INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: Adaptive Limiter in Parametric EQ  (Read 7487 times)

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Adaptive Limiter in Parametric EQ
« on: January 20, 2012, 11:24:47 am »

I couldn't find anything in the Wiki on how to use the Limiter. I looked up some other limiters and found info on the attack and release, but not the hold. Is there a description for each setting available? Knowing what they do would help implementing the limiter.

Here is my scenario:  I have a subwoofer that is able to get an actual 500 watts from my amp. The subwoofer will not reach maximum excursion with that power until below 30 Hz. Below 30 Hz I need to use the limiter for content that is above -10dB.

The limiter seems like it works on anything going over 0 dB. Is there a way to specify or use it for anything over -10dB, for example?

I would like to use the limiter on frequencies below about 30Hz. The only way I can think to do it is to move the subwoofer channel to User1 and User2, run a low pass and limiter on User1 and high pass on User2. Add User1 and User2 back to the subwoofer channel. Would this work? It would be easier if we could specify the upper and lower frequencies we want to use in the limiter itself.
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Adaptive Limiter in Parametric EQ
« Reply #1 on: January 20, 2012, 11:33:22 pm »

The limiter works in the time domain (on the waveform) as opposed to the frequency domain.  That means you can't really dial in frequencies.

However, the two are closely related.  If you limit the waveform, you're effectively limiting the excursion at low frequencies because this is the only signal that would trigger the limiter.

I have about the same problem as you at home, and haven't figured out the best way to handle limiting.  Currently I'm just letting 'Clip Protection' take care of it, but it's not perfect because huge bass spikes can push the overall level of everything else down too much.
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Adaptive Limiter in Parametric EQ
« Reply #2 on: January 26, 2012, 04:47:56 pm »

I played around with Voxengo Elephant today. It is a VST limiter plugin. I learned a few things that may be helpful.

First, Elephant has the option to increase the input gain while simultaneously decreasing the output gain. With the current method with JRiver's limiter, the limiter only kicks in when the signal is clipping. By being able to increase the input gain and decrease the output gain by the same amount, you can push the content higher to make the limiter kick in. This allows one to limit the LFE channel by compressing everything over a certain amount. There is a nice wave graph in Elephant that shows you exactly how much of the signal is being limited.

Second, any limiting in the DSP studio is affecting the waveform regardless of volume level. I know this is self-evident, but it helped me think the issue through. In order to effectively mimic the limiter on a pro amplifier, JRiver needs to provide a limiter after the internal volume control. This allows the limiter to kick in only when the volume reaches a certain level. Rather than raising the input level in the limiter, the internal volume control would raise the input level. At a certain point the limiter would kick in and prevent one from sending too high of a signal to their gear.
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Adaptive Limiter in Parametric EQ
« Reply #3 on: January 26, 2012, 04:55:23 pm »

Second, any limiting in the DSP studio is affecting the waveform regardless of volume level. I know this is self-evident, but it helped me think the issue through. In order to effectively mimic the limiter on a pro amplifier, JRiver needs to provide a limiter after the internal volume control. This allows the limiter to kick in only when the volume reaches a certain level. Rather than raising the input level in the limiter, the internal volume control would raise the input level. At a certain point the limiter would kick in and prevent one from sending too high of a signal to their gear.

Imagine playing at 5% volume when some DSP or source sends a 500% spike through.  Should the output be capped at 5% or 100%? 

By design we do the volume last, so cap at 5%.

100% might be enough to blow your speakers.
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Adaptive Limiter in Parametric EQ
« Reply #4 on: January 26, 2012, 05:37:21 pm »

I thought that was caught by the clip protection.

This week I measured the frequency response and tested the maximum output of two Dayton SA1000 subwoofer amps. One had the frequency response adjusted using essentially a Linkwitz Transform and limiting done by Seaton Sound. These were implemented my means of changing resistors, etc. in the amp. I think I read it takes about two hours to modify one amp.

The other amp was stock. I was able to use the Linkwitz Transform in JRiver to match frequency responses exactly. It only took a few minutes of using the RTA function in REW while playing the REW generated Pink Noise PN file back through JRiver and making adjustments in the PEQ. I was not able to implement any sort of limiting that worked.

The modded amp was modified with a regular home theater system in mind. In this system you would calibrate your receiver so that the test tone playback at 0 dB on the volume control would be 85 dB (or 75 dB depending on receiver). During playback of bass intensive material, the limiting on the amp would start when we reached about -15 to -10 dB on the volume control depending on the bass levels in the movie. The drivers never made any funny noises and most of the dynamics were maintained. During our KC Subwoofer GTG a few weeks ago, this system had a maximum output level of 116 dB while other systems with more capable drivers and when calibrated the same, had a maximum output of over 123 dB.

This past week during testing of the stock amp the drivers would start reach maximum excursion on War of the Worlds and other low bass material during playback in JRiver. According to the internal volume control, we were at about -20 to -15 dB. If we pushed the volume any higher, we could have damaged things. With the other amp it was impossible to damage the drivers. With higher frequency bass, like the Jericho scene in Iron Man, you could turn up the gain with no problems.

This indicates to me that to duplicate what was done to the modded amp, I need post volume control limiting that only kicks in when the volume control gets to a certain level. This would allow one to set the limiting for their own situation and prevent issues with all or certain channels (like the subwoofer channel). The limiting needs to be based on the input signal to the amp and not the actual native waveform of the audio track.
Logged

MusicHawk

  • Citizen of the Universe
  • *****
  • Posts: 796
Re: Adaptive Limiter in Parametric EQ
« Reply #5 on: January 29, 2012, 12:37:58 pm »

Is the solution to give MC a multi-band audio compressor? (Limiting is a subset of more-versatile compressing.)

This is THE method to prevent bass pumping (and other problems), universally used by radio and TV stations , on-line broadcasters, recording studios, theaters, etc. It caught on in the 1970s as very expensive hardware (originally a $20K box), but today is usually done via software.

It would be WONDERFUL if MC could integrate a multi-band compressor.

Background...

The problem of overall level "pumping" on bass peaks used to be annoying -- and often quite audible -- on over-the-air radio and TV stations, because they MUST limit max audio/modulation per FCC regs (and physics). The final stage of the audio chain is a compressor, followed by a limiter. If this is applied to the entire audio stream, the loudest sound element tends to control the overall volume, and periodic peaks that tower above the average trigger the compressor and/or limiter and force the entire audio level to be briefly reduced. When these periodic peaks are the bass line of a song, the effect is continual pumping up-and-down of the rest of the sounds in the mix.

It used to be a minor problem because analog discs could not have especially strong bass -- if the groove swing was too wide, it either distorted the adjacent groove or required spacing grooves more widely, thereby limiting total playing time. So strong bass, which might exist in the original master tape, was always limited and/or rolled-off when mastering for LPs and 45s.

But the problem increased over the years. By the late 1960s LPs were becoming shorter due to fewer tracks (10 or 11 instead of 12 to hold the retail price by reducing royalty fees), which also meant the bass level on vinyl discs could be increased. Then when CDs arrived the bass limitation virtually disappeared (obvious on "modern" recordings, and sometimes a revelation when hearing an old recording re-mastered for digital -- the bass is back!).

But as recordings gained bass, the pumping problem for radio/TV stations got worse. It couldn't be handled by fixed audio-chain EQ, which if set to reduce excessive bass would reduce ALL bass, including on records that were already balanced. The solution was and is dynamic multi-band compression (rather than limited, because raising low levels is also desirable at times; a compressor can be set to only do peak limiting when desired).

In multi-band compression, the incoming audio stream is split by frequency range into 5 or more channels (some use 8 or even 10). Each channel is processed by a compressor, then all channels are recombined as final output. This degree of control allows bass to be controlled without affecting other frequencies. (A stereo stream requires dual multi-band processors, and each band is cross-controlled so identical compression is applied to L and R to prevent sound stage image wandering.)

When properly set up, the limiting/compressing effect is very smooth when applied via 5 or more frequency-specific channels. This approach both eliminates the bass-pumping problem AND creates much more consistent and rich audio across the wide range of mixes that radio stations (and music collectors) encounter.

There are commercial multi-band compressor plug-ins, usually sold to broadcasters (on-air and on-line), movie, video production and recording studios, but usable with MC (but at a notable price). A while back I wrote an article about how I use one such plug-in (intended for Winamp) with MC:
http://www.advisor.com/story/how-get-radio-station-sound-your-own-music-collection

But I'd much rather be able to use MC's internal processing than have to add plug-ins that aren't truly MC-aware. All MC needs is to add a multi-band comrpessor!  ::)

Note that a commercial product typically is complex, but an MC version needn't be. Professional devices have many, many, many adjustments, to tune the frequency bands of each compressor channel, to adjust trigger points, attack and release times, and a lot more. Some are truly necessary -- trigger levels and release times must be very different for pop vs. classical music. But it's possible to simplify to just a few controls and a few settings -- after all, a radio or TV station's audio processing, once set, is used for everything it broadcasts.

I use my MC + compressor setup for music, movies, TV, and find the result so consistently smooth that I forget about the processing -- until I hear a system that lacks it and wish it wasn't so "lame".


Logged
Managing my media with JRiver since Media Jukebox 8 (maybe earlier), currently use Media Center for Audio/Music and Photos/Videos.
My career in media spans Radio, TV, Print, Photography, Music, Film, Online, Live, Advertising, as producer, director, writer, performer, editor, engineer, executive, owner. An exhausting but amazing ride.

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Adaptive Limiter in Parametric EQ
« Reply #6 on: January 29, 2012, 07:38:29 pm »

I'm really interested in this topic for my home setup.

For my system to be in calibration, I need my center speaker to be about -25 dB relative to the subwoofer.  This is because I use the same power amplifier for the subwoofer as for the other channels, and the center speaker is way more sensitive and subwoofers always require +10 dB.  This means 100% is -25 dB on the center, which sometimes isn't quite enough.

So I want a way to have the center be at like -15 dB at 100%, and just have the subwoofer fall out of calibration _only_ when the output level requires it.

I think this would be some type of 10 dB limiter on the subwoofer channel that does nothing most of the time, but reduces the level during really loud bass passages.

MusicHawk's suggestion of a multi-band limiter is interesting, but is it needed when we're talking about the subwoofer channel?  I mean, it's only going to contain frequencies from like 10 Hz to 100 Hz (roughly).

If this were a common enough need, could there just be a few levels of subwoofer limiting built into 'Room Correction'?  The advantage to this is that it would know when the limiter is engaged so that it could correctly calculate that the center only requires -15 dB as opposed to -25 dB (using my home setup as an example).

Thanks for any advice.

Logged
Matt Ashland, JRiver Media Center

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Adaptive Limiter in Parametric EQ
« Reply #7 on: January 29, 2012, 09:20:04 pm »

I've been thinking about this some more, and it gets complicated.

For example, I use the hardware volume on an X-Fi card.

Imagine a case where it's at 50% and a huge bass spike comes through.  It could do one of the following things:
1) Turn everything else down to keep the proper balance between the bass and other speakers
2) Limit the bass (so it falls out of calibration for the spike)
3) Do #1 but also turn the hardware volume up the corresponding amount (to say 80%).

It seems like #3 is ideal since the apparent volume stays the same and the bass is reproduced correctly.  However, this isn't possible today.

Is this an argument for running the hardware at 100% volume and always using 'Internal Volume' so we could better handle this type of thing?  I need to do some tests to see if this would even work in my case.
Logged
Matt Ashland, JRiver Media Center

MusicHawk

  • Citizen of the Universe
  • *****
  • Posts: 796
Re: Adaptive Limiter in Parametric EQ
« Reply #8 on: January 29, 2012, 09:21:42 pm »

Matt:

I haven't focused on my subwoofer's behavior in particular, because it's implicit in my MC plus multi-band compressor setup. Bass is automatically consistent, compressed/limited as needed. Each audio band has separate controls, so to control the subwoofer I simply adjust the output level of the band that handles low bass channel.

The cool thing is, it not only reduces excessive bass, it "adds" bass to recordings that are wimpy (thanks to being a compressor, not just a limiter). It's really amazing to hear solid bass in many recordings that lack bottom (I rip a lot from vinyl so encounter this frequently). And really really cool to hear CONSISTENT bass with just about every track I play, spanning many decades of recording.

The huge reason to recommend multi-band compression is the ability to achieve audio consistency. If I turn it off so the original source audio is untouched, listeners immediately ask "what happened, it sounds horrible!" I have recordings with screechy highs that need to come down, and recordings with muddy mids that need to be brightened aroud 2.5K, and they all get improved by the being handled by separate band processing -- dynamically, as they play. No two recording facilities, studio setups, mixes, or audio chains are alike, not to mention encodings, streams, etc, but listening to my setup is very smooth and satisfying. I credit the multi-band compression I'm applying.

I should admit my bias: I'm a professional broadcast engineer (among other things), so on a life-long quest to provide consistent, balanced musical enjoyment to my "audience" (whether big city radio listeners or me, my wife and friends) is bolted into my ear-brain connection.

The basic plumbing to process just bass could instead select any arbitrary audio band, then limit/compress it. Once such a "black box" exists, run several in parallel to create a multi-band compressor "device". Connect the compression controls of two such devices to create a stereo multi-band compressor.  ;D

Maybe you'd like to check out AudioProc, one of the multi-band compressor plug-ins I use (the article link I provided), to see what you can do with bass and other frequencies via this type of processing. This product is a tad pricey, but there's a similar but somewhat less-refined plug-in that is free and works pretty well with MC. I can dig up the name if you wish.
Logged
Managing my media with JRiver since Media Jukebox 8 (maybe earlier), currently use Media Center for Audio/Music and Photos/Videos.
My career in media spans Radio, TV, Print, Photography, Music, Film, Online, Live, Advertising, as producer, director, writer, performer, editor, engineer, executive, owner. An exhausting but amazing ride.

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Adaptive Limiter in Parametric EQ
« Reply #9 on: January 29, 2012, 09:28:40 pm »

I need to do some tests to see if this would even work in my case.

The noise / hum at the speakers is identical when playing silence at 0% and 100% hardware volume with an X-Fi.  So I could use internal volume to avoid the double-volume issues to make this sort of thing easier.

As an aside, I used the new Tools > Advanced Tools > Create Test Clips to create the silence to test with (and made sure 'Do not play silence' was unchecked).
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Adaptive Limiter in Parametric EQ
« Reply #10 on: January 30, 2012, 11:03:08 am »

2) Limit the bass (so it falls out of calibration for the spike)
To limit the bass seems like you are talking about reducing its volume. A limiter doesn't reduce volume. It just cuts off the peaks after the volume reaches a certain threshold. The bass is still in proper calibration with the other speakers.

Quote
Is this an argument for running the hardware at 100% volume and always using 'Internal Volume' so we could better handle this type of thing?  I need to do some tests to see if this would even work in my case.

What I am asking for only works with Internal Volume control.

Some are buying the Behringer NU6000DSP amp for use with passive subwoofers. The amp has a built-in DSP that you can control with a little software program and USB connection to the amp. If you download the remote editor software you can see the functions of the DSP. JRiver can already handle most of these DSP functions. This DSP is essentially sitting between JRiver's output (using internal volume) and the amp's input. It modifies the signal before it reaches the amp's input stage. One nice feature is that it has a peak limiter. This is what I am requesting in JRiver. You can play with the remote editor, but below is a screenshot of what the peak limiter looks like.  I've added a peak limiter to channel B.
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 42323
  • Shoes gone again!
Re: Adaptive Limiter in Parametric EQ
« Reply #11 on: January 30, 2012, 11:17:14 am »

To limit the bass seems like you are talking about reducing its volume. A limiter doesn't reduce volume. It just cuts off the peaks after the volume reaches a certain threshold. The bass is still in proper calibration with the other speakers.

This might be semantics.  If you cut a peak, you are reducing the volume since the volume is the area under the curve.

An adaptive limiter cuts the volume and then restores it at some rate.

You can also brick-wall limit with Parametric Equalizer, which just cuts the peaks.
Logged
Matt Ashland, JRiver Media Center

mojave

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 3732
  • Requires "iTunes or better" so I installed JRiver
Re: Adaptive Limiter in Parametric EQ
« Reply #12 on: March 09, 2012, 10:06:01 am »

In my living room I have a pair of dual opposed 15" subs driven by a 1000 watt amp. Last night I had a few minutes alone with them so I tried several of the new features in JRiver. I first put JRiver in the new loopback mode and ran sweeps with Room Equalization Wizard (REW). This works perfectly and is a great new feature. I did close mic measurements of the subwoofers by holding the mic close to 1 driver. The driver I am using has an inductance hump at about 55 Hz. I used a 55 Hz PEQ with Q=1 and - 5 dB to smooth things out. Then I added a Linkwitz Transform with an Fp of 35 Hz and Qp of .707. This lifted the bottom end and started the sub's rolloff to be at 35 Hz. This matches well with the room gain in my room. I unchecked "adjust volume of all channels to prevent clipping" because I was going to use the limiter.

Next I tested the limiting. I played back the Jericho scene from Iron Man since it has very high output requirements in the bass region. It is one of the few scenes I have been able to shut the amp down with do to the amps protection circuit. There are two more new features of JRiver that helped with the testing. First, the DSP's can be processed independently of internal volume. This lets one use the internal volume, but still have the signal at it maximum level for a DSP. The second feature is the new Parametric EQ 2 DSP. I probably could have added the limiter to the first PEQ DSP, but I didn't want the Linkwitz-Transform done at maximum volume all the time since it could clip. I did want to process the 2nd PEQ DSP independent of internal volume since I was going to use the limiter on it.

I added the adaptive limiter with default settings to PEQ 2, clicked "options" at the top right of the DSP, and checked "Process independently of internal volume." I next added two "Adjust the volume" filters with one before and one after the limiter. I set these to opposing values in order to force the limiter to act at a certain volume level. The first filter increases the volume and the second filter decreases the volume by the same amount. By increasing the volume on the subwoofer channel you can trigger when the limiter is kicking in and experiment with various triggers.

My first playback of the the Jericho scene was with no limiting and I increased the volume until the amp light started to flicker. The light isn't a clip light, but more an indicator of output. It starts to flicker when the amp is at half output. 3 dB more of gain takes the amp to full output and the light is solid. For the first use of the limiter I set the volume filters to +20 and -20 dB. With this setting the limiter was gently removing a lot of the peaks in the bass. I then tried +10/-10, +5/-5, and finally disabled the volume filters. I found that even with the volume filters disabled, the limiter was still in effect and the subwoofer amp light did not flicker. This is because the Linkwitz-Transform is boosting the bass enough that the signal is pushing 100% peak levels. I was able to notice a difference with each of the limiter settings. There didn't really seem to be an overall volume reduction because the "core" output is still present and just the peaks are being cut from the signal.

One thing that I didn't try is to change the volume reduction slightly to offset the loss in peaks. For example I could have used +5 dB and -2 dB for the volume filters. This would have still activated the limiter sooner, but I wouldn't have returned the bass to its original level. I would actually be playing it a little louder. You might think that this is counterproductive, but it has to do with overall driver excursion. Engaging the limiter is reducing peaks and often those are the lowest bass signals. These need a lot of driver excursion and could cause the driver to bottom out. By removing the lowest bass you can play the midbass at higher levels. The reason for using the limiter and not a high pass filter on the bass is because the high pass filter cuts everything off regardless of volume. The limiter allows one to hear all the frequencies only kicks in when you turn the volume way up. It is a means of wringing the most out of your gear without exceeding mechanical limits.

Overall I am very pleased with the latest improvements to JRiver and feel that I could completely setup a subwoofer system to be properly EQ'd and have it stay within its limits by using the DSP offered by JRiver.
Logged
Pages: [1]   Go Up