Jump to content
IGNORED

Izotope SRC


Recommended Posts

Will you also make that available to your iZotope SRC OEM clients--specifically Damien Plisson of Audirvana?

I don't personally handle licensing, but it shouldn't be hard if he contacts iZotope.

 

 

Also, can you tell me if your iZotope SRC (with full controls) could be made available for license to a hardware OEM? In other words, if someone was developing a player/DAC that had processor running a Linux variant, could they license your s/w for "embedded" use in that product?

Yes, it should be possible, provided that there's enough DSP power. Again, iZotope would be the right point of contact for such collaboration.

Link to comment

Someone named TJ in iZotope customer support, in response to my email query, said that all iZotope SRC licensees get the same algorithm that is in RX-2 Advanced. There are no dumbed-down versions of iZotope SRC. (That was in Sept. 2013, before RX-3 was released.)

HQPlayer (on 3.8 GHz 8-core i7 iMac 2020) > NAA (on 2012 Mac Mini i7) > RME ADI-2 v2 > Benchmark AHB-2 > Thiel 3.7

Link to comment

Thank you Alexey and Bob.

 

Bob, I really like those graphs you posted. Would you mind running a set for me at the A+ iZotope settings I settled on a while back after a LOT of listening? (They could probably stand some tiny tweaks since the best sounding A+ version--1.5.10--came out and I also have improved my power supply, etc.)

 

Parameters are:

 

Steepness: 7

Cut-off: 1.02

Pre-ring: 0.86

My filter max-length is at 1,300,000 and anti-aliasing (final attenuation) is at 200--in case those make any difference to your method.

 

Also, what are the input and output sampling rates you are using for the impulse file? Is there a frequency to the pulse (its not a square wave, but it also is not a DC pulse, so I assume it must have a frequency.)

 

Best,

Alex C.

Link to comment

@Alex: Sorry, but I generated those graphs a year ago, and I failed to document how I did it. It would take too long for me to figure it out now.

 

The sampling rate was 44.1. The pulse was a single sample of amplitude -12dB that I downloaded from:

Professional Online Audio Frequency Signal Generator

HQPlayer (on 3.8 GHz 8-core i7 iMac 2020) > NAA (on 2012 Mac Mini i7) > RME ADI-2 v2 > Benchmark AHB-2 > Thiel 3.7

Link to comment

Can't speak for Bob. Superdad's DAC is NOS, so he oversamples to avoid harmonic distortion. As for me, yes, I oversample to "8x" rates (352.8/384) to bypass my DAC chip's interpolation filters.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
@Alex: Sorry, but I generated those graphs a year ago, and I failed to document how I did it. It would take too long for me to figure it out now.

 

The sampling rate was 44.1. The pulse was a single sample of amplitude -12dB that I downloaded from:

Professional Online Audio Frequency Signal Generator

 

Thanks Bob. Yes my goldfish memory forgot how to do those too (I once used a trial of RX). I just really liked how your graphs showed and expanded scale and the elapsed time on the impulse. Do you at least remember what s/w you used?

 

Best,

ALEX

Link to comment
Do you guy listen to 44.1 kHz material with a DAC running at 44.1 (using its own low pass filter), or do you upsample the file and use the DAC at a higher rate ?

 

I think it's really dependent on the music you're listening to.

If you listen to "ethereal" music such as ambient, smooth jazz or so, upsampling gives a lot of "air" to the music...

but for rock music with strong rhythm foundation i find that keeping native sample rates helps focus and transients.

Link to comment
I think it's really dependent on the music you're listening to.

If you listen to "ethereal" music such as ambient, smooth jazz or so, upsampling gives a lot of "air" to the music...

but for rock music with strong rhythm foundation i find that keeping native sample rates helps focus and transients.

 

Leonardo: What DAC9s) are you listening through? You do realize that unless you are using a DAC with a old R2R ladder DAC chip with no oversampling circuitry in front of it (an NOS DAC), your DAC is up/oversampling all your 16/44.1 material--to at least 384Khz--to feed its sigma-delta DAC chip. Therefore you are already listening to an SRC filter. So the choice becomes doing it in s/w with personal control over the filter parameters, or letting your DAC do it for you. Only in very good DACs, where the designers tuned their own filters--and maybe use a lot of FPGA horsepower to do it--are you listening to something other then the highly resourced-constrained built-in filters of the DAC chip.

 

And if you do use an NOS DAC like me (PCM1704K with nothing in front of it except a bit justifying CPLD and XMOS-based async-USB>I2S board), then you MUST software upsample Redbook CD unless the heavy aliasing artifacts don't bother you.

Link to comment
Leonardo: What DAC9s) are you listening through? You do realize that unless you are using a DAC with a old R2R ladder DAC chip with no oversampling circuitry in front of it (an NOS DAC), your DAC is up/oversampling all your 16/44.1 material--to at least 384Khz--to feed its sigma-delta DAC chip. Therefore you are already listening to an SRC filter. So the choice becomes doing it in s/w with personal control over the filter parameters, or letting your DAC do it for you. Only in very good DACs, where the designers tuned their own filters--and maybe use a lot of FPGA horsepower to do it--are you listening to something other then the highly resourced-constrained built-in filters of the DAC chip.

 

I have a Mytek Stereo 192 DSD. So yes, sigma-delta modulator.

I'm not that technically aware of what's going on inside the DAC to properly answer, so thank you for your insights.

I just give you my listening impressions. I disable the upsampling functionality inside the DAC via its onboard menu because my ears tell me that software (SoX or Izotope) upsampling sounds better.

And yet, I find that keeping redbook at redbook sample rates helps making it more "solid", even though a little bit harsher.

Link to comment
I have a Mytek Stereo 192 DSD. So yes, sigma-delta modulator.

I'm not that technically aware of what's going on inside the DAC to properly answer, so thank you for your insights.

I just give you my listening impressions. I disable the upsampling functionality inside the DAC via its onboard menu because my ears tell me that software (SoX or Izotope) upsampling sounds better.

And yet, I find that keeping redbook at redbook sample rates helps making it more "solid", even though a little bit harsher.

 

Ah, the Mytek and its ESS Sabre DAC chip. Numerous mysteries and possibilities with regards to how it passes and/or processes Redbook. Maybe someone else who has one will chime in here and tell us what options it offers with regards to handling 16/44. Even with upsampling supposed disabled, as an SD DAC it has to be taking it to some other rate, but I am not an ESS maven--and they are coy about it all--so I don't know what filters you end up listening to.

 

I do hear nice things about the Mytek of course.

Link to comment
No suggestions for LH labs Geek Out and izotope SRC settings?

 

I've since switched to JRiver and upsample everything to 2xDSD (which sounds wonderful, by the way), but when I was using PCM upsampling, I didn't hear any reason to change the settings I've used for some time with other DACs:

 

Screenshot 2014-06-22 12.18.18.png

 

PS - My understanding is the way I've got it set up, Audirvana uses these settings when it is forced to resample (i.e., when the DAC does not support the native data Audirvana is trying to feed it) - if you want it to always upsample, you'll need to change "Forced upsampling" section to something other than "None" ;)

John Walker - IT Executive

Headphone - SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable Ethernet > mRendu Roon endpoint > Topping D90 > Topping A90d > Dan Clark Expanse / HiFiMan H6SE v2 / HiFiman Arya Stealth

Home Theater / Music -SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable HDMI > Denon X3700h > Anthem Amp for front channels > Revel F208-based 5.2.4 Atmos speaker system

Link to comment
Thank you Alexey and Bob.

 

Bob, I really like those graphs you posted. Would you mind running a set for me at the A+ iZotope settings I settled on a while back after a LOT of listening? (They could probably stand some tiny tweaks since the best sounding A+ version--1.5.10--came out and I also have improved my power supply, etc.)

 

Parameters are:

 

Steepness: 7

Cut-off: 1.02

Pre-ring: 0.86

My filter max-length is at 1,300,000 and anti-aliasing (final attenuation) is at 200--in case those make any difference to your method.

 

Also, what are the input and output sampling rates you are using for the impulse file? Is there a frequency to the pulse (its not a square wave, but it also is not a DC pulse, so I assume it must have a frequency.)

 

Best,

Alex C.

 

Hi Alex. I plugged a few settings into the demo of RX3 last night and had a look. Long story short, looking at the graph of the impulse response, I wasn't happy with how anything over 6 looked. (That was true whether I had pre-ring at .8 or 1.0. .8 got rid of some pre-ring, but added that energy at the back end.) Based on the looks of the frequency and impulse response graphs, I tried out a steepness setting of 6 and cutoffs of 1.01 and 1.02. Just goes to show you listen with your ears rather than your eyes, I suppose: I much preferred the settings I'm currently running, steepness 5, cutoff 1.00.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

Thanks Jud. Slightly shallower slope and lower cutoff were the direction of the parameters I have been planning to revisit anyway. And of course shallower slope results in less ringing energy overall--but more aliasing artifacts passed through and folded back down--yet better time domain response due to lower group delay (this last I learned from you!). So that could be very system dependent: bandwidth of everything from the DACs analog low pass filter, to the power amps, to the tweeters.

Actually, Mr. Lukin (iZotope author for those just joining) tells us that "Steepness" and dB are misnomers for that setting--it really is a number that translates indirectly to filter "order"--which of course does control slope steepness. But I also wonder if our sonic preference (particularly we who prefer settings at the far low end of the control versus say 22 "dB") isn't also due to liking the sound of low-order filters. It's probably more the preference for less overall ringing energy, as DF order is probably not the same as analog filter orders anyhow, but still I wonder.

 

The above is why I am looking forward to having finer control over iZotope's "Steepness" parameter. I have wanted to have half-steps or finer for quite some time.

 

Separately Jud, what manual filter control does Peter offer in XXHigh-End? Does it offer full parameter control as with Advanced IZotope SRC, or do you just get to choose among a range of pre-mixed filter recipes? Not that there is anything wrong with the latter, in fact I would love to hear his "Arc Prediction" filter (on my Mac inside A+ please!).

And I will shortly be trying the new beta of stand-alone HQPlayer on OS X (getting a newer, more powerful mini just for that--and so I can pass my 2010 Core 2 Duo to my son whose G5 iMac just bit the dust).

Link to comment
Thanks Jud. Slightly shallower slope and lower cutoff were the direction of the parameters I have been planning to revisit anyway. And of course shallower slope results in less ringing energy overall--but more aliasing artifacts passed through and folded back down--yet better time domain response due to lower group delay (this last I learned from you!). So that could be very system dependent: bandwidth of everything from the DACs analog low pass filter, to the power amps, to the tweeters.

Actually, Mr. Lukin (iZotope author for those just joining) tells us that "Steepness" and dB are misnomers for that setting--it really is a number that translates indirectly to filter "order"--which of course does control slope steepness. But I also wonder if our sonic preference (particularly we who prefer settings at the far low end of the control versus say 22 "dB") isn't also due to liking the sound of low-order filters. It's probably more the preference for less overall ringing energy, as DF order is probably not the same as analog filter orders anyhow, but still I wonder.

 

The above is why I am looking forward to having finer control over iZotope's "Steepness" parameter. I have wanted to have half-steps or finer for quite some time.

 

Separately Jud, what manual filter control does Peter offer in XXHigh-End? Does it offer full parameter control as with Advanced IZotope SRC, or do you just get to choose among a range of pre-mixed filter recipes? Not that there is anything wrong with the latter, in fact I would love to hear his "Arc Prediction" filter (on my Mac inside A+ please!).

And I will shortly be trying the new beta of stand-alone HQPlayer on OS X (getting a newer, more powerful mini just for that--and so I can pass my 2010 Core 2 Duo to my son whose G5 iMac just bit the dust).

 

Hey Alex. J

Less group delay comes from a higher pre-ringing setting (closer to linear phase: 1.00 = linear phase = no group delay) rather than lower steepness.

Re XXHE, you just get the one filter (PeterSt and users are experimenting with a second) with no adjustments for the filter. You do get the largest range of adjustments for the operation of the program and the computer’s (Windows) OS that I’ve ever seen anywhere, including some things I’d never even thought of that turn out to have audibly significant impacts (except if connected to the new Phasure NOS1a DAC).

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
  • 2 months later...
The updated iZotope SRC in RX 4 now has more gradations for filter Steepness, fully compensated latency (for easy null-testing), and less time-domain ringing in intermediate-phase modes.

 

Dear Alexey,

That is great news! Thanks for letting us know. Will the finer gradations of filter steepness be somehow given to existing iZotope SRC licensees--specifically Audirvana Plus--for inclusion in their products?

 

And is the reduced time-domain ringing something that those of us setting all our own iZotope parameters will benefit from? In other words, is this improvement benefiting the SRC engine underlying all your filters and not just intermediate-phase presets?

If so, then again I ask if this is an update your licensees will have access to?

 

Regards,

Alex C.

Link to comment
In other words, is this improvement benefiting the SRC engine underlying all your filters and not just intermediate-phase presets?

The reduced ringing benefit only applies to intermediate-phase filters, i.e. those whose Pre-ringing is between 0 and 1. Linear-phase and minimum-phase filters only have the other two improvements.

Link to comment
The reduced ringing benefit only applies to intermediate-phase filters, i.e. those whose Pre-ringing is between 0 and 1. Linear-phase and minimum-phase filters only have the other two improvements.

 

Excellent, thank you! I enjoy an iZotope filter with a "pre-ring" setting of 0.86, so that improvement will apply to me. And I also look forward to using the finer "steepness" gradations as well. You might recall my being one of those here who wished for just that.

 

I will alert Damien at Audirvana to the availability of your updated SRC engine. Is there a cost to your s/w OEMs for this update?

 

Thanks again,

Alex C.

Link to comment

[Mind Blown – in from the wilds]

Greetings, this is my first post to this forum. I’d like to detail the iZotope SRC settings available in Sony’s Sound Forge Pro v.10 for its “Maximum Quality” simple setting; and, the ranges and variability for its parameters. Sound Forge has a ‘batch processing mode’ which I’ve been using to up sample (to date) my collection of HD files in either 24/88 or 24/96 to 176/192KHz sampling rate. One of the 3T Seagate USB drives on sale at Costco (now?) makes file size irrelevant.

Sound Forge Pro v.10 – iZotope 64-bit SRC “Maximum Quality” simple settings

Max SR: 192KHz

Steepness: 150 (range 0-200, vary by 1)

Filter Length: 500,000 (range 10K – 2M, vary by 1)

Cutoff Scaling: 1.00 (range 0.5 – 2.0, vary by 0.01)

Alias Suppression: 200.00 (range 50 -200, vary by 0.01)

Prering (%): 100 (vary by 1)

-----

To date about 20 HD albums have been upscaled. I am using a new DAC with a ‘Listen’ Minimum Phase switch which is what I’ve been using. No critical listening to date. I discovered this thread with a search trying to find out how to use the SRC’s advanced settings. The Maximum Quality – simple setting (as above) has seemed quite good in casual listening. A dedicated Atom-based Foxconn nettop (not connected to the internet) is the server I use – not recommended. Processing an album with it (24/96 to 24/192) usually takes a couple hours (time isn’t an issue).

John – Acoustic Line Source Research

p.s. I lost my first post, the “Quick Reply” mode was dropping every other letter, and then I wasn’t allowed to post after registering? This message was composed offline – hope it works this time.

 

Link to comment
I have a Mytek Stereo 192 DSD. So yes, sigma-delta modulator.

I'm not that technically aware of what's going on inside the DAC to properly answer, so thank you for your insights.

I just give you my listening impressions. I disable the upsampling functionality inside the DAC via its onboard menu because my ears tell me that software (SoX or Izotope) upsampling sounds better.

And yet, I find that keeping redbook at redbook sample rates helps making it more "solid", even though a little bit harsher.

 

FYI: The ESS Sabre has a two pass oversampling process with the standard settings. The first pass is an 8x FIR filter, symmetrical, linear phase, with two options for the roll off: fast (in other words a steep slope to suppress all alias products) and slow (a shallower slope, allowing less ringing, but more alias leakage). The second pass is another 8x with a IIR filter. These two filters in sequence result in 44.1 PCM files being brought up to DSD rates before the final DAC section. The first filter pass can be turned off (if the DAC maker enables this feature) but the second stage of oversampling is always active. If Mytek allows you to turn off the first stage filter, and you do so for playback of 44.1 files, you are going to be listening to a lot of in band noise, you might like it though, that is up to you.

 

Of course, with the shallow slopes most folks using Audirvana + oversampling seem to prefer, they are also listening to quite a bit of noise resulting from alias products folding back into the audible bandwidth. Without being able to test the analog output of the DAC with something like an AP there is no way to know for sure how much artifacts these shallow slope filters are producing.

SO/ROON/HQPe: DSD 512-Sonore opticalModuleDeluxe-Signature Rendu optical with Well Tempered Clock--DIY DSC-2 DAC with SC Pure Clock--DIY Purifi Amplifier-Focus Audio FS888 speakers-JL E 112 sub-Nordost Tyr USB, DIY EventHorizon AC cables, Iconoclast XLR & speaker cables, Synergistic Purple Fuses, Spacetime system clarifiers.  ISOAcoustics Oreas footers.                                                       

                                                                                           SONORE computer audio

Link to comment

Of course, with the shallow slopes most folks using Audirvana + oversampling seem to prefer, they are also listening to quite a bit of noise resulting from alias products folding back into the audible bandwidth.

 

Guilty! :)

 

I'll keep doing further experimentation (especially trying settings from people who actually know something about filter design) to try to obtain improvements. Equally important, I'm going to try to understand more about the difference between euphony and accuracy.

 

P.S. Hope you're doing well, haven't had the pleasure of running into you here in a while.

 

Edit: Re oversampling, I find it sounds best to me if done offline using software made by someone who knows about filtering.

 

 

Sent from my iPhone using Tapatalk

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...