Any use trying the H9000 for summing?

[joeydego]

Hear me out on this. Maybe futile but an interesting off label use would be some kind of emulated summing. I have 8 analog and 8 ADAT channels currently connected. Can’t I just output them all to the same stereo out and use an algorithm to simulate what a summing box would do? Is this laughably a bad idea?? Which algos if any would be suitable??

[Puppeteer]

I’ve thought about doing this as a kind of submix for analog to ADAT. Not sure if there is an algorithm for doing this, but it won’t be hard to build one in VSIG.  I’ll have a look at it soon and see what’s possible.

[joeydego]

Here’s a holy crap is this possible idea: these preamp cards they’re working on, how about sending analog and ADAT out of the daw, thru the cards and back into on a 2 channel summing bus?

or how about designing a card specifically for this maybe with a few matching algos? Come on, I don’t want to have to buy a 2 bus plus or something. Would rather give eventide my dough.

[Puppeteer]

Here you go.

Here’s a simple stereo summing algorithm.  Will mix 4 stereo inputs into a single stereo output.  All knobs are in dB.  0dB is full volume, no attenuation.

Choose whichever 8 inputs you want and whichever 2 outputs.

https://godlike.com.au/index.php?id=420

I haven’t tested it yet, but it should work.  I’ll upload the compiled algorithm next time I log into my studio.

Enjoy.

[Pombero]

Wow this is a really interesting idea. I currently use a Dangerous 2-Bus+ for analog summing, but hey my mind is now working on ways to use the H9000. I wonder if there is a creative way to combine the two… Even with 16 channels on the Dangerous I still run out of channels.

[Puppeteer]

You could put one on every buss, or you can cascade them.  I also did a 32 channel version, which I’ve uploaded here – https://godlike.com.au/index.php?id=420

I have no idea if you can have 32 inputs on a H9000 buss, but I can assign that many in VSIG, so give it a go.

[John Baylies]

> I have no idea if you can have 32 inputs on a H9000 buss, but I can assign that many in VSIG, so give it a go.

The H9000 currently supports a maximum of 8 ins and 8 outs per algorithm.

It seems that if an alg with more than 8 ins or outs is uploaded, it will crash the DSP, and you’ll have to softly reset your H9000.

https://cdn.eventideaudio.com/manuals/h9000/content/appendix/startup_seq.html#soft-reset

[Puppeteer]

Thanks John, I’ll hide that download

[Puppeteer]

It might be worth adding a warning to VSIG when more than 8 inputs or outputs are selected, to let people know that they will crash to H9000

[SonikStrike]

any update about this algo??

 

I think we just need some algos that can give different and nice shades of saturation

[Puppeteer]

I’ll have a look to see what I can do to add some saturation to the channels.

[joeydego]

Puppeteer wrote:

I’ll have a look to see what I can do to add some saturation to the channels.

 

Please keep us in the loop if you come up with something cool for this.

[Puppeteer]

I’ve added drive to this algorithm.

The algorithm, a chain and a couple of presets are available here

https://godlike.com.au/index.php?id=420

Enjoy.

[Puppeteer]

I’ve got 15 additional presets for this algorithm available at https://www.buymeacoffee.com/godlikeprod/e/72137

$1 per preset, free for members

[John Baylies]

Hi Puppeteer, I’m not sure if you mention this someplace or not, but keep in mind that in order for folks to import your presets, they’ll have to import the algorithm to the same “location”/ID# that the algorithm was in when you created the presets.

We’ve been meaning to make a script that’ll allow algorithm presets to be converted for any algorithm ID#, but I don’t have a timeline for that…

[Puppeteer]

Thanks mate, I’ll make a note of that.

The stereo summing w Drive 6 should be installed at location 10117

That script would be awesome.  There doesn’t seem to be a way to install an algorithm at a specific location, either, except to open the vsig file and then to upload to a specific location.

Any possibility of adding that to the import algorithm function of emote?

Unfortunately presets appear to be binary files, so no easy way for me to build a tool to reassign ID’s

[Puppeteer]

I’ve added a manual with workarounds for installing this to the correct ID (and how to use it)

https://godlike.com.au/fileadmin/godlike/patch/h9000/manuals/Stereo_Summing_Mixer_with_Gain_User_Guide.pdf

Just one more question while I’m here, is there a limit to the number of user algorithms ID’s that we can use on the H9000?

[John Baylies]

I’ll make a note about adding that to Emote’s import algorithm function.

Vsig’s “Algorithm Location” upload method only supports IDs from 10101 to 10199, but if you have an Algorithm with ID 10199 and use Vsig’s Next empty algorithm location, or use Emote > menubar > Algorithm > Import Algorithm, then the algorithm will appear in ID 10200. I don’t think I’ve ever gone past 10202, but I don’t think there’s a limit.

all .9k files are actually .gz files, so if you change the file extension, then unzip the files, you can see the raw json.

`Python
>>> import gzip
>>> import json
>>> data = gzip.open(‘/media/joe/UNTITLED/Really Large Room_976608312_A.9kp’).read()
>>> o = json.loads(data)
>>> o
{u’lv-obj.0′: -0.3000999987125397, u’rd7-obj.0′: 108.7991027832031, u’lmingl-obj.0′: 0.0, u’rd2-obj.0′: 70.59700012207031, u’alg_gain’: {u’out_gain’: {u’values’: [0.0], u’subscription’: None}, u’mute’: {u’values’: [0], u’subscription’: None}, u’bypass’: {u’values’: [0], u’subscription’: None}, u’tempo_mode’: {u’values’: [0], u’subscription’: None}, u’wet_dry_mix’: {u’values’: [100.0], u’subscription’: None}, u’in_gain’: {u’values’: [0.0], u’subscription’: None}}, u’lmaxgl-obj.0′: 200.0, u’outgain-obj.0′: 0.0, u’rd5-obj.0′: 94.29779815673828, u’glide-obj.0′: 100.0, u’intrm-obj.0′: 0.0, u’d3-obj.0′: 0.0, u’rd3-obj.0′: 85.69619750976562, u’rd6-obj.0′: 78.60150146484375, u’rmaxgl-obj.0′: 0.0, u’pdly-obj.0′: 0.0, u’dsize-obj.0′: 96.0, u’d2-obj.0′: 0.0, u’diff-obj.0′: 100.0, u’d1-obj.0′: 0.0, u’subscriptions’: {}, u’rd12-obj.0′: 123.0925979614258, u’vrbglide-obj.0′: 0.0, u’rdcy_1-obj.0′: 31, u’dv2-obj.0′: -100.0, u’rv-obj.0′: -3.0, u’dv1-obj.0′: -100.0, u’gldrate-obj.0′: 0.3499999940395355, u’metadata’: {u’name’: u’A’, u’program_crc32′: 976608312, u’program_id’: 2764032, u’version’: 0, u’is_factory’: False, u’preset_id’: 2764033, u’crc32′: -444381573, u’program_name’: u’Really Large Room’, u’id’: 2764032, u’software’: u’unknown’}, u’mspan-obj.0′: 100.0, u’wav-obj.0′: 1, u’rmingl-obj.0′: 200.0, u’rd9-obj.0′: 157.1882934570312, u’mdepth-obj.0′: 1.380200028419495, u’fb1-obj.0′: 0.0, u’lf-obj.0′: 190.0, u’rd4-obj.0′: 58.0, u’rd8-obj.0′: 114.0, u’h-obj.0′: 12500.0, u’gldspeed-obj.0′: 0.1000000014901161, u’rd1-obj.0′: 54.39770126342773, u’hf-obj.0′: 10500.0, u’d4-obj.0′: 0.0, u’rd11-obj.0′: 141.902099609375, u’mrate-obj.0′: 8.0, u’fb2-obj.0′: 0.0, u’hv-obj.0′: -0.5001000165939331, u’rdcy1-obj.0′: 431.0393981933594, u’rsize-obj.0′: 70.0, u’rd10-obj.0′: 134.1905059814453, u’drywet-obj.0′: 50.0}
>>> o[‘metadata’]
{u’name’: u’A’, u’program_crc32′: 976608312, u’program_id’: 2764032, u’version’: 0, u’is_factory’: False, u’preset_id’: 2764033, u’crc32′: -444381573, u’program_name’: u’Really Large Room’, u’id’: 2764032, u’software’: u’unknown’}
>>> o[‘metadata’][‘program_id’]
2764032`

These IDs are 32 bit integers that encode the bank and algorithm number as well as the algorithm ID.
program_id is the ID of the algorithm it belongs with, but preset_id also matters, because it is based on the program ID.

`
python
>>> (o[‘metadata’][‘program_id’] >> 16) & 0xff
42
`

That’s bank 42

`
python
>>> (o[‘metadata’][‘program_id’] >> 8) & 0xff
45
`

that’s algorithm 45 within bank 42, i.e. 4245 “Really Large Room”

`
python
>>> o[‘metadata’][‘preset_id’] & 0xff
1
`

this is preset 1 for that algorithm.

I hope that helps..

[Puppeteer]

Thanks so much for that.  I’ll see if I can find a gz parser for lua. If I can, I should be able to build a cross platform tool to reassign the id.

[Puppeteer]

Hi John,

Are the program_id or preset_id included in the crc check?

[Puppeteer]

Woohoo – halfway there.

>>> testLoad()
Preset File Exists: D:\Temp\Blistering.9kp
D:\Temp\Blistering
File Exists
Algorithm Name: Stereo Summing Mixer w Drive 6
Preset Name: Blistering
Old Bank: 101
Old Algorithm: 17
Preset Number: 0

 

[John Baylies]

nice! I think that nothing in the metadata section is included in the crc check.

[Puppeteer]

Thanks, that will make things easier. Does the preset ID need to be set, separately for each preset, or is it OK to have them all the same as the program_ID?  The preset’s I’ve exported all have this set to 0. I assume the H9000 just increments them as it imports them, if the program_ID and preset_ID point to the correct algorithm.

[joeydego]

Im going to check this out tonight. 1000 thanks to Puppeteer!! Coffee coming your way!

[Puppeteer]

joeydego wrote:

Im going to check this out tonight. 1000 thanks to Puppeteer!! Coffee coming your way!

Thanks for the support, mate.

[John Baylies]

Does the preset ID need to be set, separately for each preset, or is it OK to have them all the same as the program_ID? The preset’s I’ve exported all have this set to 0. I assume the H9000 just increments them as it imports them

I’m not sure, but that’s what I’d assume as well.

[Puppeteer]

Getting closer.  I decided to allow the tool to change multiple presets in one go.

Just need to get the execute button working and see if works on OSX.

 

[joeydego]

I have a mix in studio one where I’m summing everything down to 4 stereo channels. I’m sending those outputs to adat 1-8 which hits my Eventide. Thats summing out to adat 1-2 which is the input of the Track I’m recording the final mix on. This works as anticipated, except when I bypass the Eventide, the expected result is to hear all 8 channels, just unaffected. I’m actually only hearing my drums (which is adat 1/2 out of my Apollo). Am I not understanding my routing correctly?

[Puppeteer]

I think the bypass at the top just connects the inputs of the block directly to the output of the block, so you would need 8 outputs connected on the output side, which my algorithm doesn’t currently support. When you press bypass

I can probably program in a bypass button that just routes the inputs straight to the outputs, and then you can connect all 8 outputs from the algorithm to ADAT 1/2 on the output matrix of the H9000.  Would that be useful? or I could have the bypass internally mix them to the stereo output at unity gain, though there may be the risk of clipping in this case.

In the meantime you could work around it by also connecting ADAT Ins 3-8 To ADAT Outs 3-8 (in addition to your existing connection), maybe through a block with Octal I/O where you turn the gain down to 0.  If you bypass both blocks, you would get the bypass you are chasing.

I think this will happen with any block in the H9000 that has more inputs than outputs.

[joeydego]

This makes sense now, thanks. I just wanted to compare the effect to the original and tweak accordingly. Very good work, really goes from mild to wild with saturation. This algorithm was definitely needed. Creates a whole other use for the H9000.

[Puppeteer]

I’m glad it’s useful.  Let me see what I can do about adding the bypass.  It might need a new set of presets, but happy to send them to you.

[Puppeteer]

I’ve created a new Octal version of this algorithm and added in the bypass.  Download available here – https://godlike.com.au/index.php?id=420

Only VSIG at the moment. I fly back to the studio tomorrow where I can unplug the USB drive.  There’s a new manual.  I suggest you route it like this.

Let me know if this does the bypass you are chasing.

 

[joeydego]

Puppeteer wrote:

I’ve created a new Octal version of this algorithm and added in the bypass. Download available here – https://godlike.com.au/index.php?id=420 Only VSIG at the moment. I fly back to the studio tomorrow where I can unplug the USB drive. There’s a new manual. I suggest you route it like this. Let me know if this does the bypass you are chasing.

 

Hey there. Really grateful for your work with this. I’ve never worked with vsig, can you (or anyone) tell me how to import this one? Tried using the typical import algo and it doesnt see the file.

[Matthias Adloff]

Oops. That’s a very strange idea in my opinion! Not laughably silly, but strange.

Two considerations from my technical idiots point of view:

1. Using analog summing may or may not affect the sound, especially the stereo width and/or perceived depth, for the better. But let’s keep this subjective question aside. The H9000 is a digital machine, so it can only do an emulation of analog circuits. Which tons and tons of plugins will do as well. So, if you just go from ADAT/MADI or whatever digital input to another digital output, it will actually sum channels by doing the most simple math (sample value + sample value for n times, where n is the number of tracks), not by moving electrons on the audio path. Which is considered to do the trick alone.

2. If you use any A/D or D/A conversion from the H9000, things are different. People say that they like out-of-the-box summing better. And yes, it does change the transients a lot, actually! But you don’t need a summing box for that, just a DA/AD loop to achieve the same effect. Generate a pure square wave in you DAW, send it to the loop and voilà, it looks completely different now! Admittedly, this is not summing anything, but who knows what happens on your output? I mean, you ears cannot process digital data, so just another D/A has to be done. And then you have it all. Not?

Just my 2 cents.

 

[Fender17]

One may do square wave in digital, but it is not possible to do it in analog. Even if one could use infinite frequencies, which is not the case, exactly square wave is not reachable. Also one would need infinite amount of energy…

So, no magic, that if one makes square wave in digital, when converted to analog, it is not square wave. Math says, it cannot be exactly suare..

https://cs.wikibooks.org/wiki/Praktick%C3%A1_elektronika/Spektrum_sign%C3%A1lu#/media/Soubor:Synthesis_square.gif

[Matthias Adloff]

Cool, thank you!  It was already clear to me that a speaker cannot move anywhere near a square wave, but I didn’t know the model with n sine waves to approximate the result.
Using that model for summing, would it mean that each channel has to be processed, and the result again? Or are there other interferences in an analog summing circuit that makes emulation extremely difficult?

Sorry for those questions, I’m just too curious 🙂

[Fender17]

No, more channels would not help…

That animation shows, how analog signal looks like if one aproximates square wave with sine. It gets better as one uses more and more sines, but look at the end of upward and beginning of downward part. There is considerable up for short time, before it settles at desired value. It is inevitable….

Simply, perfect square wave does not exist in analog world.

The summing in the picture is just mathematical abstraction, not real summing in the gear. If one creates square wave in digital, it has infinite number of sines reaching infinite frequency (if using infinite sample rate)… But because of limited samplerate, digital may represent just frequences up to half of sampling rate. In fact, D/A generates higher frequencies, but they have to be cut off after conversion to analog, they are just “reflections” of lower frequencies.

There are few peculiar things in D/A and A/D that are not very intuitive. It is just pure math…

But back to original topic… I guess, there is not much difference if the summing of signals is done analog or digital. Beside, that analog summing always introduces some kind of alteration/colouring of signal (probably unrecognizable, but still present), while digital does it just perfectly. Unless one deliberately models analogish sound in digital summing…

[Puppeteer]

You’ll need to use the VSIG method as outlined in the manual.

I’ll try to get the algorithm and presets up in a few hours. Just been working on a paid project.

 

joeydego wrote:

Puppeteer wrote:

I’ve created a new Octal version of this algorithm and added in the bypass. Download available here – https://godlike.com.au/index.php?id=420 Only VSIG at the moment. I fly back to the studio tomorrow where I can unplug the USB drive. There’s a new manual. I suggest you route it like this. Let me know if this does the bypass you are chasing.

Hey there. Really grateful for your work with this. I’ve never worked with vsig, can you (or anyone) tell me how to import this one? Tried using the typical import algo and it doesnt see the file.

[Puppeteer]

OK, the updates are now up on my website – https://godlike.com.au/index.php?id=420

I’ve packaged both the original and new algorithms and presets together.  There are 2 algorithms, the original and the one that can be bypassed.

Regarding the utility of this algorithm, the original purpose was just to provide the ability to mix down 4 stereo pairs to 1 stereo pair, ie a simple summing mixer.  I added in a saturation control for each input which emulates the analog warmth that you get from pushing into tubes on an analog mixer, or if set to hard clip, the transistor clipping when a channel overdrives.

No it’s not truly analog, but it has the advantage of a lower noise floor compared to analog, and gives really pleasing results, at least to my ears. It also has a variety of choices for the saturation type, which are subtly different. Also, it’s a damn sight cheaper than an analog summing box (assuming you already have a H9000).

[joeydego]

Circled back around on this as I’ve been doing a lot more work lately. This works fantastically!! Thank you. The 8 in 8 out algo seems algos 3-8 are dead and send no sound, everything comes out of 1 and 2. Still works great and actually is kinda the concept, except I wanted to get 8 stereo pairs in my session and take it from there. 6 of 1 and a half dozen of the other maybe.

 

This idea may seem silly, but I’m no longer considering a 4k dollar summing chain. I’m thrilled with the results.

[Puppeteer]

Hi mate,

Thanks for the feedback.  The block is designed to sum the inputs.  The other 6 outputs are active when you bypass the module, so when you bypass from the panel’s control, the audio is routed directly through the block, as if it isn’t there.

I could add another option to have seperate outputs for the inputs as you describe, where each channel is saturated and the volume can be adjusted.  It’s a reasonably easy change to it. Essentially I could put a mix setting where everything is mixed to Out 1/2 as it currently does, and when turned off, just gives 8 outputs after saturation and volume adjustment. Let me know if this would be valuable.

I’m also looking at building an octal channel strip at some stage in the future, so each of the 8 channels will also have parametric EQ, compression and a gate, with the option to gang each pair as a stereo pair.  Not sure on timing. Things are hectic at this end.

I was really pleasantly surprised at how good the saturation is on the H9000. I often use it in place of my tube pre-amps if I just want that colour.

[joeydego]

Ahhh so this makes sense. I can just route everything to the inputs and don’t need 8 stereo channels in my session to get it back.

if I may ask one feature: it would be tremendous to make a quad mono block otherwise identical. Kicks and snares, bass etc to go there. I’d pay a fair fee for this algo.

[Puppeteer]

Quad Mono block is easy, though you will get the same thing if you just hook up every second input and leave the other side unconnected.  Probably easier for me to do the octal version to be honest, with 8 mono ins and outs. Less VSIG to change.

I’ll try to have a look at it over the next couple of weeks.  It will make a nice change from the insane complexity of the 4 voice synth

[Puppeteer]

I’ve done this update now.  I’ll release it in a couple of days.  You can use it as  4 x stereo or 8 x mono.  You can bypass mix + warmth, bypass just the mix (so 8 x warmers only with 8 outputs), or mix down to stereo as you currently can.

[joeydego]

Much appreciated

[Puppeteer]

Just realized I need to make a change to it so that that mono inputs aren’t panned hard left and right.  I’ll add pan controls into the algorithm so you can pan each of the inputs in the stereo field. I’ll see if I can set them to be balance controls for the stereo sends.  Give me a few more days to sort this out. I’m going to have to add a bit of code to make this work and work sensibly.

 

[Puppeteer]

Just a quick update on this.  I’m still working on it.  The implementation of panning/balance is complicated VSIG spaghetti for this particular algorithm.  I’ve mainly got it working, but there’s a few quirks that I need to fix to make it user friendly.  I’ve got some urgent stuff I’ve had to address outside of the music world, so will get back to this in a couple of weeks.

[joeydego]

Thank you

[Puppeteer]

I’ve resolved the panning issue, and this is nearly ready.  I have some UI questions.

Would you prefer Each page to represent 1 channel (ie the page has Drive, Mix, Pan), or would you prefer all Drive controls on a page, all Mix on a page and all Pan on a page?

Secondly is a single order ok, as it currently is, or would you like to choose the order on each channel.  More customization, but more tedious to set.

Here’s the final block diagram of the algorithm.

[joeydego]

I think each page representing one channel would be more useful, but both ways work fine too.
single order as it is works great, too.

[Puppeteer]

I think I have it finished. JoeyDego, if you are comfortable loading to the H9000 via VSIG, and exporting the algorithm from the H9000 to the USB stick, I could send you a copy for testing.  I’m away from the studio for 3 weeks, so can’t export the algorithm until I get back.

Hit me up at https://godlike.com.au/index.php?id=contact if you’d like to test it.

[Puppeteer]

Here’s what they look like in Emote.

[Puppeteer]

Here’s the upgraded algorithm. This time we’ve upgraded our Mixing algorithm with drive to include the option to use it as 8 Mono channels, each with pan, and an added balance control when used in stereo mode. The download includes our older algorithm, for stereo use, and the manual.

Available now for $5 or free to our Buy Me A Coffee members.

https://www.buymeacoffee.com/godlikeprod/e/143383

[joeydego]

Sent you a couple coffees this morning. Thanks for the work on this. It really does work better than it should.

[joeydego]

Can you elaborate on the differences if any between 10107, 08 and 09??

Looks like 10110 is the finished product and will definitely serve my needs.

[Puppeteer]

The differences are as follows.

Stereo Summing Mixer with Drive (6) is 4 x stereo input channels to 1 stereo output. No bypass available.

Stereo Summing Mixer with Drive (7) is 4 x stereo input channels to 1 stereo output. There is a bypass control that patches the 8 inputs straight to the 8 outputs.

Stereo Summing Mixer with Drive PRO allows either 8 x mono input channels or 4 x stereo input channels.  Each channel has a pan (in mono) or a balance (in stereo mode). Bypass has 3 options. 1) Channels are mixed down to stereo pair 1/2. 2) The 8 outputs have the drive applied only, making this an 8 channel saturator, in either stereo or mono mode. 3) The 8 inputs are connected to the 8 output for true bypass.  This algorithm has a rearranged UI with 1 page per channel.

I haven’t tested the Pro version extensively yet, so if you find something that doesn’t work as expected, please let me know.  The wiring is very complicated for this one, and there is a possibility that I got 1 or 2 wires mixed up.  I’ll do some presets for this one shortly as well, to provide some useful starting points.  If you have some favourite presets, feel free to send them through and I’ll include them in the package.

 

Hope this helps.

[Puppeteer]

I’ve discovered an issue with the bypass wiring of the new block. I’ll post an update soon. Dry signal is bleeding to the outputs at all bypass settings at the moment.

 

[Puppeteer]

I’ve fixed the bypass issue and also fixed an issue with the panning being reversed.  The gain controls are now in percent rather than dB, which will give better control around the top end.  While the knob display’s values to the nearest percent, you can type in decimals if you need finer control.

You can download this at the same link as before at https://www.buymeacoffee.com/godlikeprod/e/143383, and it’s also in the message section for subscribers.

If you have previously downloaded this, please delete the old version from your H9000 and load this one instead.

[Author]

Posts