Print

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

[JimH]

ran a quick test using a single filter (LS 80Hz +2dB Q=0.707) and then repeating that n times, was fine til I tried it with more than 6 biquads which resulted in a burst of 0dBFS high frequency content. Tweeter and ear melting stuff basically hence I think feature is currently unsafe to use til that bug fixed (I run these tests silently as I've nearly melted my ears more than once in my life doing this sort of thing!)

Darn!  But thanks for testing.

So would it be OK, in your opinion, to set a limit of 6 for now?  We'd fix the problem you found, but not until a later build.

[Matt]

Jim, I'm working on fixing it this morning.  Today's build will be good (hopefully!).

Thanks for testing!

[mattkhan]

Darn!  But thanks for testing.

So would it be OK, in your opinion, to set a limit of 6 for now?  We'd fix the problem you found, but not until a later build.

I think it depends what the problem is, I only tested a v simple case which tripped this problem but it doesn't mean it can't be tripped in some other way (using fewer biquads)

[Matt]

Testing appreciated.

Thanks!

[JimH]

Testing appreciated:
https://yabb.jriver.com/interact/index.php/topic,132343.0.html

Thanks!

We're releasing it tomorrow.

[mattkhan]

the problem (with biquads 7 & 8 ) is fixed in the latest build

[Matt]

Don't ask if I paid bribes!

[JimH]

the problem (with biquads 7 & 8 ) is fixed in the latest build

Whew!  Thanks for reporting.

[mwillems]

So I was paddling around this morning, and it's not a huge deal, but I notice that the highpass and lowpass filter Q in DSP Studio seems to be rounding to two decimals.  So for example, if I enter .7071 which is the correct Q for a 2nd order Butterworth, it rounds it off to .71 which isn't quite right.  Any chance of more precision there?

Also it might be worth considering setting the default to something other than "1" as that's not an expected filter slope at any order so newly created filters will have unexpected behavior for naive users.  To get back the old default behavior (where all filters are butterworth by default) it probably would make sense to set the default to whatever the appropriate Q is for a butterworth filter at that order (e.g. .7071 for 2nd order butterworth, 4th order butterworth has 1.307, etc.).

I put this here to continue the ongoing conversation, but let me know if I should post this in the build threads instead.  Thanks again for all the cool improvements this cycle!

[JimH]

I just moved this out from the beta board so you can see how much mwillems and mattkahn contributed to the new DSP capabilities.  Thanks again! 

And thanks to Matt (JRiver) for all the blood sweat and tears he put into making their dreams come true!

[Matt]

Coming next build:
Changed: Parameters in the Parametric Equalizer dialog show up to five digits of precision instead of two.

[BryanC]

So I was paddling around this morning, and it's not a huge deal, but I notice that the highpass and lowpass filter Q in DSP Studio seems to be rounding to two decimals.  So for example, if I enter .7071 which is the correct Q for a 2nd order Butterworth, it rounds it off to .71 which isn't quite right.  Any chance of more precision there?

Also it might be worth considering setting the default to something other than "1" as that's not an expected filter slope at any order so newly created filters will have unexpected behavior for naive users.  To get back the old default behavior (where all filters are butterworth by default) it probably would make sense to set the default to whatever the appropriate Q is for a butterworth filter at that order (e.g. .7071 for 2nd order butterworth, 4th order butterworth has 1.307, etc.).

I put this here to continue the ongoing conversation, but let me know if I should post this in the build threads instead.  Thanks again for all the cool improvements this cycle!

Aside, I was under the impression that Q is actually S in MC? https://wiki.jriver.com/index.php/Parametric_Equalizer#Q_or_S

[mwillems]

Aside, I was under the impression that Q is actually S in MC? https://wiki.jriver.com/index.php/Parametric_Equalizer#Q_or_S

So that's actually something that should have been fixed as part of this effort per Matt upthread.  Feel free to test!

Coming next build:
Changed: Parameters in the Parametric Equalizer dialog show up to five digits of precision instead of two.

Thanks Matt!

[BryanC]

So that's actually something that should have been fixed as part of this effort per Matt upthread.  Feel free to test!

Oh wow, I missed that!

[Matt]

Also it might be worth considering setting the default to something other than "1" as that's not an expected filter slope at any order so newly created filters will have unexpected behavior for naive users.  To get back the old default behavior (where all filters are butterworth by default) it probably would make sense to set the default to whatever the appropriate Q is for a butterworth filter at that order (e.g. .7071 for 2nd order butterworth, 4th order butterworth has 1.307, etc.).

Could you expound on this a bit?  A new Low-pass is set to 12 dB/octave and 1 for Q.  Are you saying it should be 12 dB/octave and .7071?

You can change the slope, but that doesn't change the Q.

Thanks.

[BryanC]

Now that I don't have to convert Q->S I'm more likely to play around with custom parametric EQs. Could it be possible to add more than two and to rename them in the sidebar? I know this is somewhat possible with preset files, but it's way nicer just selecting/de-selecting them for AB.

[mwillems]

Could you expound on this a bit?  A new Low-pass is set to 12 dB/octave and 1 for Q.  Are you saying it should be 12 dB/octave and .7071?

You can change the slope, but that doesn't change the Q.

Thanks.

So a first order butterworth (6db/octave) has a Q of .5, a 2nd order butterworth (12db/octave) has a Q of .7071.  But higher order filters will have multiple poles and not all filters that have the same product Q have the same shape.  I think the two poles of the 4th order butterworth are a Q of .541 and Q of 1.31, which make the product Q .7071.  But if you change the poles you can wind up with the same product Q and a different slope instead. 

To see more info, mattkhan linked a code example above from his beqdesigner program that shows how he modelled/calculated the Q factor for various pass filters: https://github.com/3ll3d00d/beqdesigner/blob/f2991ac3b78669ec237826ef18ca4f2afa885648/src/main/python/model/iir.py#L1208

This complexity was part of why I think mattkhan suggested offering presets for different filter shapes and orders.  As it stands getting things right for higher order filters requires cascading several and setting the Q's just so.

[Matt]

So what should we change the default to?

[mattkhan]

Aside, I was under the impression that Q is actually S in MC? https://wiki.jriver.com/index.php/Parametric_Equalizer#Q_or_S

Note that that only ever applied to shelf filters

[mwillems]

So what should we change the default to?

I'd be interested to hear mattkhan's thoughts on this, but if we have to pick only one default (rather than defaults that shift by order), I think .7071 is the sanest default (i.e. least likely to result in surprising behavior). 

[Matt]

Next build:
Changed: Low and high pass filters have a default Q of 0.7071.

[mattkhan]

I agree the default should be a 2nd order Butterworth (so q=1/sqrt2)

I think a custom q pass filter is a pretty niche thing though, people generally just want the usual named filters of various orders.

[d_pert]

We can set DSP for individual DLNA servers in MC. Why not also JRiver's 'own' (proprietary) server(s)?

I'd really like to add headphone-related DSP to the stream from my MC server to JRemote on my iPhone.

Thanks!

See also:
https://yabb.jriver.com/interact/index.php/topic,123093.msg852498.html#msg852498