Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Trumpetguy]

The ETCs are good even before DRC - in fact I don't see much improvement but then, I'm no expert at all. So please forgive my ignorance 

You are probably right. DRC seem to give good results in most rooms, but also seems to give the little extra in already good listening environments. The already good ETC suggests this room is aready quite all right from the start.

In regard to B&K house curve: I personally find it dull. With especially many classical recordings the atmosphere from the concert room disappears. As far as I've read in the white-paper from B&K the curve resembles the acoustics from a concert hall/larger music arena, which apparently makes it ideal for playback of close-up recordings.

While probably true, I nevertheless prefer a completely flat curve for most recordings (mind you, I listen to classical music where far-field recording isn't uncommon... as far as I know).

I find my taste to be changing every now and then. I do not believe one curve is the ultimate answer for all recordings and situations. For rock and pop, B&K sounded really good. But I can imagine the curve not being quite as good for classical music. I listen to classical music a lot, and want to test it on large orchestral works and chamber music. Looking forward to that.

With movies I have no idea. I'm still trying to figure out if dvd/blu-ray discs are remastered with a flat frequency respons or is attenuated in the upper range to compensate for screen dampening and the distance from speaker to listeners in the cinema room.

I know what my preference is - flat curve on movies is a boring experience. With movies, I am not after what is correct, only what is entertaining. To get that, my experience is that using the ears is the best thing. I have been using something close to a pear shaped curve for a couple of years. An 8dB hump below 500Hz. Now I am going to test B&K. It may give more flesh to voices.

[AudioVero]

I can provide the correction curve, but I yet do not have any way to measure actual room response with filter activated. Is it the first you are requesting? If it is the latter, could you please describe how I could accomplish that?

I have asked for the correction curve itself, but not for the corrected room response curve.
As stated in a previous message I advise to check for the correction (influenced by measurement and the defined target). If e.g. someone says that the B&K curve sounds dull he can easily see this in the correction curve.

[Mitchco]

...Or maybe luck has nothing to do with it?

I designed it that way or tried to.  I have high efficiency speakers http://i1217.photobucket.com/albums/dd381/mitchatola/barnettestylec.jpg  At 100db SPL @ 1w/1m sensitivity, even with 12db correction, the sound is still quite dynamic.
 
Here is an overlay of the measured uncorrected (purple) and with DRC enabled (blue) frequency response at the listening position.  http://i1217.photobucket.com/albums/dd381/mitchatola/jan29rawanddrcfreqrespoverlay.jpg  I could get away with 6db of correction if I did not want to extend the low end down to 20Hz. I think 6db correction got me to 32Hz.  No sign of overload yet and I like to listen reasonably loud.
 
I attribute this to the upfront work I did using a distance laser measure to accurately and precisely set up an equilateral triangle between listening position and speaker symmetry.  This sets up DRC for success.

I have a ~10ft equilateral triangle in my 30’ x 15’6” x 8’ room.  My speakers are set up along the long wall.  http://i1217.photobucket.com/albums/dd381/mitchatola/Roomside.jpg  With respect to the ETC and diffuse field, I tried to set up a reflection free zone from the speakers to the listening position.  Which means the listening position needs to be away from the rear wall.  http://i1217.photobucket.com/albums/dd381/mitchatola/carpetrear.jpg  I am about 1/3 of the width of the room from the back wall and the speakers are 1/6 of the width of the room from the front wall.

I voiced the speakers to the room by listening to music (in the bass range) and moving each speaker slowly to and from the front wall until I found the spot where all the bass notes from the song sound the most even in loudness.  Use the balance control to voice one speaker at a time.  You know you don’t have the spot when some bass notes disappear (null) or are very loud (peak).    Once I got them in the ballpark, I used REW to measure and verify and fine tune.  After a 3 iterations, I was done.  Here is the uncrorrected frequency response at the listening position.  http://i1217.photobucket.com/albums/dd381/mitchatola/jan29leftspeakerrawfreqresp.jpg  Pretty good response on its own given the constraints of my room setup. 

How did I measure?  As you saw in the rom pics, I have a good size table and couch.  I moved those off to the side of the room and set up the calibrated omni mic (isolated from floor vibrations) at the listening position facing straight up and ear height.  No multi-seat measures yet as it seems to sound good anywhere in the room.

Re:  narrow peaks and dips.  Here is the unsmoothed frequency response at the listening position with DRC. http://i1217.photobucket.com/albums/dd381/mitchatola/jan29drcrawfreqrespDRC.jpg  I as understand it, the measurement software has more measuring resolution than our ears can hear.   We can start discriminating frequency differences between 1/3 and 1/6 octave range.  I could be wrong, but that is my understanding.

So when you see my 1/6 octave smoothed frequency response, it seems like every little narrow dip and peak was corrected.  That is not the case as you can see in the raw response.  But for our ears purpose, 1/6 octave smoothing is just outside our ears ability to differentiate.  At least that is my understanding.

[AudioVero]

  I as understand it, the measurement software has more measuring resolution than our ears can hear.   We can start discriminating frequency differences between 1/3 and 1/6 octave range.  I could be wrong, but that is my understanding.

So when you see my 1/6 octave smoothed frequency response, it seems like every little narrow dip and peak was corrected.  That is not the case as you can see in the raw response.  But for our ears purpose, 1/6 octave smoothing is just outside our ears ability to differentiate.  At least that is my understanding.

The pulse response underlying the frequency plot has a certain length. The frequency chart shows the behaviour over the full pulse length. Imagine the duration of the pulse response e.g. with 1 hour and you will get even more peaks and dips.
So the frequency plot has to be understood as showing a steady state situation. There is no information in the chart at which time a certain frequency has happened.

But in reality we hear a stream of incoming and continuously changing soundwaves. So our ear does not take a snapshot for a certain time, compute the frequency response and then uses it. It is a common misunderstanding of the frequency plot.
Even a 1/n-octave smoothing based on averaging the frequency plot for a 1/n octave does not show the reality because the underlying data are depending again on the pulse length. The error is just not as obvious because of the averaging algorithm.

A better way to interprete a pulse response is the frequency dependent windowing. By nature in the windowed pulse response the data with a bigger time distance from a given point of interest are windowed out and do not influence the frequency plot anymore. So you put more weight on the soundwave the actually happens. By nature such a frequency plot looks like smoothed but it is not a smoothing. It is simply the FFT result of a momentary situation.

This way a correction based on such a frequency dependent windowed pulse also is correcting the momentary sound. It does not correct a measured pulse response over its full length.
Of course now there are again 1001 strategies of how to carry out the frequency dependent windowing, leading to different results.
Anyway FDW is now more and more common. So Audiolense and also Acourate basically apply FDW.

So just use the unsmoothed frequency plot as an indicator on how much noise and reflections are present in your room. With a better acoustical room treatment you will get a better plot = less peaks and dips. In other cases check the smooth frequency response, preferably a result of a frequency dependent windowed pulse.

[Trumpetguy]

I have asked for the correction curve itself, but not for the corrected room response curve.
As stated in a previous message I advise to check for the correction (influenced by measurement and the defined target). If e.g. someone says that the B&K curve sounds dull he can easily see this in the correction curve.

Here is the filtered frequency responses, correction curves and simulated final frequency responses (Audiolense model). I used time domain correction and linear phase option.

The first curve with almost flat target:




The second with something similar to the Bruel & Kjär target:


It was not me that said B&K sounds dull, but I admit it is quite different from my old near-flat curve. I may end up preferring it all the same, though. I would appreciate any opinions on the corrections. What puzzles me is that there is a limit of +6dB correction, but even so the correction curve amplitude is way larger at some frequencies.

Here is a ETC of the actual measurement:



My room is almost square (HxWxD 2.35x3.95x4.2m). Both reflective windows (behind the speakers) and absorbers at first refection points on  the sides and in the ceiling. Some would say a bit too much dampening. My sofa is placed about 1.2m from the back wall. I do not have space for an equilateral triangle, but almost, and the distance to each speaker is as good as identical. The eagle-eyed will see that the measurement was done with slightly different distance to the speakers, as the two first ETC peaks are about 0.04ms apart.

[hulkss]

Of course now there are again 1001 strategies of how to carry out the frequency dependent windowing, leading to different results.

I thought fixed point per octave (FPPO) was a common method that works well:
From SMAART documentation:

[Mitchco]

Here is the filtered frequency responses, correction curves and simulated final frequency responses (Audiolense model). I used time domain correction and linear phase option.

Thanks for posting your graphs.  Looks good!

If you get a chance and want to try an experiment, open up Audiolense and click new target.  Click on the minimum phase radio button.  Right click on the red point on the chart.  That should pop up a dialog with several tabs.  One tab should be labelled Data (thanks hulkss for the tip) Click on it and enter the following:

200      -0.5
2000     -3
20000   -6

That's what I use as a minimum phase target and my interpretation of the B&K target looking closely at the chart from the article.

To my ears, on my system, it produces the best tonal (or spectral) balance compared to several other target variations I have tried. Not too bright, not too dull, but just right.

The big suprise for me was how this target resolves the best 3D soundstage as well.   Not too far forward, not too far back, but just right.  At least that is how it sounds on my system, to my ears. 

Relative to other targets, I did not think a couple of db adjustment either way would make a big audible difference, but it did, not only in tone quality, but the soundstage seems sensitive to small adjustments as well.  At least on my system and to my ears.  That's why I think it is important to enter the B&K target numbers directly into the Target Designer's  Data page for accuracy, as opposed to drawing on the graph with the mouse.

I listen mostly to both low and hi res rock, pop, blues, world, folk, and some classical.

Curious to hear your (and others) feedback if you try this target out.

[Trumpetguy]

Mitcho: I have made the correction curves, and will report back during the weekend. Could you explain one thing - you suggest a 6dB decrease at 20kHz. In your measured room response in one earlier post, the attenuation was about 10dB. How can you apply a -6dB target and then measure -10dB? BTW I tried -10dB, and that sounded dull

[Mitchco]

Mitcho: I have made the correction curves, and will report back during the weekend. Could you explain one thing - you suggest a 6dB decrease at 20kHz. In your measured room response in one earlier post, the attenuation was about 10dB. How can you apply a -6dB target and then measure -10dB? BTW I tried -10dB, and that sounded dull

Trumpetguy - great question, I wondered that myself.  If you look at the measurement again, it is closer to -9db at 20KHz.  Here is a link to my Audiolense target and simulated response:  http://i1217.photobucket.com/albums/dd381/mitchatola/Jan29Audiolense.jpg    You see the target is indeed set at -6dB at 20KHz.  However, my horn tweeter rolls off right at 20KHz.  One could argue it is actually closer to -8db right at 20Khz.  That would mean there is 1dB or so discrepancy at 20KHz between simulated and measured.
 
While there are many variability’s, I think the fundamental issue is that I used a completely different computer (laptop) and sound card combo to measure the frequency response using REW versus the much more expensive HTPC and sound card (Lynx L22) combo I used for taking the Audiolense measurements.

Here is the frequency response of my Lynx L22 card. http://i1217.photobucket.com/albums/dd381/mitchatola/lynxl22-1.jpg  It is ruler flat right out to 50KHz., which is the measurement limit in REW.  I used this to take the Audiolense measurements.  Now have a look at the frequency response of the onboard sound chip in the laptop I used to measure the frequency response with REW:  http://i1217.photobucket.com/albums/dd381/mitchatola/thinkpadx301soundcardfreqresp.jpg  Yes, I used the corresponding sound card cal's for both measuring in REW and Audiolense. 

Additionally, in order to take the measurement in REW, I used the loopback capabilities of my Lynx card, but rather than going through Convolution in JRiver, since I can’t get at its input (would be a great feature to add in JRiver btw), I had to host ConvolverVST in Savihost, which then hosted the Audiolense generated filters.  More variability…

In hindsight, I should have used the exact same gear (i.e. HTPC and Lynx L22) for both REW and Audiolense measurements to remove this variability from the equation.   Next time…

Personally, I believe it is more the slope of the B&K target curve than anything else that contributes to what I am hearing and would recommend going with the B&K target numbers (i.e. -0.5dB @ 200Hz, -3@2Khz, and -6@20KHz) as the baseline.  Do you have any way (i.e., soundcard loopback or virtual audio cable) of measuring the actual frequency response?

[mojave]

Do you have any way (i.e., soundcard loopback or virtual audio cable) of measuring the actual frequency response?

I just posted a guide for using the RTA function of REW. It is useful for measuring the actual frequency response of your signal chain with EQ or convolution.

How to use the RTA feature of REW with JRiver

[JimH]

I just posted a guide for using the RTA function of REW. It is useful for measuring the actual frequency response of your signal chain with EQ or convolution.

How to use the RTA feature of REW with JRiver

Thanks, mojave!

[Mitchco]

I just posted a guide for using the RTA function of REW. It is useful for measuring the actual frequency response of your signal chain with EQ or convolution.

How to use the RTA feature of REW with JRiver

Ask and yee shall receive!  Thank you mojave!  This should work perfect!

PS.  I should have listed the other parameters that may have helped contribute to my 20KHz variability:

- mic cal file ends at 19,998Hz.  I wonder what happens right at 20KHz? 
- band limited the laptop sound card REW loopback measurement with a swept sine wave to 20KHz.
- band limited the REW speaker measurement by sweeping to 20KHz.

In contrast, the Audiolense measurements were band limited to 24KHz by the swept sine wave.

Next time I will increase everything to 25KHz and try the RTA!

[hulkss]

I just posted a guide for using the RTA function of REW. It is useful for measuring the actual frequency response of your signal chain with EQ or convolution.

I would much prefer to be able to feed an REW sweep (impulse response measurement) through MC DSP.

An RTA measurement is much different than measuring the system impulse response. Yes, the results will be the same if you set a long rectangular window for the FTT, however, this approach does not exclude any reflected sound from the measurement.

This is not at all how your brain perceives sound, there is much more importance on the direct sound which arrives first than reflections which arrive later to your ear. An RTA treats it all the same.

A sweep with Frequency Dependent Windowing (FDW) for processing the FFT (as posted by Uli below) is the best way to measure your system and is how Acourate and Audiolense work. This is also illustrated in the Fixed Point Per Octave (FPPO) example I posted below. Impulse response analysis in REW provides similar capability.

[adolfotregosa]

Hi guys. I'm new here! First post actually and I have to say this topic is really interesting!

Sorry to hijack the thread like this, I'm sorry for it, but I need some help. Just like NickF was, I'm waiting for my microphone kit to arrive so i can start playing.
But i have some questions that i really haven't found answer by searching.

1 - First of all which audiolense version should I invest in? I need at least the Surround version. But should I buy the XO ? Do you guys think is it worth the extra money?

2 - The mic position !! I haven't found any picture ! Horizontal position ? vertical position ? angled ? For both calibrations? I mean which calibration position for music (2.1) and 5.1 for movies i have to use?

3 - I use hdmi from ati/amd 6970 card into a onkyo 609, thats for output but how about input? I have to use the onboard realtek mic jack, is this alright ?? I took out a Asus D2x when i started using hdmi, should i use it instead for mic in ?

4 - What volume level do I configure in windows for the mic ? 100% ? and boost at zero ? and what about the kit pre-amp volume knob ?

5 - audiolense likes asio but since i'm using hdmi out from ati/amd i don't have asio. Should I use asio4all ? Pity it doesn't work with wasapi like J River, or ?

6 - onkyo volume !! how do i set it? So i can measure correctly without causing damage to the speakers?

7 - What do I do about the calibration/speaker settings that were made using Onkyo MCACC EQ ? Do i keep the speaker distance and levels it measured ? makes sense leaving it alone but...

A lot of questions, I know, but it's my first time doing this !!

Thank you for all for this great forum.

[hulkss]

1 - First of all which audiolense version should I invest in? I need at least the Surround version. But should I buy the XO ? Do you guys think is it worth the extra money?

Active XO is a major benefit definitely worth the investment IMHO (I assume you know what Active XO requires to implement).

2 - The mic position !! I haven't found any picture ! Horizontal position ? vertical position ? angled ?

Get an omnidirectional mic and aim it straight up at ear level when seated. Compensate for high frequency directional fall-off in frequency response of the mic in the mic cal file or in your target response. Most mics come with data like this:

[NickF]

I'm only a beginner here but I'll give my answers from recent experience!

1 - which audiolense version should I invest in?

Good question. I went for Surround because I wanted to cover stereo for my music and surround for my movies but didn't want active crossovers.  However, I soon found that there were features I needed in XO such as corrections of more than 6 dB so i paid for an upgrade.  After all, it's only money!!

2 - The mic position

The mic is omnidirectional and needs to be mounted vertically.  It comes with a clip with a large thread which I didn't recognise so I have taped it to an old camera tripod.  I place the tripod on my listening seat and put cushions around it to avoid any reflections and vibrations.  The mic and pre-amp come with no cables.  All connections are RCA Phono.  I asked a question about the quality of the cable but hulkss answered with a general answer about using quality cables.  My view, and I am a novice in convolution, is that this measurement path is critical in the setup.  It needs to hear every deficiency in the playback path and environment so it is important not to introduce any additional deficiencies or interference into the measurement path.  So I have made some good quality cables.

3 - I use hdmi from ati/amd 6970 card into a onkyo 609, thats for output but how about input? I have to use the onboard realtek mic jack, is this alright ?? I took out a Asus D2x when i started using hdmi, should i use it instead for mic in ?

I am lucky to have an RME card which has inputs and outputs.  I have never rated Realtek very highly so I don't know about the quality of the mic input.  Remember that you are using a mic pre-amp so you really want a line input.  I will leave it to others to advise here.

4 - What volume level do I configure in windows for the mic ? 100% ? and boost at zero ? and what about the kit pre-amp volume knob ?

I set my amp to a normal listening level and the pre-amp to about mid position.  When you do the Measurement with Audiolense, when the measurement ends, it will display the highest level and, consequently, the part most likely to be clipped or distorted.  If you get either, back off the volume so you get a nice sine wave.  You need to reach a point where the volume is loud enough to expose resonances in the room without distortion.  It's a case of trial and error.

5 - audiolense likes asio but since i'm using hdmi out from ati/amd i don't have asio. Should I use asio4all ? Pity it doesn't work with wasapi like J River, or ?

You need to get the best quality path from mic to Audiolense.  I have no experience with asio4all.  Someone else may help here.

6 - onkyo volume !! how do i set it? So i can measure correctly without causing damage to the speakers?

See my comments in 4 above.  Start low until you know what levels Audiolense will output then increase to the level that you would listen at.

7 - What do I do about the calibration/speaker settings that were made using Onkyo MCACC EQ ? Do i keep the speaker distance and levels it measured ? makes sense leaving it alone but...

See my post further up this thread.  You need to remove all corrections in your Onkyo.  I don't know whether you can keep them stored and just switch them off.

Nick

[adolfotregosa]

  !! thank you all ! Really nice of you guys to help

I felt silly, I forgot I will be using a pre-amp so obviously I'll have to use line in, thank you for that one ! So still leaves, realtek line-in or asus d2x? New question. Line in expects stereo input or mono will also work ? When making the cables I'll use a mono jack? ; stereo jack and only use the tip connection? ; stereo jack and short/use both channel inputs ? What did you do Nick?

"I assume you know what Active XO requires to implement" nop, no idea, i think.. !! I presume using separated amplifiers per speaker type with a electronic crossover between the pre-amp and the power amps? I'm I making any sense here?  ? ? That is out of my league

I asked about surround vs XO because of the extra functions like Audiolense True Time Domain correction and multiseat readings, if that is another thing that could help my sound system then i'll go for XO version.

To disable mcacc, by reading Onkyo user manual i just need to use pure audio mode..hmm but that disables routing the bass to the subwoofer. I'll have to wait for the Mic and experiment on this one.

Any input and correction on any thing is welcome !

Again thank you

[NickF]

Line in expects stereo input or mono will also work ?

Yes, mono will work. Audiolense just found my input and only needs a mono input so it's stereo jack and only use the tip connection.

"I assume you know what Active XO requires to implement"

Hulkss mentioned this.  I don't know what he meant so maybe he could elaborate.  I just installed Audiolense and placed the licence file in the program folder and it worked.

I presume using separated amplifiers per speaker type with a electronic crossover between the pre-amp and the power amps?

This is an option in XO but you don't have to do this and I would advise that you don't try to start with this level of complexity.  You need to look at the Speaker Setup and choose the right match for your setup.  Remember that you will need two filters, one for stereo and one for surround, so you need to go through the whole process twice.  Having done my first attempt at surround yesterday, I now know that multiseat is essential.  I tend to listen to music alone and want the best possible stereo image and single measuring point gives this.  For movies, I need two seats.  With single point, there was absolutely no image in the second seat.  I will try multiseat today.

To disable mcacc, by reading Onkyo user manual i just need to use pure audio mode..hmm but that disables routing the bass to the subwoofer.

I can't help here.  I use 6 channel analogue input to my Denon.  There are other Onkyo users here or on other forums who may be able to help.

Nick.

[Trumpetguy]

1 - First of all which audiolense version should I invest in? I need at least the Surround version. But should I buy the XO ? Do you guys think is it worth the extra money?

Active XO is a major benefit definitely worth the investment IMHO (I assume you know what Active XO requires to implement).

2 - The mic position !! I haven't found any picture ! Horizontal position ? vertical position ? angled ?

Get an omnidirectional mic and aim it straight up at ear level when seated. Compensate for high frequency directional fall-off in frequency response of the mic in the mic cal file or in your target response. Most mics come with data like this:

The XO version is the only with True Time Domain correction, the other versions are frequency corrections only (2.0 or multichannel). TTD is worth the xtra cash. I use XO in an ordinary 7.1 multichannel setup, with two subs on the .1 channel. Makes it really simple to integrate the subs, with separate cross-over filters for all the 7 high-pass channels. Maybe to surround version have that functionality, I don't know. 

hulkss: I have never understood why the mic should be pointed upwards. I have however noticed that the mic direction is a much debated subject. All my measurements have been done with the mic pointed horizontally forward, exactly in the middle of the main speakers. Should I redo all my measurements? I am not worried, I do that from time to time...   

[Mikkel]

About mic-direction: I strongly recommend reading this thread http://www.hometheatershack.com/forums/rew-forum/20293-full-range-eq-no-go-3.html

Take good notice of Wayne A. Pflughaupt's comments.

Vertical orientation is fine with an omnidirectional mic if you have a correction file to compensate for the roll-off at the high frequencies. It probably also is the only convenient orientation if using a multi-channel setup. Otherwise you'd have to turn the mic between each sweep...

As Wayne in the long thread points out, if pointing the mic upwards you get a lot of reflections from the ceiling (if not treated) instead of a more direct loudspeaker response (if pointed towards each speaker and positioned horizontally). But since DRC is about room correction I guess that isn't a problem. But do remember to use a correction-file.


Best regards,
Mikkel

[hulkss]

I have never understood why the mic should be pointed upwards.

The need exists in Digital Room Correction with Time Domain Correction to measure all speakers (including surrounds and rears) from a single point with similar measurement characteristics. By pointing the mic up, you can aim it with the same relative angle (about 90 degrees) to all of your speakers. The best average angle may not be exactly straight up.

You may have noticed that measurement mics are very small at the tip where the measurement sensor is. This is to achieve good omnidirectional measurement characteristics. Human hearing is the most sensitive to sounds from the front and sides. You want the mic to measure all those speakers the same so they will have the same timbre after correction.

As to measuring reflections from the ceiling: The Time Domain Correction Technique is expressly intended to correct for low frequency reflections, so you do want to measure them from all directions as equally as possible. At higher frequencies the measurement time window is shortened so those ceiling reflections are not measured at all. This is possible because you are usually measuring about three feet off the floor. The direct sound path length from speaker to mic is much shorter than a reflected ceiling path so the reflection is easy to eliminate with a shortened measurement window.

[NickF]

Vertical orientation is fine with an omnidirectional mic if you have a correction file to compensate for the roll-off at the high frequencies. It probably also is the only convenient orientation if using a multi-channel setup.

So just for clarity for adolfotregosa, he has bought the MP-1r-KIT Measurement kit from Audiolense which does come calibrated so he should use it vertically, confirmed by hulkss.  Adolfotregosa needs to remember to load the calibration file, supplied on a CD, into Audiolense before he starts to take measurements.

A question, though.  Is the mic calibrated in a vertical position?  Presumably it is.

Nick.

[Trumpetguy]

A question, though.  Is the mic calibrated in a vertical position?  Presumably it is.

I would guess the mic is calibrated in a very narrow, completely anechioc chamber. Like a small box with absorbant walls. Orientation should not be an issue then.

[adolfotregosa]

Seriously, thank you all. I'm really anxious to get the Mic and start working on this. I really want to see what I can pull of my sound system. I've been learning so much since I started building it. lol it never ends 

Does JRiver MC also use convolution while bit-streaming dts, dts-ma, etc movies files? Or it only works for music?

[hulkss]

Does JRiver MC also use convolution while bit-streaming dts, dts-ma, etc movies files? Or it only works for music?

You can use the DSP engine for everything. It's up to you. I have a zone for 2 ch music playback and a zone for multi-channel video playback that are slightly different.

[stealth82]

It would be great if MC allowed to directly record from the mic a played sweep.

Actually I tried using Audacity (ASIO version) to record the sweep played by MC but since MC is using ASIO as well the sound cards figures as in use and I cannot record from any other external program.

I'm doing this because MC allows me to mute channels and apply xover filters useful when measuring any non full range speaker, like the sub or the side speakers. E.g.: joining sub and center to have a correct impulse response to be processed by DRC.

Actually I had to use DirectSound to do so but it's rather tedious to play from MC, quickly switch to Audacity and press REC, and then stop it. Not to mention the record would take place in the same environment in which will be applied; also I would so keep everything in the ASIO realm.

Am I thinking it wrong?

[Trumpetguy]

It would be great if MC allowed to directly record from the mic a played sweep.

Actually I tried using Audacity (ASIO version) to record the sweep played by MC but since MC is using ASIO as well the sound cards figures as in use and I cannot record from any other external program.

I'm doing this because MC allows me to mute channels and apply xover filters useful when measuring any non full range speaker, like the sub or the side speakers. E.g.: joining sub and center to have a correct impulse response to be processed by DRC.

Actually I had to use DirectSound to do so but it's rather tedious to play from MC, quickly switch to Audacity and press REC, and then stop it. Not to mention the record would take place in the same environment in which will be applied; also I would so keep everything in the ASIO realm.

Am I thinking it wrong?

First - using ASIO you will always lock audio stream to one application. You will need to generate and record from the same application. Why not use REW? Or download Audiolense and do it there, with full cross-over functionality. The test license allows you to, but has an attenuation built into the filter you would generate (until you pay ). With Audiolense doing the full-range and/or non-full range sweeps is straight forward.

The main reason I see for MC to generate sweeps and record room response from an analog in channel would be to measure the actual room response with a DRC filter and JRiver convolver in the signal chain.

[stealth82]

First - using ASIO you will always lock audio stream to one application. You will need to generate and record from the same application.

Yeah, I knew that, that's why I wrote my post.

Why not use REW? Or download Audiolense and do it there, with full cross-over functionality. The test license allows you to, but has an attenuation built into the filter you would generate (until you pay ). With Audiolense doing the full-range and/or non-full range sweeps is straight forward.

REW doesn't use ASIO, REW max allowed sample rate is 48kHz and I didn't see options for measuring a 5.1 system, just left/right and sub. Did I overlook the feature?

Audiolense is not free and I'm not certainly taking it into account right now. Besides, what's the purpose of measuring everything with it when you can't put it into "production" because of the attenuation thing...

The main reason I see for MC to generate sweeps and record room response from an analog in channel would be to measure the actual room response with a DRC filter and JRiver convolver in the signal chain.

That is another interesting pro.

[hulkss]

what's the purpose of measuring everything with it when you can't put it into "production" because of the attenuation thing...

I have not had any trouble with an "attenuation thing."

[Trumpetguy]

Audiolense is not free and I'm not certainly taking it into account right now.

Why, because its not free? Is that a requirement? Does that apply to hardware also?

Besides, what's the purpose of measuring everything with it when you can't put it into "production" because of the attenuation thing...

The measurement is as good as anything in the trial version, but the filter is attenuated. Even with a 20dB attenuation you will get a good impression of the performance, but with reduced dynamic range. If you get a license, you can use the same measurement and regenerate the filter. I think that is a pretty good deal.

[stealth82]

Thanks guys for your answers but they don't address any of my current concerns.

I said I'm not interested in using/buying Audiolense - I'm using DRC right now and I've already done the 5.1 DRC configuration by using together JRiver and Audacity for recording the sweeps. I just wanted to deal with less programs and keeping the exact environment I'm using when doing the playback with JRiver - that is distances SPL leveling etc. etc.

[NickF]

Thanks guys for your answers but they don't address any of my current concerns.

I said I'm not interested in using/buying Audiolense - I'm using DRC right now and I've already done the 5.1 DRC configuration by using together JRiver and Audacity for recording the sweeps. I just wanted to deal with less programs and keeping the exact environment I'm using when doing the playback with JRiver - that is distances SPL leveling etc. etc.

It isn't clear to me whether you are using or planning to use any convolution product in your solution.  You talk about doing DRC but are you just using the Equaliser and Room Correction features in MC's DSP Studio?  The real value-add of products like Audiolense is the "intelligence" in the creation of the correction filters to correct for the behaviour of the whole listening environment.  You aren't going to duplicate or replace this with the other DSP components.

I can certainly see value in MC generating the sweep and recording the result to create the parameters required for Room Correction and Equalisers without getting into the full complexity of convolution.  This would simplify the setup of these DSP components and could be a first step to a complete convolution solution.  I suspect that Matt may have this in mind.

Nick.

[)p(]

It isn't clear to me whether you are using or planning to use any convolution product in your solution.  You talk about doing DRC but are you just using the Equaliser and Room Correction features in MC's DSP Studio?  The real value-add of products like Audiolense is the "intelligence" in the creation of the correction filters to correct for the behaviour of the whole listening environment.  You aren't going to duplicate or replace this with the other DSP components.

Nick.

Drc is also the name of a free program to create filters similar to audiolense etc.

http://drc-fir.sourceforge.net/

I also use it. I especially like the result I get when I use its option to compute a target frequency response based on a model of the psychoacoustic perception of the corrected frequency response. see:
http://drc-fir.sourceforge.net/doc/drc.html#htoc31

[NickF]

Drc is also the name of a free program to create filters similar to audiolense etc.

Sorry, my mistake.  Not enough acronyms, apparently!

Nick.

[stealth82]

It isn't clear to me whether you are using or planning to use any convolution product in your solution.  You talk about doing DRC but are you just using the Equaliser and Room Correction features in MC's DSP Studio?  The real value-add of products like Audiolense is the "intelligence" in the creation of the correction filters to correct for the behaviour of the whole listening environment.  You aren't going to duplicate or replace this with the other DSP components.

I'm already using it and I'm certainly not talking about eq and room correction features. But, having them perfectly set up help me when recording the sweep, because I'm replicating the exact environment in which the filter will be applied.

The free DRC solution though, hasn't been created having the multi-channel setup in mind. So, in order to create correct filters for SW and other speakers I need to do some tricks:

e.g. for measuring the center channel:
Copy L to C
High filter 80hz on C
Copy L to SW
Adjust gain (- or + db) for SW to keep C and SW balanced
Low filter 80Hz on SW
Mute all the speakers except for C and SW
Play the sweep from MC and Record the Sweep from Audacity

e.g. for measuring the SW:
Copy L to C
High filter 320hz on C
Copy L to SW
Low filter 320Hz on SW (generally the xover is 2/octave above the normal playback setting)
Mute all the speakers except for C and SW
Play the sweep from MC and Record the Sweep from Audacity

I would like to better the last step and play and record the sweep through ASIO from JRiver.

[stealth82]

Sorry, my mistake.  Not enough acronyms, apparently!

Nick.

yes, exactly. I'm referring to DRC as the homonym solution: http://drc-fir.sourceforge.net/

[mojave]

REW doesn't use ASIO, REW max allowed sample rate is 48kHz and I didn't see options for measuring a 5.1 system, just left/right and sub. Did I overlook the feature?

ASIO supported was introduced in REW in version 5.01 on 7/11/11. You can use any sample rate supported by your audio device and you can select which input and outputs to use. By only changing the output in the soundcard setup in REW, I can measure any speaker in my system.

[nwboater]

ASIO supported was introduced in REW in version 5.01 on 7/11/11. You can use any sample rate supported by your audio device and you can select which input and outputs to use. By only changing the output in the soundcard setup in REW, I can measure any speaker in my system.

Could you please explain the relationship between MC, REW and your soundcard for this testing? Since we can't send the REW sweeps to MC I assume you still use a pink noise file in MC and the RTA in REW with no direct interaction between the two programs. So it would seem that you are measuring the speakers without DSP in MC, thus not seeing the results of any EQ done in MC?

I'm confused (nothing new!)

Thanks,
Rod

[mojave]

I was just talking about using REW to measure your system to see what is going on before making changes. After you make changes, the RTA method is the only one that works for me and you can't use REW in ASIO mode anymore. In case you missed it, I made a guide on How to use the RTA feature of REW with JRiver.

[nwboater]

I was just talking about using REW to measure your system to see what is going on before making changes. After you make changes, the RTA method is the only one that works for me and you can't use REW in ASIO mode anymore. In case you missed it, I made a guide on How to use the RTA feature of REW with JRiver.

Thanks for the clarification. I had seen your helpful guide - thanks for your efforts on it!

I have been using the RTA in REW with the pink noise files for awhile now. Just wondered if there was some trick you were using that I missed.

I don't fully understand this whole convolution thing, but am wondering if there might be some way using it to get the REW Sweeps into MC. Then we could use REW to measure the system performance after corrections instead of only before. And of course we could use Waterfall, impulse and the full compliment of REW features.

Rod

[stealth82]

ASIO supported was introduced in REW in version 5.01 on 7/11/11. You can use any sample rate supported by your audio device and you can select which input and outputs to use. By only changing the output in the soundcard setup in REW, I can measure any speaker in my system.

Interesting but, still, it doesn't allow to play 2 speakers contemporaneously. e.g. center + sub

[JJJ]

To be more specific on the last post, I know how to measure my room, I know how to create filters in REW but I've not seen a set of instructions as to how to load a filter into JRiver.  Maybe I need to just play around some more in JRiver but I have access only via RDP on an iPad...  I do not want to purchase Audiolense.  I would rather apply the money to Theta's forthcoming implementation of Dirac Live.

[mwillems]

Read it and the threads it linked to but unfortunately (unless I missed something) it didn't really help at all. Maybe I need an idiot's guide.

What format are you generating the REW filters in?  How many channels are you trying to convolve at once?  

If it's just stereo you should be able to generate a stereo WAV filter that you can load directly into JRiver's DSP studio "Convolution" module.  If you're trying to do multichannel convolution you'll need to write a config file.  Convolution (in my experience) is complicated to do right, and easy to do wrong, but if you can provide more specific info I can talk you through some of the ins and outs.

One solution that bypasses convolution (and only sacrifices some functionality) is to look at the correction filters that REW is dialing in and to manually transcribe them into the parametric equalizer in JRiver.  You can find REW's proposed filters in REW's "EQ Filters" window: http://www.hometheatershack.com/roomeq/wizardhelpv5/help_en-GB/html/eqfilters.html#top

In the linked example, you'd just make a new entry in the JRiver DSP Studio parametric equalizer module for each of the those filters.  You'd probably want to choose "adjust a frequency," and then you'd enter the relevant frequency, gain, and bandwidth (or Q).  That would effectively replicate what REW is trying to accomplish without convolution.  That has some advantages (much less latency and processor use) and some disadvantages (can't correct phase as readily), but is much easier for multichannel setups.

[JJJ]

What format are you generating the REW filters in?  How many channels are you trying to convolve at once?  

Just stereo for now.  I'm dabbling while I wait for Theta to implement Dirac Live.

If it's just stereo you should be able to generate a stereo WAV filter that you can load directly into JRiver's DSP studio "Convolution" module.

........

One solution that bypasses convolution (and only sacrifices some functionality) is to look at the correction filters that REW is dialing in and to manually transcribe them into the parametric equalizer in JRiver.  You can find REW's proposed filters in REW's "EQ Filters" window: http://www.hometheatershack.com/roomeq/wizardhelpv5/help_en-GB/html/eqfilters.html#top

In the linked example, you'd just make a new entry in the JRiver DSP Studio parametric equalizer module for each of the those filters.  You'd probably want to choose "adjust a frequency," and then you'd enter the relevant frequency, gain, and bandwidth (or Q).  That would effectively replicate what REW is trying to accomplish without convolution.  That has some advantages (much less latency and processor use) and some disadvantages (can't correct phase as readily), but is much easier for multichannel setups.

Thanks.  I've made it as far as observing the EQ Filters window and studying "house curves".  I was not sure how to then take the information I see in REW and get it into JRiver. When I save the filters I do not get a file with the suffix .WAV.

[mwillems]

Just stereo for now.  I'm dabbling while I wait for Theta to implement Dirac Live.

Thanks.  I've made it as far as observing the EQ Filters window and studying "house curves".  I was not sure how to then take the information I see in REW and get it into JRiver. When I save the filters I do not get a file with the suffix .WAV.

What suffix are you seeing?  You may need to change your export settings.  When I get home I'll open up REW and try and be more specific.

[JJJ]

Let me respond when I get home   

[JJJ]

What suffix are you seeing? 

.req

It's not editable in a text editor

[mwillems]

.req

It's not editable in a text editor

.req is just REW's storage format, not a usable filter format.  Under the file menu in REW, look for the option to export the filters as a WAV.  You'll want to make sure that you're exporting the filters as a stereo WAV and not a mono WAV.

http://www.hometheatershack.com/roomeq/wizardhelpv5/help_en-GB/html/file.html#top

Once you've got a .WAV saved, try loading it in JRiver's DSP studio convolution module.  One important thing to keep in mind is that convolution filters are sample rate specific.  A convolution filter generated using a 44.1 KHz sample rate will not sound right with 48 KHz files. There are two solutions to that problem:

1) Pick the sample rate you use most often, and generate your REW filter using that, and then use JRiver's DSP engine to resample everything to that sample rate (this sometimes introduces problems, but works for some folks), or

2) Make a different convolution filter for every sample rate you plan on using, and then use JRiver's automatic convolution sample rate switcher (a check box in the convolution module).   In order to work, the switcher requires that the files be named in an extremely specific fashion:

xxxx2.0_441
xxxx5.1_48
etc.

Where:
xxxx = Arbitrary Filter Name
2.0 or 5.1 = number of channels, and
441 or 48 = Sample rate

If that all starts to sound like a lot of work (and it can be to get it right), you can try the previous approach I mentioned (above) and just manually transfer the EQ filters and see how that sounds before worrying about phase.  Phase can be important, but uneven frequency response is a much more noticeable issue in my experience.  

Let me know if you hit more roadblocks, I'm happy to help!

[JJJ]

Hey thanks a lot.  I will give this a solid look over the weekend! I guess I will have issues generating a sweep at higher bit rates to generate filters for high-res audio but 44.1 and 48 kHz should not be a problem.

Very much appreciated!