Print

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

[NickF]

This is new territory for me but I am really fascinated and want to give it a try.  The trouble is, I don't think I completely understand the process for getting things set up.  The one thing I have done is to order a measurement mic.  I have gone for the MP-1r-KIT Measurement kit from Juice.  Bernt has confirmed that these are individually calibrated.  I now need to know what software I need to do the setup.  Many products have been mentioned in this thread.  Could one of you experienced guys outline the steps and what product is needed at each step?

Nick.

[hulkss]

Could one of you experienced guys outline the steps and what product is needed at each step?

You need MC plus a Digital Room Correction program.

The leading options are:
Audiolense http://www.juicehifi.com/index.html
Acourate http://www.acourate.com/
DRC http://drc-fir.sourceforge.net/
Ultimate Equalizer http://www.bodziosoftware.com.au/

Another solution, more of an automated equalizer:
Room EQ Wizard http://www.hometheatershack.com/roomeq/

As far as I know they all achieve good results. They will not achieve the same results, even from the same measurement.
So your results may vary. The most important thing to remember is it all comes down to developing a target response for your system in your room that you like. So you end up depending on your ears.

The posts referenced here are a great read on how to get good results and combine your ears with verification measurements.
http://yabb.jriver.com/interact/index.php?topic=68828.msg463993#msg463993

[psam]

This is new territory for me but I am really fascinated and want to give it a try.  The trouble is, I don't think I completely understand the process for getting things set up.  The one thing I have done is to order a measurement mic.  I have gone for the MP-1r-KIT Measurement kit from Juice.  Brent has confirmed that these are individually calibrated.  I now need to know what software I need to do the setup.  Many products have been mentioned in this thread.  Could one of you experienced guys outline the steps and what product is needed at each step?

Nick.

I don't consider myself experienced on this, but some basic steps are:

1) You hook up your microphone to your PC and put it at your listening position / Lazy Boy or whatever

2) You start the software of your choice (Audiolense in your case I presume)

3) Perform a new measurement using the buttons on the right. The software will start a whooping sound (frequency sweep) from the very low to the very high frequencies through the speakers, while the expected sound pressure level is considered to be the 0db point.

This takes place once for each speaker. The microphone picks up the resulting sound as it reaches your listening position and the software records it.

4) Using the software, an analysis of the measurement is used to determine the equalisation required to improve the sound in your listening area. For example, if there is a dip of 6db at 2.3 KHz, this means that vocals may be sounding hollow.

Aiming to eliminate this anomally, a correction is required. In detail, a boost of 6db at 2.3 KHz, if our goal is to achieve 0db.

These infinite equalisations across the frequency spectrum can be viewed like a graph.This graph (with some limitations) is like an inverse mirror of the measurement graph.

The result is called a filter and this filter (a PC file actually) is saved on your PC.

5) Using MC17, you go to Player-> DSP Studio -> Convolution and check its box.
In Settings of the same pop up window, you select the filter file you have created, and from that point onward MC17 applies this correction to all music you play.


This is simplistic, but I hope you get a rough idea!

[Trumpetguy]

There are also inifinite number of things you can do and traps to fall into when doing the measurement. Two nice things to consider, which I found important:

1) In a system with center channel, the mic will most probably be located directly in front of the speaker, with the sensor in your nose-tip location and yourself outside the room. The back of your chair will reflect significantly, and results impulse response which is (almost the) magnitude of the original signal. And it will be 1-3ms delayed. This should be avoided, as it does not reflect the room response when you are in the room. It also creates results in strange filters. Use an absorber to replace your body.
2) I have a very difficult room, and filters based on sweet spot only measurements tend to result in pre-ringing anomalies. Creating the filter based on multiple measurements in a 3D array around the sweet spot gives a tremedously better result. E.g. sweet spot + 30cm away in all directions (back/in front/left/right/above/below). Audiolense supports multiseat filters.

These are just some things found by experience, there are more.

Then you will need to spend some time finding the targer response of your taste. My preference for music is a 3dB slope from the bass roll-off to 20kHz. For movies, using LFE and two subs, the bass rolloff is at a lower frequency, and I use an almost 8dB linear slope.

[NickF]

Thanks for the advice guys.  I am looking forward to getting started but face a frustrating 2 week wait for the measurement mic to arrive!

A couple of questions:

Does the Audiolense created filter and convolution take account of delays due to speaker distance or is this done separately in "Room Correction" in MC?

I assume, from comments above, that I can set up a different filter for music and movies.  Are these filters selected automatically and, if so, how?

Nick.

[Trumpetguy]

Thanks for the advice guys.  I am looking forward to getting started but face a frustrating 2 week wait for the measurement mic to arrive!

A couple of questions:

Does the Audiolense created filter and convolution take account of delays due to speaker distance or is this done separately in "Room Correction" in MC?

I assume, from comments above, that I can set up a different filter for music and movies.  Are these filters selected automatically and, if so, how?

Nick.

AL will automatically correct for distance for the individual speakers. You can manually edit the correction in terms of ms delay if required. I have never found that necessary.

You can define different zones in MC. The convolver can be set up with individual filters for individual zones. This works really good!

[candycane]

Do folks think it's OK to try to correct all dips and peaks with DRC, or is there a rule-of-thumb that says a dip or peak correction beyond a certain number dBs is better or must be done with room treatment?  I think I get it that room treatments with DRC is optimal in general, but given the range of options it would be great for some tips on what kinds of correction can't be handled well with DRC and room treatment is essentially mandatory.

Related to that, I have a vague understanding that if a peak or dip is related to a room mode (resonant frequency of a room, generally in the bass region), then room treatment is the only way to address this, but knowing when a dip or peak is related to a room mode or not is beyond my current understanding.

 thanks for any insight!

[NickF]

I am no expert in this but I had assumed that this technique, as it is end-to-end, will compensate for frequency and time based deficiencies anywhere in the chain, whether amp, speakers or room.  I guess where I am a bit nervous, not having mine set up yet, is that I may lose some of the "warmth" of my Tannoys.  I suppose time will tell.  But why, otherwise, wouldn't we want to correct for any deficiencies?

Nick.

[NickF]

I just had a thought about steps in the process for setting this up.  I am sure many of us have receivers of one type/brand or another, and many of these have mechanisms for compensating for room and speaker setup.  My Denon certainly does.  So we need to disable any setup of the receiver and have all of the compensation in MC Convolution.

Nick.

[NickF]

Another thought - I shall need to go and have a lie down!!  I have seen mention of taking measurements at mutiple positions in the listening room.  The Audyssey system on my Denon supports this although I have never used it.  How can this work?  Surely, the system can only create a correct audio image for one point (the sweet spot) in the room.  What is the thinking behind this and how can it work?

Nick.

[Trumpetguy]

Another thought - I shall need to go and have a lie down!!  I have seen mention of taking measurements at mutiple positions in the listening room.  The Audyssey system on my Denon supports this although I have never used it.  How can this work?  Surely, the system can only create a correct audio image for one point (the sweet spot) in the room.  What is the thinking behind this and how can it work?

Nick.

That was me. The short answer is that multiposition measurements do create one filter, but averages the correction.

A room will give slightly or very different response at different locations. My room is difficult, and even 30cm from the ideal sweet spot, bass response in particular is very different. Ideally, if you are the only one listening, you should make one measurement in the sweet spot and create the filter. In difficult rooms with lots of nodes and bass problems, single-point correction can create quite unwanted anomalies at and outside the sweet spot. In my difficult room, the DRC will then try and compensate a lot in the 50-80Hz range. The resulting filter is not good, and gives pre-ringing in the bass (sounds like swoooopp! for certain sounds, like kick drums). To avoid this, I had to 1) find the optimum placement of speakers and sofa, 2) create 1 (one) filter made from multiple measurements in a grid around ideal sweet spot. Step 2) is what is called multiseat (Audysseay does that, as do Audiolense), and is usually intended to create a filter that averages the correction such that several people at least have some kind of audio improvement. The recommendation is then to measure in a quite large area of the room. My step 2) involves measurements up to 1m from sweet spot, and works really well in sweet spot and for the two adjacent seats.

[Trumpetguy]

Do folks think it's OK to try to correct all dips and peaks with DRC, or is there a rule-of-thumb that says a dip or peak correction beyond a certain number dBs is better or must be done with room treatment?  I think I get it that room treatments with DRC is optimal in general, but given the range of options it would be great for some tips on what kinds of correction can't be handled well with DRC and room treatment is essentially mandatory.

DRC can not make magic. Narrow freq-band dips are not disturbing the perception of audio, and you should not worry much about these. A general rule of thumb would be to not try and increase dips more than around 6dB, even if I have seen people reporting good results with 9 and 12dB increases. A 6dB increase of a dip means you have lowered all other frequencies by 6dB, i.e. halved your power amplifier's headroom. Also, remember that the x dB increase has to be played by the speaker. You can get really unwanted results from your speaker if you force it to play 6, 9, 12 dB higher at one or several narrow dips.

Peaks is a different story. A typical room often have several dB peaks in the 40 to 50Hz range. This is easily (?) flattened out with DRC. In my room, this 40Hz peak is accompanied by a broad dip from 50-80Hz, which cannot fully be brought up in a 2.0 setup. With the immense power of two subs in the 7.1 setup, this is less of a problem, as the two subs can be placed such as to minimize the dip in the first place. And then DRC uses some head room of the subs to increase the response up to the target curve.

Related to that, I have a vague understanding that if a peak or dip is related to a room mode (resonant frequency of a room, generally in the bass region), then room treatment is the only way to address this, but knowing when a dip or peak is related to a room mode or not is beyond my current understanding.

Again, peaks can the lowered with good results. Dips, and especially when you are at or close to a null, should not be corrected with DRC. Despite the name, bass traps can be helpful in increasing bass response in a room, as they absorb sound waves that otherwise would be reflected and would cause negative interference.

[hulkss]

There are also inifinite number of things you can do and traps to fall into when doing the measurement. Two nice things to consider, which I found important:

1) In a system with center channel, the mic will most probably be located directly in front of the speaker, with the sensor in your nose-tip location and yourself outside the room. The back of your chair will reflect significantly, and results impulse response which is (almost the) magnitude of the original signal. And it will be 1-3ms delayed. This should be avoided, as it does not reflect the room response when you are in the room. It also creates results in strange filters. Use an absorber to replace your body.

I have four separate seats that are not massive in a big room (sound can travel under and between them). I get good results if I recline the center seats to avoid this reflection problem (measure where your head is when setting more upright).

I get even better results if I remove the seats and then average multiple measurements in the center seating area. This way I am correcting for mainly the lower frequency room boundary effects without irregular short seat surface reflections screwing up the measurements at higher frequencies. Those short reflections change totally if the seat is occupied, empty, reclined, etc. Removing that variable completely seems to work in my case. I don't hear the difference if someone sits next to me so I don't want to screw up the direct response of the speakers in an ineffective attempt to compensate for a seat reflection.

[NickF]

Yippeee!  My mic has arrived!  It should be an interesting weekend getting my system set up.

Trumpetguy and hulkss, surely the measurements should be done with the listeners in their seats, otherwise you will be compensating for effects which won't exist in a real situation.  You have left me somewhat confused.  ?

Anyway, that sort of fine tuning is somewhere down the road.  I'll start with the basics and get used to Convolution first.

Nick.

[hulkss]

surely the measurements should be done with the listeners in their seats, otherwise you will be compensating for effects which won't exist in a real situation.

You should attempt to correct for low frequency room effects (boundary reflections). At higher frequencies (above the Schroeder frequency) correct the direct sound from the speakers. Get anything away from the measurement mic that would cause a short distance reflection into it and screw up your measurements. People don't reflect much bass, they are too small.

Time windowing of your measurements is needed with a shorter window at high frequencies.

[Trumpetguy]

Trumpetguy and hulkss, surely the measurements should be done with the listeners in their seats, otherwise you will be compensating for effects which won't exist in a real situation.  You have left me somewhat confused.  ?

I have seen that this topic can cause debate temperature to rise , but personally I do not have any strong reasoning for either/or. I did some early test in/out of the room, and it didn't make a noticable difference so I leave the room simply because this is simpler. I find it difficult to place the mic in the middle of my own head  , that is one thing. Placing it just beside will cause reflections, causing impulses which will disturb high frequencies simply because of the close proximity to the mic. On the other hand, your body would absorb and reflect energy, an effect that could be relevant for your filter. You will need to figure this out yourself.

Anyway, that sort of fine tuning is somewhere down the road.  I'll start with the basics and get used to Convolution first.

Wise strategy

[AudioVero]

Please assume a live event in your living room, a band is playing nicely. The arriving sound will float around you, reflected by ears, shoulders and the body. Despite this influence on the sound I'm pretty sure you will here a perfect live event and be happy with the (assumed nice) sound. Because your brain has learned to interprete the influence of your body and your ears. It simply ignores it or even does not recognize it.

Now assume that you are able to measure with a microphone in the middle of your head. It will record all the influences of your body on the sound. A later correction will compensate for the influences.  Thus the sound of a speaker reproducing perfectly the live event of the band will be altered by the correction. And you will hear a wrong sound !

IMO it is not correct to record a sweep sitting in the chair and with te microphone in the mouth or in the ears.

[Trumpetguy]

IMO it is not correct to record a sweep sitting in the chair and with te microphone in the mouth or in the ears.

LOL, you could be locked up or severly hurt to put a microphone into your head, regardless of opening...

Thanks for the reasonable argument on why the room should be left empty during measurement.

[NickF]

Well I'm on the verge of giving up already!  I have downloaded both Acourate and Audiolense to try them.  I rashly thought that I would get a 30 day trial of a functioning product, like we do with JRiver.  No such luck.  Audiolense produces a file but it contains no correction.  What is the use of that?  I opened Acourate and could make no sense of it.  At least I was able to make progress with Audiolense so I decided to buy it - 250 Euros!!  Now I can't even activate it.  The product wants to send an email with the hardware id, whatever that is.  My HTPC doesn't have an email client - why would it?  So I have now spent 411 Euros and have got nowhere.  I can't even post a message asking for help on the Audiolense Forum without the topic being approved by a moderator.  What a mistake!!

It makes you realise what a brilliant organisation JRiver is.

Nick.

[JimH]

Thanks, Nick, but I'm sure Audiolense will find a way to get you going.

[lasker98]

Nick,

I had the same issue installing Audiolense since the computer had no internet connection. Here's the instructions I received:

"Next step for you is to install Audiolense and generate a HardwareID for license activation. You can generate a hardware ID from the license menu or by running AudiolenseHID.exe that sits in the program folder."

Once you have that generated, send it to Audiolense and you'll receive your activation.

Good luck,

Bill

[NickF]

Nick,

I had the same issue installing Audiolense since the computer had no internet connection. Here's the instructions I received:

"Next step for you is to install Audiolense and generate a HardwareID for license activation. You can generate a hardware ID from the license menu or by running AudiolenseHID.exe that sits in the program folder."

Once you have that generated, send it to Audiolense and you'll receive your activation.

Good luck,

Bill

Thanks Bill, I'll give that a try now.

Nick.

[Trumpetguy]

I believe Audiolense trial is fully functional (once you get it going, and you will it is just not 100% intuitive...), but with a heavily attenuated filter. Some 10 or 20dB. Should make you able to evaluate it, but you will of course get limited dynamics.

[hulkss]

Well I'm on the verge of giving up already!  I have downloaded both Acourate and Audiolense to try them.

Nick,
Don't give up, the rewards are too great. These two programs do work and produce excellent results, however, they are also capable of making a mess. You will need to do a lot of reading before you understand what you are actually doing. I can not listen to audio without digital correction now - it is amazing when you get it right.

Start reading here: http://yabb.jriver.com/interact/index.php?topic=68828.msg463993#msg463993
and then some about room acoustics and technical digital filter stuff if you can handle it. Just developing a good target response is a fair amount of work and has a profound impact on results.

[AudioVero]

Well I'm on the verge of giving up already!  I have downloaded both Acourate and Audiolense to try them.  I rashly thought that I would get a 30 day trial of a functioning product, like we do with JRiver.  No such luck.  

Unfortunately it is not possible to get filters that run as trial filters for 30 days. No chance that a filter stops working after a trial period. If you know such a solution then please tell me.
As a previous full working Acourate trial version also has been hacked in the past I have been forced to create just a "dummy" version that does not allow to create the final filters.

So now there is a free logsweep recorder (www.acourate/AcourateLSR2Setup.exe) and I offer to compute filters based on the measured pulse responses and to treat two stereo music tracks of your own choice. This allows you to listen to the filtered tracks and to compare the result with the original tracks or other correction products. This service is free of charge. Isn't that fair?

[NickF]

Unfortunately it is not possible to get filters that run as trial filters for 30 days. No chance that a filter stops working after a trial period. If you know such a solution then please tell me.
As a previous full working Acourate trial version also has been hacked in the past I have been forced to create just a "dummy" version that does not allow to create the final filters.

So now there is a free logsweep recorder (www.acourate/AcourateLSR2Setup.exe) and I offer to compute filters based on the measured pulse responses and to treat two stereo music tracks of your own choice. This allows you to listen to the filtered tracks and to compare the result with the original tracks or other correction products. This service is free of charge. Isn't that fair?

Sorry Uli, I didn't mean to be rude, I just found myself a bit bewildered and disappointed.  Yes, I appreciate the problems of time expiry of filter files.  I didn't realise you offered this service.  It does seem very reasonable.  I may try your logsweep recorder as I get more experienced.  Thanks.

Nick.

[NickF]

Well, I'm up and running, or should I say walking!!  It's a bit of a steep learning curve but I've made a start.  Bernt was quick off the mark this morning with my activation file.  Thanks Bernt.

I'm starting with stereo for my music.  I have taken Trumpetguy's advice and set up a Zone for Movies and one for Music so I can use different filters.   I can switch between these automatically with my iPad remote. 

I have done my first set of measurements after much fiddling with mic mounting and positioning, connections to my RME sound card and adjustments to volume and pre-amp gain.  

I created a target, probably one of the most critical steps, I think.  I really like the sound of my Tannoy Berkleys.  They are vintage speakers but have been re-coned and I have upgraded the crossovers with high spec components and they are superb.  I have shaped the target to retain some of their natural behaviour.  

I produced the correction file and loaded it into Convolver and...... it works!!  Even this first attempt reveals a much sharper and more detailed sound stage with so much less clutter from the room reflections.  I am now having a problem dragging myself away from listening to go back to tuning!

Thanks for all the help and advice so far.

Nick.

[NickF]

I'm a firm believer in quality interconnects.  So how important are the interconnects between mic and pre-amp and pre-amp and sound card?  The mic to pre-amp is about four metres long so that could get quite expensive!

Nick.

[hulkss]

I'm a firm believer in quality interconnects.

Quality interconnects are important. I will give you an Engineer's recommendation, not Audiophile.
This is what I do:

1. Use all balanced interconnections. Read this: http://www.jensen-transformers.com/an/an003.pdf and this: http://www.rane.com/note110.html
2. Use brand name professional cable. I use Mogami: http://www.mogamicable.com/additional/best_cable.php
3. If you want to make your own cables to custom lengths, use Mogami with connectors from: http://www2.neutrik.com/us/en/audio/203_11/XLR_Cable_Connectors_group.aspx
4. Use hardware that takes full advantage of balanced cables. I recommend this design developed by Bill Whitlock of Jensen Transformers: http://www.thatcorp.com/datashts/THAT_1200-Series_Datasheet.pdf

I use two of these (shorter length) to connect my 16 channel DAC to the amps:

[Trumpetguy]

hulkss, du you use the original breakout cables from Lynx, or are these some other brand? The original cables are probably splendid, but they seem kind of thin and unprotected (don't know how to put it).

A side question: I have unused input and digital connectors hanging which are not terminated. Should there be some end resistor or other type of termination of these?

[NickF]

Well it's been an interesting day playing with this solution.  There have been some distinct audible improvements in the stereo image of my music and improved clarity of very low frequencies.  I have the CD 'This Fire' by Paula Cole and the first track, Tiger, has some incredible low frequencies.  Without Convolution, you feel the bass but with Convolution, you can clearly hear the cadence and shape of the sound.  But as I continued to listen, it became clear that there is a distorted edge to many sounds, a harsh, fuzziness.  I have swapped back and forth between Convolution and the pure path through MC and there is no doubt which I prefer.  Something is wrong with Convolution.

Tomorrow I will try ConvolverVST to see if that is any better but, at the moment, I couldn't live with Convolution.  Sorry Guys.

Nick.

[hulkss]

hulkss, du you use the original breakout cables from Lynx, or are these some other brand?
A side question: I have unused input and digital connectors hanging which are not terminated. Should there be some end resistor or other type of termination of these?

I bought these cables: http://www.markertek.com/Cables/Audio-Cables/ADAT-Cables/TecNec-Cables-Connectors/25MA-XM-M05.xhtml?25MA-XM-M05
I'm sure the Lynx ones are fine.

Never seen anyone use xlr terminator caps, you can buy them though.

[hulkss]

Something is wrong with Convolution.

Nick,
More likely something is wrong with your filter. Poor filter = poor convolution result.

I hope you have a decent target response curve. Read this: https://docs.google.com/viewer?a=v&pid=explorer&chrome=true&srcid=0B97zTRsdcJTfY2U4ODhiZmUtNDEyNC00ZDcyLWEzZTAtMGJiODQ1ZTUxMGQ4&hl=en_US
Note that the best sounding response (top line p24) has a full 10 dB decrease from 20 - 20kHz.

Depending on the power response (not frequency response) many loudspeaker systems need a "midrange compensation dip" of a few dB centered at 2 kHz to remove harshness. Keep in mind that the microphone and computer do not hear like your ear and brain, the target curve has to compensate. A flat target will sound very lame.

[NickF]

Thanks, hulkss.  It's very early days for me so I haven't reached any conclusions yet, just first impressions.

I spent quite some time investigating and shaping my target.  It is 10dB down across the spectrum so I'm fairly sure I have enough digital headroom for the correction to avoid clipping.  Is there any way to show whether MC is seeing or dealing with clipping?

I will take another look at the shape of, it today.  The effect that I am hearing isn't due to an over-emphasis of mid-range frequencies.  It is a distorting effect added by the process.  It manifests as a fizz on the edge of purish mid-range sounds, particularly female vocals.

I will take some more measurements today, look more closely at the target and do some comparisons between different convolution engines.

Nick.

[AudioVero]

Please note that the correction counts but not the target !

Now
correction = target - (interpreted) measurement

This means that depending on the interpretation of the measurement (like smoothing, frequency dependent windowing and other treatments) also a target must change to get an identical correct correction. It does not help to read articles as mentioned above because each correction algorithm has its own philosophy to interprete the measurement (indeed this also causes different targets and results).

So you should always study the correction curve. If it has a rising slope then you should not wonder about more brightness. If it is strongly falling you should not wonder about a dull sound.
Try to understand what the correction is changing, the correction curve show this clearly.

[NickF]

Thanks for the advice, Uli.  As a complete novice, I welcome all input. 

I had assumed that the target, a component of the process over which the user has complete control, would be an important area to focus.  Surely it is intended to drive the final behaviour of the solution.  I assumed that the correction, created by the tool, not the user, is essentially a response to the target, which is what your formula says.

So you are saying study the correction and understand what effect it will have on the solution.  Are you also saying that we should modify the correction if we think it is necessary?  I am using Audiolense.  Does it allow me to modify the correction?

Nick.

[AudioVero]

So you are saying study the correction and understand what effect it will have on the solution.

Yes.

 Are you also saying that we should modify the correction if we think it is necessary?

Yes.

I am using Audiolense.  Does it allow me to modify the correction?

Yes, modify the target accordingly.

[Trumpetguy]

Thanks for the advice, Uli.  As a complete novice, I welcome all input. 

I had assumed that the target, a component of the process over which the user has complete control, would be an important area to focus.  Surely it is intended to drive the final behaviour of the solution.  I assumed that the correction, created by the tool, not the user, is essentially a response to the target, which is what your formula says.

So you are saying study the correction and understand what effect it will have on the solution.  Are you also saying that we should modify the correction if we think it is necessary?  I am using Audiolense.  Does it allow me to modify the correction?

Nick.

With AL you should also play around with window widths in the filtering section. Different windowing (number of cycles) give different correction results. You should make sure your correction curve does not have any nulls (or very sharp and deep dips).

[hulkss]

If your not doing digital XO, here'e the Audyssey Multeq Pro product in a VST plug-in.
http://www.ikmultimedia.com/arc/features/

I've used Multeq Pro with good results. Their "secret" is to combine a number of measurements to average out measurement anomalies. This also applies to DRC software solutions.
http://www.audyssey.com/sites/default/files/attachments/onesheet_howtomulteq.pdf

You can combine multiple measurements in Audiolense and see the effect in the simulated response. Multiple measurements in different central seating locations will help with measurement confidence, understanding of room response issues, and should help to provide a better overall filter solution.

Another very good article, it's 21 years old and describes the digital correction challenges we are just beginning to work now in our home theaters.
http://www.audiosignal.co.uk/Resources/Digital_room_equalisation_A4.pdf

[Mitchco]

I had assumed that the target, a component of the process over which the user has complete control, would be an important area to focus.  Surely it is intended to drive the final behaviour of the solution.
 
Nick.

Nick, yes, the target is the key.  I have found the following target from this excellent paper: http://www.bksv.com/doc/17-197.pdf to have the most natural timbre at the listening position.  The target points are -0.5db at 200Hz, -3 at 2KHz and -6 at 20KHz.



Using REW http://www.hometheatershack.com/roomeq/ (awesome measurement software and fantastic support on their forum, just like JRiver!) I measured the following frequency response at the listening postion with the Audiolense filters loaded into JRiver's fantastic Convolution:



As hulkss has mentioned, a flat frequency response at the listening position will sound (etch a sketch) bright.  The B&K target curve above produces the most natural timbre and soundstage at the listening position IMHO.  I have tried many variants of this target and end up coming back to this one as being the most natural sounding to my ears. YMMV.

I suggest you head over to the Audiolense forum http://groups.google.com/group/audiolense as there are many tips being shared that you may find beneficial.

Happy Listening!

Mitch

[hulkss]

Very nice result. From 40 hz to 20kHz a 10 dB tilt is what your measurement shows and is almost identical to the paper I referenced by Harman.

[Mitchco]

Very nice result. From 40 hz to 20kHz a 10 dB tilt is what your measurement shows and is almost identical to the paper I referenced by Harman:

hulkss, thanks.  I had a look at the Harman paper you referenced: https://docs.google.com/viewer?a=v&pid=explorer&chrome=true&srcid=0B97zTRsdcJTfY2U4ODhiZmUtNDEyNC00ZDcyLWEzZTAtMGJiODQ1ZTUxMGQ4&hl=en_US  Slide 24, the top curve:



And yes, it is almost identical - good correlation to 2 different papers showing the same preferred target response.  Btw, combining the left and right responses = flat at 20Hz.  So relative to the target, I am easily within +-3db from 20Hz to 20KHz.  I am so satisfied with the response of my system, I am done.  Just listening to music now.  With Convolution built-in to JRiver, I try and feed everything through it - sounds great!

[Trumpetguy]

Truly a nice (or impressive) frequency response. Would you care to post impulse response and/or ETC as well? Is your room acoustically treated, or do you do huge digital enhancements?

[NickF]

Well I have made some progress.  Firstly, thanks for all the support and encouragement!

The distortion - I suspected there was something going on within the solution to cause this so I decided to drop the whole target by 6 dB.  By the way, thanks Mitchco, I have used your B&K target as the basis and like it.  This improved the situation but the problem was finally resolved by turning off Clip Protection and, epecially, Normalise filter volume.  The latter is a killer, it completely destroys the quality.  With these corrections, the distortion is now gone.  The problem is that it raises the noise floor by 6 dB, not that this is a real problem, but I now need to go back to the target and see what the minimum reduction needs to be.  That will take a few iterations.

I note that hulkss has posted something on clipping within the Convolution path and that clip protection is currently applied at the end rather than during.  Are you hearing something too?  My view is that clip protection is attacking an effect, not the cause.  You need to ensure that you never reach a clipping situation.  However elegantly clip protection works, it is no different from AGC and that is to be avoided at all costs.  The original recording fitted within the digital sample ceiling and you need to keep it there.

One thing about Audiolense which is confusing me at present is the shape of the correction filter.  There are some dips in my room/speaker response which it is trying to correct but the correction waveform flattens out in several places as though it hits a max threshold for correction.  I thought that dropping the target would help this.  In theory, it should but it doesn't.  Anyone know why this is?  I can't find any option to adjust it.  Is it a defect or is there some other reason?

Well my enthusiasm is back up and, hopefully, over the next few days, I will get my stereo solution right and I can move on to the surround correction for my movies.

Nick.

[hulkss]

There are some dips in my room/speaker response which it is trying to correct but the correction waveform flattens out in several places as though it hits a max threshold for correction.  I thought that dropping the target would help this.  In theory, it should but it doesn't.  Anyone know why this is?  I can't find any option to adjust it.  Is it a defect or is there some other reason?

The amount of + correction in the Audiolense filters defaults to a limit of +6 dB. This can be changed in the correction procedure designer (maybe not in all software versions). This needs to be increased with caution.

Clipping is bad for sure. The JRiver clip protection is not bad though. I believe it will only mess with the audio if you are actually clipping. Something is off with the gain levels in DSP somewhere.

Normalize filter volume should be OK too. Maybe Matt can comment. I don't have issues with clipping or the convolver normalization. I was pointing out that I could see levels above 100% in the DSP chain yet I was not activating clip protection. Matt explained that the floating point processing allows this without problems. The signal has to be back down below 100% only at the last step in the chain.

[Mitchco]

Truly a nice (or impressive) frequency response. Would you care to post impulse response and/or ETC as well? Is your room acoustically treated, or do you do huge digital enhancements?

ETC’s using Audiolense TDD: http://www.hometheatershack.com/forums/rew-forum/51865-thank-you-rew-2.html#post477314

Latest ETC, but using only frequency correction: http://i1217.photobucket.com/albums/dd381/mitchatola/jan29etcfrequencycorrection.jpg

What is interesting to note when you compare the TTD versus frequency correction only ETC’s, is that the TTD does a good job of supressing the early specular reflections.

Waterfall  Full range: http://i1217.photobucket.com/albums/dd381/mitchatola/jan29waterfallfullrange.jpg

Waterfall 15Hz to 300Hz: http://i1217.photobucket.com/albums/dd381/mitchatola/jan29waterfall15to200hz.jpg

You can read the details, plus more graphs: http://yabb.jriver.com/interact/index.php?topic=68828.msg463993#msg463993

No acoustic treatment, save a heavy carpet between the listening position and speakers.   Not because I don’t think acoustic treatment works, it does, and I have experience with acoustic treatment as can be seen here:  http://www.computeraudiophile.com/blogs/Hear-music-way-it-was-intended-be-reproduced-part-5   The issue for me is cost.   To really get the room right using acoustic treatment costs in the thousands…

I use 12db of correction, but could get away with 6db and get -3db at 30Hz.  I have tried up to 18db of correction, but settled on 12.

Hope that helps.

[Trumpetguy]

Again, the ETCs are just as impressive as your frequency responses. No impulses above -15dB before ~20ms. And all that without any particular acoustic treatment. I believe you should consider yourself lucky with your room. Or maybe luck has nothing to do with it?

With 12dB increase, do you experience any loss of dynamics? Audiolense has a default setting of max +6dB. It sure helps explaining the straight frequency response (in addition to the graph smoothing), but I would expect some really noticeable loss of headroom. I know I would in my room. With a room with narrow-band dips, and even having sufficient amplifier power, it will be a really challenging task for a speaker to play +12 and even +18dB louder in these narrow frequency ranges. How do you solve these things?

Only asking because you seem to have solved somthing that I have been struggling with.

BTW, I tested some target curve close to the B&K curve yesterday. Sounds very different from what I am used to, but it created a really pleasant, warm, still dynamic soundscape. Flat -0.5dB <200Hz, then sloping to -10dB at 20kHz. It is a real challenger to my old near-flat target.

[AudioVero]

BTW, I tested some target curve close to the B&K curve yesterday. Sounds very different from what I am used to, but it created a really pleasant, warm, still dynamic soundscape. Flat -0.5dB <200Hz, then sloping to -10dB at 20kHz. It is a real challenger to my old near-flat target.

Can you post some pictures of the correction for your old target and the B&K target?

[Mikkel]

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 

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).

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.


Best regards,
Mikkel

[Trumpetguy]

Can you post some pictures of the correction for your old target and the B&K target?

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?