa and movements of the listening position, so we are talking up to about 500 Hz (about 70cm). Do you see any critical issues in applying your method however limited up to 500 Hz? Does linear phase work with RePhase cause any mess if the file you are working on is limited?
You can absolutely focus on any frequencies you want to. I don't see any reason not to address 20-20,000Hz if we have the opportunity to though. As for using rePhase, it will work happily with any range of frequencies so go for it! Have a play and see what sounds best to your ears.
@@PassionforSound Because there are differences after the Schroeder frequency, the 180, 200, 300 Hz due to reflections etc with curves that rise and fall depending on small movements and therefore I find little sense in having a theoretical flat curve for a head that must be stationary as in a recording studio as at work. So I'm talking about passionate listening, yes, but for pleasure and perhaps to share with other people. Of course, during the measurement phase, I could create a cloud of points around the head within 50 cm and take an average. And here comes another question, the best average to do is the RMS avg + Phase avr tool? Always in case I then have to make a linear phase filter in RePhase. Thank you very much for the attention
I have deliberately not delved too deep into this because there are SO many factors to consider. If you're working with a larger listening area, taking a real time measurement in REW is a good idea to even out the FR across multiple locations.
Im a little slow at this, but would you use correction on the internal devices eq/peq as well as roons eq/peq? I would think it would be one or the other but I don't know..
Hi, please rate the Topping l70 both as an amplifier and as a preamplifier? And especially the opinion in relation to the Topping A30 pro and the Topping L50? Thank you
It's shocking to me machine learning algorithms are not more widespread in the digital signal processing arena to facilitate detail retrieval. Theoretically, a versatile algorithm properly trained on a catalogue of sounds and music should be able to upscale any signal to say for example PCM64bit/768kHz or any arbitrary (lossless) encoding that is desired. Like (Google's) VP9 codec for audio. Like an artist drawing an ideal waveform on a canvas. All solid state electronics would become obsolete. Digital signal processing algorithms are so useful and powerful to use before converting the signal to analog.
That's a very interesting idea! I wonder if they could provide the accuracy needed down to a timing accuracy of 1-2 micro seconds to achieve a sound that's completely natural to our brains.
There is no point to convert something that starts as PEQ and outputting FIR, when the programs you mentioned already support PEQ (Parametric EQ). FIR from REW is only useful when you use RePhase (which you are not doing here).
As discussed in the video, PEQ will address both channels equally and therefore cannot account for the different influences in the room on the left and right speakers. FIR allows for distinct filtering of the left and right channels to create more precise correction and a better resultant sound. Also, depending on the devices used, a FIR filter may provide more adjustment capabilities than an EQ with less bands (e.g. 10-band stereo PEQ vs an FIR with up to 20 filters per channel)
You can absolutely can PEQ left and right channels separately@@PassionforSound . I don't know how much experience you have on DSP but I've been doing PEQ and digital room correction (65k FIR filters) for a long time. As I mentioned, starting with PEQ and converting it to FIR just adds extra CPU loading for no reason when you can just leave it as PEQ for Roon, JRiver, HQPlayer, whatever you use. FIR is useful with REW + RePhase (not REW alone as you demonstrated)
In the devices this was mostly intended to address (e.g. EverSolo DMP-A8, DMP-A6, Roon, etc.) the PEQ will apply to both channels simultaneously (unless there's a setting in Roon I haven't explored). After that, there's also the lack of filters on the PEQ for those devices (10 PEQ vs 20 per channel on FIR), not to mention the simplicity of loading an FIR vs manually copying across EQ settings.
I played also a lot with lots of method in REW alone or with rePhase. I managed to fully iron out EQ and phase, or just use simple EQ, and tried everything in between, including room mode dips with clever phase filtering.
Neumann KH 310 with dedicated studio converter… these are very good and powerful Class AB amp, which is important as they have naturally really good flat phase response despite being 3-way speakers, plus EQ correction does not push them to distort even with 6dB boost that requires double effort from amp to perform.
At the end of the day all failed and were much more useless then the powerful algorythm of my Sonarworks SoundID, but likely Dirac and IK ARC are safer too. Note: I still measured the SoundID correction with REW and applied further fine tuning as it is not flawless, but I got much tighter response, and coherent response without manual phase correction.
Because the problem is always IR caused delay and pre-ringing, or sometimes post-ringing side effect. :((
This hasn’t been mentioned or presented in your vid. Listening test is always the most important beyone freq response or phase response smoothing.
Michael Jackson’s - Beat it… the kick part right after the intro will always reveal if you went too far with phase linearization, and always always check distortion ratio and group delay, which is much more important in our brain’s process when we are listening to music.
finally a simple explanation of using REW and repahse
I'm glad I could keep it simple!
EQ is by far the greatest thing in audio. Free, highly effective and allows for speaker/headphone designer to not compromise as much between FR and technical performance.
What’s funny is I’ve tried those Oratory settings before on my headphones and everytime I just like the headphones as they are stock and turn them off. Same goes for pad rolling, never once found a pad that beats out what the headphone shipped with.
@@mikeg2491 The Oratory presets are usually personalized for taste, but still my experience is the opposite in that essentially every headphone is better with than without. Same experience with pads though with IEMs some tips have been way better than the stock ones.
It's always going to be personal and I'm starting to think it depends on each listener's priorities. Those who prioritise tone above all else will gravitate towards EQ. I'm in the other camp though where the slight timing, imaging and resolution smearing created by EQ is enough to detract from the listening experience. To be clear, I'm not promoting one side over the other - what matters is that people enjoy the music.
Room correction is essential in unprepared living rooms. I currently use Roon with 2 convolution filters in succession. First is for two channel correction. Second for full response including subwoofer.
I agree! I hadn't thought about running two filters in succession in Roon. That's a great idea.
Fantastic video! Very clear and instructive. Thanks for sharing!
My pleasure! Glad I could help.
Thanks for this. I’m setting it aside to fully absorb and implement with my DMP A-8 asap.
I hope you love the results!
I've closely tracked the whole video now. Brilliant. Time to buy the microphone and download REW. Thanks again !!@@PassionforSound
Great! Hope you love the results!!
I know nothing about this except that the final result is obviously a better sound
tailored to my taste and room. I am wondering if there are any introduction type
tutorials along with what you teach here? Sorry but it is all new to me. Thanks for
posting this.
My pleasure! I'm glad it sounds better to you.
What kind of tutorials are you looking for?
♥
Look at Lachlan playing with EQ. Is this the best toy he's had for ages.
LOL. No. Only for speakers still for me. 🙂
Lachlan, thanks for this video. I've made a first pass effort using the rePhase method you describe and have gotten good results, but I need to work on it some more. A suggestion for you is to run REW with and without the convolution filter engaged to see the results for IR, phase and EQ and then explain what one should look for and how to interpret the results. I've done this with the filters I've created, and I see the changes, but I have only a superficial understanding of acoustics. Again, very nice video! 👍
Glad you've started trying it out. Finding the right room curve / target curve can take some fiddling.
I actually thought about remeasuring with the filters running on the DMP-A8 just to see the phase shifts created. I can't do it with the rePhase version because that has to go via Roon and I can't measure via Roon when using REW.
Hi, Lachlan. I've figured out a way to test the filters in Roon with REW. I will e-mail you with the files and protocol. @@PassionforSound
Thank you. I'm swamped at the moment so if I'm slow to reply (I think I saw your email come through), feel free to chase me up.
Another nice one! Saved to watch properly another day. Am looking for a streamer. Eversolo was top on the list cos it works with Apple Music but I don’t like the DAC inside it.
Maybe check out the Volumio Rivo. I haven't heard it, but they make good products based.on past experiences.
@@PassionforSound thank you!
Brillant video! It works even with the DMP A6 . If i understand well, REW EQ filters recommend me 6.3 db headroom for my 10 filters. Does it mean i have to reduce the EQ Gain by 6.3db in the DMP A6, config 1 menu, EQ gain as well? Fortunately, my 300W has enough juice for it but i must listen at 64% instead of 45% volume for the same 85db output.
That's correct about the gain settings for headroom. And yes, now that the DMP-A6 has the EQ functionality, this will work for the A6 too. 🙂
The deeper dips in the response are *probably* there because of a cancellation. These are often from a floor reflection (or ceiling) that is a path that is about half the wavelength longer than the direct path (from the driver to your ears). And these dips are from the specific mic position. OR it is from phase cancellation between the drivers because they are at different positions on the speaker. In the crossover region
Which are going to happen differently at different listening positions - so compensating for the room/set up is problematic, in my opinion.
I would measure the speakers at 1 meter on the listening axis - this removes the floor/ceiling cancellations. You need to use the listening axis that is midway between the center of the tweeter and midrange/bass driver; so the distance is the same to the two drivers. And I would average the left and right speakers.
Then you are not trying to compensated for "physical" things that are "outside" the speakers themselves. These "physical" things are going to vary for all the other possible listening positions - so the changes are *not* correcting the sound for any other listening position.
This process can only go so far (as you mention) and so it is best to use it to "fix" the frequency response deviations of the speakers. Speaker placement and room treatment is the best way to get their response to match.
Thanks for the info, Neil. I can't recall now which set of measurements I used for the examples in this video so I'm not sure where the mic was placed or if I used a real time measurement across a small range of movement with the mic.
Hello and thanks for the very useful video.
A question: why not focus on frequencies whose length does not depend on the movements on the most common sofa and movements of the listening position, so we are talking up to about 500 Hz (about 70cm).
Do you see any critical issues in applying your method however limited up to 500 Hz?
Does linear phase work with RePhase cause any mess if the file you are working on is limited?
You can absolutely focus on any frequencies you want to. I don't see any reason not to address 20-20,000Hz if we have the opportunity to though.
As for using rePhase, it will work happily with any range of frequencies so go for it! Have a play and see what sounds best to your ears.
@@PassionforSound Because there are differences after the Schroeder frequency, the 180, 200, 300 Hz due to reflections etc with curves that rise and fall depending on small movements and therefore I find little sense in having a theoretical flat curve for a head that must be stationary as in a recording studio as at work. So I'm talking about passionate listening, yes, but for pleasure and perhaps to share with other people.
Of course, during the measurement phase, I could create a cloud of points around the head within 50 cm and take an average.
And here comes another question, the best average to do is the RMS avg + Phase avr tool? Always in case I then have to make a linear phase filter in RePhase.
Thank you very much for the attention
I have deliberately not delved too deep into this because there are SO many factors to consider. If you're working with a larger listening area, taking a real time measurement in REW is a good idea to even out the FR across multiple locations.
Im a little slow at this, but would you use correction on the internal devices eq/peq as well as roons eq/peq? I would think it would be one or the other but I don't know..
Great question! You'd normally only use one or the other and Roon is generally the more powerful.
Hi bro im new to the IEM world and i want to buy the KZ castor bass version, can u pls tell me what's ur thoughts on Castor bass version
I haven't heard that one. Sorry.
Hi, please rate the Topping l70 both as an amplifier and as a preamplifier? And especially the opinion in relation to the Topping A30 pro and the Topping L50? Thank you
I don't currently have that one to test. Sorry
@@PassionforSound thanks
It's shocking to me machine learning algorithms are not more widespread in the digital signal processing arena to facilitate detail retrieval. Theoretically, a versatile algorithm properly trained on a catalogue of sounds and music should be able to upscale any signal to say for example PCM64bit/768kHz or any arbitrary (lossless) encoding that is desired. Like (Google's) VP9 codec for audio. Like an artist drawing an ideal waveform on a canvas. All solid state electronics would become obsolete.
Digital signal processing algorithms are so useful and powerful to use before converting the signal to analog.
That's a very interesting idea! I wonder if they could provide the accuracy needed down to a timing accuracy of 1-2 micro seconds to achieve a sound that's completely natural to our brains.
I like my speakers totally flat.
That can be achieved 🙂
There is no point to convert something that starts as PEQ and outputting FIR, when the programs you mentioned already support PEQ (Parametric EQ). FIR from REW is only useful when you use RePhase (which you are not doing here).
As discussed in the video, PEQ will address both channels equally and therefore cannot account for the different influences in the room on the left and right speakers. FIR allows for distinct filtering of the left and right channels to create more precise correction and a better resultant sound.
Also, depending on the devices used, a FIR filter may provide more adjustment capabilities than an EQ with less bands (e.g. 10-band stereo PEQ vs an FIR with up to 20 filters per channel)
You can absolutely can PEQ left and right channels separately@@PassionforSound . I don't know how much experience you have on DSP but I've been doing PEQ and digital room correction (65k FIR filters) for a long time. As I mentioned, starting with PEQ and converting it to FIR just adds extra CPU loading for no reason when you can just leave it as PEQ for Roon, JRiver, HQPlayer, whatever you use. FIR is useful with REW + RePhase (not REW alone as you demonstrated)
In the devices this was mostly intended to address (e.g. EverSolo DMP-A8, DMP-A6, Roon, etc.) the PEQ will apply to both channels simultaneously (unless there's a setting in Roon I haven't explored). After that, there's also the lack of filters on the PEQ for those devices (10 PEQ vs 20 per channel on FIR), not to mention the simplicity of loading an FIR vs manually copying across EQ settings.