TTS Audio Tools

[BenIfin]

Sorry, didn't see your first post for some reason... it doesn't accept 24000, you need to put 23999.9999 there. And the same for the start frequency iirc (0.0001 instead of 0).

Also, I suggest setting the duration to 20s at least, to have better resolution on the spectrogram, and level to 0 (you can turn it down later on if needed)

Thanks :)

It wont do a 20 sec sweep - limited to 10 sec sweep - is that still OK for testing ? or should I look elsewhere for a sine sweep generator ?

 

[DLC86]

Thanks :)

It wont do a 20 sec sweep - limited to 10 sec sweep - is that still OK for testing ? or should I look elsewhere for a sine sweep generator ?

Yeah, there are some limitations if you don't leave him a donation... Anyway, the best alternative is Room Eq Wizard (and not only to generate sine sweeps)

 

[DLC86]

TTS Spectrogram Generator v2.2​

Changelog:

v2.2

• You can now load audio files of any length as the script now uses web workers for the processing, this avoids browser timeouts and allowed to remove file size limits.
• When selecting a narrower time range, only the selected range is processed to create the graph, this leads to shorter processing times in those cases.
• Added a progress bar to give a visual feedback about the processing advancement.
• Added cancel button to stop the processing without the need to refresh the page.
• Added support for stereo and multi-channel files, the graph represents the mono sum of all channels in this case.
• Various other fixes and improvements.

v2.1

• Improved mobile visualization in landscape orientation
• Added File Extension to the filename parameters, if selected this adds the extension of the loaded audio file to the PNG filename.

Online version: https://shop.thetonescientist.com/pages/audio-tools
Html file download: https://drive.google.com/drive/folders/1HJyhP9YWJMq7t7coCwXSCqMe83FpGHKm?usp=drive_link

 

[BenIfin]

OK ... used REW to make a 20 sec 0 - 24k Linear Sweep.

Used the same Amalgam Capture for each - a Carr Mercury V on the gain side with the Cpature Gain set on 10 - super super distorted for Pic 1 and Pic 2.

Pic 3 is a Slammin Mofo Complex JCM with Gain on 10 - it is one of his hyper accurate NAM Captures


NOTE:- to get the .WAV files I placed Tonex and then NAM in my Audio Track EFX - inserted the Sine Sweep into the Track and then printed the resulting .wav file

Do these pics make sense / look like they should ?

Pic 1 is Tonex

Pic 2 is NAM [Standard]

PIC 3 - NAM COMPLEX

 

[DLC86]

Do these pics make sense / look like they should ?

Yes they're fine, Tonex has that kind of "doubling" in the graph due to their bad SRC, if you switch the project to 44.1 kHz (which is Tonex native sample rate) and use a 0-22.05 kHz sine sweep, that disappears.

 

[BenIfin]

Looking at the NAM Standard vs NAM Complex, does the Spectrograph reveal much ? - I dont know

Also, am I right in seeing that Tonex applies a pretty harsh Low Cut above 15K ?

 

[DLC86]

Looking at the NAM Standard vs NAM Complex, does the Spectrograph reveal much ? - I dont know

Share them!

Also, am I right in seeing that Tonex applies a pretty harsh Low Cut above 15K ?

Yup

 

[DLC86]

Looking at the NAM Standard vs NAM Complex, does the Spectrograph reveal much ? - I dont know

Oops... I realized now that you had already added the "complex" graph in your previous post, might have been right after my reply...

Anyway, what does complex represent exactly? Nam standard architecture with one of the super inputs?
Cuz in this case it looks a little worse than standard actually... But that might be due to different level of the files (dry/wet mode on tonezone normalizes the files and so resulting models don't match the original volume of the reamp and of the standard model), so be sure "Normalize" is checked in the spectrogram generator.

Also, if the reamp already contains aliasing, super inputs don't work and in some cases also make aliasing worse. This can happen with real amps too cuz aliasing could be introduced by distortion in the audio interface's input or output circuitry.
Don't push it too much, always leave at least 6 dB of headroom even on the output, by turning down the input file level if necessary, cuz that peaks at 0 dBFS.

You can also use the spectrogram generator to check if the audio interface introduces aliasing, by loading a recording of the reamped sine sweep.

EDIT: here are some pretty self explanatory examples, all reamps of the real analog pedals.

First 3 different output levels:




And then input without and with clipping

 

[BenIfin]

OK - 4 pics here

=> 1 x RME Interface only
=> 1 x NAM Standard - NAM Gain on 10 & NAM B/M/T all on 5
=> 1 x NAM Hyper- NAM Gain on 10 & NAM B/M/T all on 5
=> 1 x Helix Native JCM 2203 with Gain, Master Volume and B/M/T/P all on 10 <- for comparisons sake

1 x Sine Sweep through my 13 Year Old RME UCX by itself - with no NAM Block

2 x Slammin Mofo Standard NAM of Cranked Plexi with Gain in NAM on full B/M/T all on 5

3 x The same Slammin Mofo Hyper NAM of Cranked Plexi with Gain in NAM on full on full B/M/T all on 5

4 x Helix Native JCM 2203 Gain on Full Master Volume on 10 and B/M/T/P all on 10 also

Is my RME UCX showing its age ? -or- is this pretty normal even for a current day interface ?

The Hyper NAM "looks" a little less dense overall but more noticeably less dense as the sweep moves into the high range over time ?

I'm assuming (?) that these Spectrograms show that NAM Standard and NAM Hyper have a very long way to go (?) to catch up with - in this case - Aliasing in a stupidly cranked Amp in a top tier physical modeler (?)

And some people still "bang-on" about the Helix for being so old (?)

 

[DLC86]

OK - 4 pics here

=> 1 x RME Interface only
=> 1 x NAM Standard - NAM Gain on 10 & NAM B/M/T all on 5
=> 1 x NAM Hyper- NAM Gain on 10 & NAM B/M/T all on 5
=> 1 x Helix Native JCM 2203 with Gain, Master Volume and B/M/T/P all on 10 <- for comparisons sake

1 x Sine Sweep through my 13 Year Old RME UCX by itself - with no NAM Block
View attachment 39310


2 x Slammin Mofo Standard NAM of Cranked Plexi with Gain in NAM on full B/M/T all on 5
View attachment 39311


3 x The same Slammin Mofo Hyper NAM of Cranked Plexi with Gain in NAM on full on full B/M/T all on 5
View attachment 39312

4 x Helix Native JCM 2203 Gain on Full Master Volume on 10 and B/M/T/P all on 10 also
View attachment 39315


Is my RME UCX showing its age ? -or- is this pretty normal even for a current day interface ?

The Hyper NAM "looks" a little less dense overall but more noticeably less dense as the sweep moves into the high range over time ?

I'm assuming (?) that these Spectrograms show that NAM Standard and NAM Hyper have a very long way to go (?) to catch up with - in this case - Aliasing in a stupidly cranked Amp in a top tier physical modeler (?)

And some people still "bang-on" about the Helix for being so old (?)

Your interface shows a lot of distrotion and aliasing and that shouldn't happen, as I said check your levels, that happens if clipping or distortion occurs in the audio interface.

Interface alone with no amp should look like just a single line going from bottom-left corner to top-right corner (and a couple very faint harmonics at max)

Interface plus real amp should look like this instead:

PS: hyper accuracy captures don't use the anti-aliasing inputs, so the aliasing is pretty much the same as standard captures, i.e. the equivalent of having no oversampling.

 

[BenIfin]

Yep - had a heap of distotion as I never re-checked my levels Re-did them

1 = RME Alone
2 = Helix Native 2203 all Controls on 10
3 = Amalgam Carr Mercury Hi Gain Plexi Style with all Controls on 10

Please correct me if Im wrong but Helix Native seems (?) to have way lower aliasing than NAM Standard (?)

1 - RME

2 - Helix 2202

3 - Amalgam Carr Plexi NAM Standard

 

[DLC86]

Helix Native seems (?) to have way lower aliasing than NAM Standard (?)

Obviously, helix has oversampling, NAM standard doesn't.

 

[DLC86]

From the geometry of the aliasing we can also tell the oversampling rate of a given plugin/modeler.

Here are some examples:

48 kHz (no oversampling)



2x


4x


8x


16x


32x



PS: @BenIfin your helix native graph doesn't seem to match any of these geometries though, I suspect it might be running at 44.1 kHz natively and oversample from there... I'll make some other graphs at 44.1 kHz

PPS: nope, 44.1 kHz graphs don't match either... maybe they're using different oversampling for different parts of the circuit. Anyway, looks pretty close to 4x

 

[DLC86]

I made another tool that I hope can be useful...

TTS Null Tester v1.0​

Load two audio files and it will automatically align them using cross-correlation (if the misalignment is within 1000 samples) and set the level of the second file to find the maximum cancellation.
It will spit out a bunch of data like absolute LUFS and dB rms of the subtraction, as well as those relative to the two loaded files, plus other more technical data like ESR, MSE, MAE and MRSTFT.
You can also download a wav file of what's left in the audio after the subtraction.



Download the html file here: https://drive.google.com/drive/folders/12aFRirV0jf6rWGBPYNJlCNpSoMLfMIyz?usp=drive_link
Or use it online here: https://shop.thetonescientist.com/en-eur/pages/null-tester

Known issues: for longer files the alignment takes quite a bit of time and needs optimization. Also, the progress bar and the log is not updated in real-time when the alignment algorithm is running. If it takes too long just be patient.

 

[BenIfin]

Hey Paisano !

I haven't done a NAM Capture before, but seeing as I have now set up 3 or 4 Helix Amps exactly as I want them, I am going to do some with ToneeZone3000.

A quick question.

The TZ3000 site has a link to its preferred NAM Standard Training file which I will use - but I am keen to download and also try training file that reduces aliasing further to compare.

I cant seem to find this "aliasing reduced" anywhere ?!? ... best I could find was a mention of something like "super input + new X Standard NAM" if this means anything (?)

Any chance you could please point me in the right direction to find and download this "new" / "improved" training file ?

Many thanks :)

 

[DLC86]

Hey Paisano !

I haven't done a NAM Capture before, but seeing as I have now set up 3 or 4 Helix Amps exactly as I want them, I am going to do some with ToneeZone3000.

A quick question.

The TZ3000 site has a link to its preferred NAM Standard Training file which I will use - but I am keen to download and also try training file that reduces aliasing further to compare.

I cant seem to find this "aliasing reduced" anywhere ?!? ... best I could find was a mention of something like "super input + new X Standard NAM" if this means anything (?)

Any chance you could please point me in the right direction to find and download this "new" / "improved" training file ?

Many thanks :)

They're available somewhere in the nam group, but for convenience I just uploaded them here: https://drive.google.com/drive/folders/13Df-GlKYTkpkx83Rwp-6QsdiQcJFGDxG

I suggest using TTSv10 or TTSv10 long

 

[BenIfin]

They're available somewhere in the nam group, but for convenience I just uploaded them here: https://drive.google.com/drive/folders/13Df-GlKYTkpkx83Rwp-6QsdiQcJFGDxG

I suggest using TTSv10 or TTSv10 long

Huge thanks ! :)

Once I've loaded the sweep and run it through my [test modeler] Amp, do I need to load both this new "long sweep" and the "recorded track" as some sort of wet/dry pair (?) -or- just the recorded track ?

Thanks again :)

 

[DLC86]

Huge thanks ! :)

Once I've loaded the sweep and run it through my [test modeler] Amp, do I need to load both this new "long sweep" and the "recorded track" as some sort of wet/dry pair (?) -or- just the recorded track ?

Thanks again :)

Yep, both as the dry/wet pair.

Mind you that this mode in tonezone throws off the levels calibration if you use it

 

[BenIfin]

Yep, both as the dry/wet pair.

Mind you that this mode in tonezone throws off the levels calibration if you use it

^ Does this mean the resulting NAM file is "incorrect" -or- just that it comes out either louder or quieter that it should (?)

 

[DLC86]

^ Does this mean the resulting NAM file is "incorrect" -or- just that it comes out either louder or quieter that it should (?)

It comes out louder. I'm not sure about the input gain (aka "send level") but it seems to be right in my models