Pre-ringing, minimal phase filters (and others), and apodizing...

[TheOctavist]

Thanks- the link still works for me. TBH I just used used Phil Leigh's custom convert.conf settings. I appreciate the relative complexity of the sampling rate conversion, but since its being done on my multi core PC I am not sure it really matters in the way it might do for a dsp chip on a dac. It's worth pointing out that Lots of dac automatically upsample to 96 or 192 anyway, and if so I think it's better done by SOX on balance.

can someone walk me through how to do this with the SBT?

I have no idea. a step by step as the touchtoolbox guy shows would be helpful..

[VirusKiller]

Just stumbled on this thread

I'm currently using -a (allow aliasing), 85.0 to get the most shallow drop-off SoX will allow, and a gain of -2.75 to avoid clipping in most situations. The replaygain keyword isn't necessary.

I have started writing my own up-sampling utility, but it's a very long term project.

One thing to be aware of is that, IIRC, the version of SoX that ships with LMS has a bug in it which prevents 24-bit FLAC output (it erroneously writes out 16 bits). This is fixed in later builds of SoX which I have installed manually on my server.

[Archimago]

Just stumbled on this thread

I'm currently using -a (allow aliasing), 85.0 to get the most shallow drop-off SoX will allow, and a gain of -2.75 to avoid clipping in most situations. The replaygain keyword isn't necessary.

I have started writing my own up-sampling utility, but it's a very long term project.

One thing to be aware of is that, IIRC, the version of SoX that ships with LMS has a bug in it which prevents 24-bit FLAC output (it erroneously writes out 16 bits). This is fixed in later builds of SoX which I have installed manually on my server.

Hi VK!

Nice seeing you around. Thanks for your work. I have been using your suggested settings a while now.

Intellectually and empirically we know what's being done with the upsampling and pre-ring suppression... But do you think this is perceptible? I'd be curious to see if anyone has ever been able to ABX the difference (should be quite easily doable!)...

[Wombat]

Hi VK!

Nice seeing you around. Thanks for your work. I have been using your suggested settings a while now.

Intellectually and empirically we know what's being done with the upsampling and pre-ring suppression... But do you think this is perceptible? I'd be curious to see if anyone has ever been able to ABX the difference (should be quite easily doable!)...

The legendary pre-ringing is something that should be a non-issue with DACs that work correctly.
Once there were some false implemented hardware that caused intermodulation of the pre-ringing into the audible band. This is no problem anymore.
Unfortunately it is easy to show as fancy picture that even can be "improved" with magic filtering.
This is audiophile tuning at its best. A graph that shows anomalities that dispaear when filtered. This must be better...
Also remember that this ringing normaly happens above 21kHz and there are several reasons it shouldn´t be a problem. Even think about that many tweeters don´t produce much content there.
In reality nearly all non-linear filters will cause lots of time smearing even starting at frequencies around 10kHz that surely change the audible signal but look soo good for the pre-ringing picture.

I seldom take marketing writings but this is what Benchmark thinks about it and i think they are right:
http://www.benchmarkmedia.com/discus...dizing-filters

I doubt you can produce some signal that causes your Transporter to hickup.

It should be possible to produce a file with a huge amount of pre-ringing with a "maximum phase" filter to do some abx. If you like i may have a looky how to produce such a file.

[Archimago]

The legendary pre-ringing is something that should be a non-issue with DACs that work correctly.
Once there were some false implemented hardware that caused intermodulation of the pre-ringing into the audible band. This is no problem anymore.
Unfortunately it is easy to show as fancy picture that even can be "improved" with magic filtering.
This is audiophile tuning at its best. A graph that shows anomalities that dispaear when filtered. This must be better...
Also remember that this ringing normaly happens above 21kHz and there are several reasons it shouldn´t be a problem. Even think about that many tweeters don´t produce much content there.
In reality nearly all non-linear filters will cause lots of time smearing even starting at frequencies around 10kHz that surely change the audible signal but look soo good for the pre-ringing picture.

I seldom take marketing writings but this is what Benchmark thinks about it and i think they are right:
http://www.benchmarkmedia.com/discus...dizing-filters

I doubt you can produce some signal that causes your Transporter to hickup.

It should be possible to produce a file with a huge amount of pre-ringing with a "maximum phase" filter to do some abx. If you like i may have a looky how to produce such a file.

Thanks Wombat. Benchmark gives a nice explanation. I'm aware of the phase changes but didn't know that minimum phase filters could cause foldback distortion they talk of...

[Mnyb]

The legendary pre-ringing is something that should be a non-issue with DACs that work correctly.
Once there were some false implemented hardware that caused intermodulation of the pre-ringing into the audible band. This is no problem anymore.
Unfortunately it is easy to show as fancy picture that even can be "improved" with magic filtering.
This is audiophile tuning at its best. A graph that shows anomalities that dispaear when filtered. This must be better...
Also remember that this ringing normaly happens above 21kHz and there are several reasons it shouldn´t be a problem. Even think about that many tweeters don´t produce much content there.
In reality nearly all non-linear filters will cause lots of time smearing even starting at frequencies around 10kHz that surely change the audible signal but look soo good for the pre-ringing picture.

I seldom take marketing writings but this is what Benchmark thinks about it and i think they are right:
http://www.benchmarkmedia.com/discus...dizing-filters

I doubt you can produce some signal that causes your Transporter to hickup.

It should be possible to produce a file with a huge amount of pre-ringing with a "maximum phase" filter to do some abx. If you like i may have a looky how to produce such a file.

i'm not that good at the exact technique used in this . But Is it so that some of these filters actually have a lot off treble drop off before 20kHz ?
Image foold back is not nice , is that the same as aliasing ? or do you only call it aliasing when doing analog to digital ?
I see this kind of problems a lot in non audio applications in my work.
So the benchmark guys think you actually are hearing the apodizing filter and not the removal of preringing ?
I've read that before somewhere so I don't think they are the only ones with this pow.

[Wombat]

Heh! Thats why it is still marketing writing by benchmark. They must mean simply "Aliasing" with this "image fold back"
What is for sure that this so called Apodizing/minimal phase filters often are used together with filters that don´t filter strong enough at 22kHz so these do alias for sure.
If the filter is steep enough some aliasing can´t hurt, it will only happen at very low level. An apodizing filter that does work for sure is the Meridian one, no aliasing and no pre-ringing but frequencies above 18.5kHz change in phase and amplitude. Of course that is what i doubt really changes audibility with modern DACs. It is just that Meridian has a nice patent here to sell and attract.

Also always think about what you do resample. Many 44.1kHz material already is filtered steep at below 21kHz. Using a strong ringing filter that does this at 21.5kHz for example doesn´t hurt at all because there is no content that can ring

One funny thing i noticed together with this is that even Pyramix seems to have a strange vision about Apodizing. Look at the graph here for "Pyramix 7 (Apodizing Filter)" http://src.infinitewave.ca/
It strangely has a linear phase response and aliasing. Huh? This must be a simple standard filter that allows aliasing, no sign of apodizing at all. I wonder if Pyramix has more strange settings for their filters and thats why some "experts" hear Pyramix filter sounding better with strange non-linear settings.