A Madlibs-like Modular Synthesizer?

Cwejman, Livewire, TipTop Audio, Doepfer etc... Get your euro on!

Moderators: Kent, Joe., luketeaford, lisa

Post Reply
wrngtrls
Learning to Wiggle
Posts: 40
Joined: Wed Jul 15, 2020 2:32 am
Contact:

A Madlibs-like Modular Synthesizer?

Post by wrngtrls » Mon Sep 07, 2020 4:34 pm

So, I'm a Middle School English Teacher, teaching from home, and loving being about to noodle on my modular synths that much sooner after works ends.

I think I can create a Madlibs-style synth/sampler that plays coherent sentences by pressing keys, but I need some help thinking about the most cost-saving yet multifaceted and interesting to actually interact with way to design it.

I'm thinking about starting with a sampler (like say the Bitbox MKII/Micro, or Rossum Assimil8or) to record spoken words and organizing them into folder files "Subject Nouns" "Prepositions" "Verbs" "Adjectives" "Object Nouns" etc. And then taking some kind of sound assigning module (like Polyend's Preset, or Hermod Pyramid, or Rossum's Control Forge, or other things I'm ignorant of) and basically designing a creative, on-the fly sentence writer where you could press buttons and it would generate nonsense poems for kids to engage with. Sentences such as:

"The aardvark" "dances on" "Wednesdays."
-or-
"My brother" "smells like " "a bag of beets."
-or-
"Susan" "found" "a raging wildfire".

It needs to be a tactile modular synth to keep kids interesting, so no DAW business. I'd love something that I can easily just upload student voices to throw in the mix.

How could this Madlibs Modular Synthezier be created? What modules would be perfect for this that I don't know about. (I have a Medusa and Assimil8or, that's all to start).

Thanks!

Arneb
Veteran Wiggler
Posts: 740
Joined: Thu Jul 04, 2019 4:50 pm

Re: A Madlibs-like Modular Synthesizer?

Post by Arneb » Mon Sep 07, 2020 6:48 pm

Hm, the Assimil8or doesn't look like the best choice for this to me. Seems to be rather resampling-focused. You'll want something which has CV over file choice... I'd probably get a bunch of TipTop Ones, or the VPME QD.

Then, you'd need a bunch of random CV sources for said CV over file choice inputs. DivKid RND STEP looks good. Or, if you want more immediate control, a bunch of offset sources. In that case my gut feeling tells me to go with faders over knobs for kids, e.g. NE Lapsus Os.

To get those single-word samples to sound like a sentence, you'll need to automatically trigger the second sample when the first has ended. So, a bunch of envelope followers with EOC outputs. You could rack Maths twice, or go for Doepfer A-135-4C into Circuit Abbey VERTY.

Finally, you need something to trigger the whole sentence chain. And you want it to be kid friendly. Let's face it, there is only one correct choice.

yellowecho
Wiggling with Experience
Posts: 297
Joined: Fri May 09, 2014 1:20 pm

Re: A Madlibs-like Modular Synthesizer?

Post by yellowecho » Mon Sep 07, 2020 9:56 pm

You could probably do this with the ALM Squid. Put each different sentence parts on different channels and use cues and pools to jump between them.

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Tue Sep 08, 2020 1:48 am

Cool project!

Arneb seems to have most of the overall signal flow worked out for you. There's lots of ways to generate random CV, but I think there would be plenty of randomness even if you just had kids slide a slider or dial a pot that selects a number from 1 to 40. The kids can supply the "random" element.

I have a Bitbox Micro and I was looking to get CV control of sample selection. But alas... it's not possible. I just checked Assimil8or and Squid Salmple... same there too. I've got a Tiptop ONE and it's great. I also have an Erica Pico Drums, which has CV for sample selection. The nice thing is that these smaller sample players are much more immediate and easy to work with.

I really like the idea of recording the kids' voices, which would make it way more interactive and personalized. I know this would introduce significantly more design complexity, but I think it would be a lot more fun for them. Instruo Arbhar has a microphone and CV control over six 10-second samples. For each part of the sentence, you could record all six buffers with the kids and then run the full patch.

And hell yes... that Big Honking Button module is perfect!

User avatar
asgood
Learning to Wiggle
Posts: 16
Joined: Wed Sep 21, 2016 8:55 am
Location: Pittsboro, NC, USA

Re: A Madlibs-like Modular Synthesizer?

Post by asgood » Thu Sep 10, 2020 9:03 am

I LOVE this idea!

One possibility would be to do all of the samples of a certain type (say, "nouns") into one sample, with plenty of room between them, and then use slices and slice control over CV / midi to trigger them.

A thing I love about the Bitbox mk2 is you can resample into an empty slot and save those files.
So you could have a sample slot each for "Subject" "Verb" "Object". Then trigger those with whatever (intentional or random or anywhere in between), and when you find something you like, resample into the Bitbox.

User avatar
EATyourGUITAR
has no life
Posts: 5798
Joined: Tue Aug 31, 2010 12:24 am
Location: Providence, RI, USA

Re: A Madlibs-like Modular Synthesizer?

Post by EATyourGUITAR » Thu Sep 10, 2020 12:18 pm

MAX/MSP
WWW.EATYOURGUITAR.COM <---- MY DIY STUFF

djdiscord
Learning to Wiggle
Posts: 22
Joined: Thu Apr 09, 2020 2:20 am

Re: A Madlibs-like Modular Synthesizer?

Post by djdiscord » Thu Sep 10, 2020 4:03 pm

I'm not sure why people are saying the Assimil8or wouldn't work well here, I'm not sure that they understand the features well. You can very finely tune CV control over 8 different samples (if you merge the samples together as the same file) per sample channel/track.

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Thu Sep 10, 2020 4:49 pm

djdiscord wrote:
Thu Sep 10, 2020 4:03 pm
I'm not sure why people are saying the Assimil8or wouldn't work well here, I'm not sure that they understand the features well. You can very finely tune CV control over 8 different samples (if you merge the samples together as the same file) per sample channel/track.
While Assimil8or can technically get the job done, AFAIK it does not have CV control over per-channel sample selection (meaning only zones or chopping of single samples is possible). I'd be delighted to hear otherwise, of course. But I've read the manual and also the updates and I'm not seeing it. SEE UPDATE BELOW. ASSIMIL8OR DOES HAVE CV FOR SAMPLE SELECTION.

For this particular project, I was making the assumption (based on his request that it be done "easily") that chopping samples in real-time context did not fit this use case. I think the idea is that this is all going to be done somewhat on-the-fly in front of the kids.

But now that we've introduced these two different ways of doing it (separate samples vs zone/chops of single samples), let's see what the OP thinks. They've been conspicuously quiet here. It would be nice to know what they are thinking, right?
Last edited by tumblz on Mon Oct 19, 2020 12:46 am, edited 1 time in total.

OHEXOH
Common Wiggler
Posts: 233
Joined: Mon Mar 28, 2016 2:32 pm
Location: Portland
Contact:

Re: A Madlibs-like Modular Synthesizer?

Post by OHEXOH » Fri Oct 16, 2020 6:28 pm

tumblz wrote:
Tue Sep 08, 2020 1:48 am
I have a Bitbox Micro and I was looking to get CV control of sample selection. But alas... it's not possible. I just checked Assimil8or and Squid Salmple... same there too. I've got a Tiptop ONE and it's great. I also have an Erica Pico Drums, which has CV for sample selection. The nice thing is that these smaller sample players are much more immediate and easy to work with.
If you treat it as a multisample collection of WAVs - you could have a word for every note on the scale. Or you could simply split a single sample file into 100s of slices.
Gear for sale (Reverb): https://bit.ly/2Sb90oc

wrngtrls
Learning to Wiggle
Posts: 40
Joined: Wed Jul 15, 2020 2:32 am
Contact:

Re: A Madlibs-like Modular Synthesizer?

Post by wrngtrls » Sun Oct 18, 2020 2:31 am

Arneb wrote:
Mon Sep 07, 2020 6:48 pm
Hm, the Assimil8or doesn't look like the best choice for this to me. Seems to be rather resampling-focused. You'll want something which has CV over file choice... I'd probably get a bunch of TipTop Ones, or the VPME QD.

Then, you'd need a bunch of random CV sources for said CV over file choice inputs. DivKid RND STEP looks good. Or, if you want more immediate control, a bunch of offset sources. In that case my gut feeling tells me to go with faders over knobs for kids, e.g. NE Lapsus Os.

To get those single-word samples to sound like a sentence, you'll need to automatically trigger the second sample when the first has ended. So, a bunch of envelope followers with EOC outputs. You could rack Maths twice, or go for Doepfer A-135-4C into Circuit Abbey VERTY.

Finally, you need something to trigger the whole sentence chain. And you want it to be kid friendly. Let's face it, there is only one correct choice.
Whoa - I totally didn't realize people weren't ignoring this post!

This big honking button is amazing and perfect trigger event.

My novice wiggling is getting in the way of understanding all of what your envisioning here, but this is exactly why I asked for help.

I love the idea of a BHB triggering RND STEP triggering TIPTOP ONE just for "nouns" Then the same set up just for "verbs;" "adjectives" etc. It would make the system huge but instructional. The micro sd cards on the TTO look really helpful to switch samples with ease.

Why Maths twice?

Thanks for your suggestions!

wrngtrls
Learning to Wiggle
Posts: 40
Joined: Wed Jul 15, 2020 2:32 am
Contact:

Re: A Madlibs-like Modular Synthesizer?

Post by wrngtrls » Sun Oct 18, 2020 2:40 am

tumblz wrote:
Tue Sep 08, 2020 1:48 am
Instruo Arbhar has a microphone and CV control over six 10-second samples. For each part of the sentence, you could record all six buffers with the kids and then run the full patch.
I thought about this one because of the condenser mic in the module. Giving kids a Zoom and having them load up words and phrases on an sd card was what I was thinking, but if the machine can just record on the fly, this might be neat too.

User avatar
dragulasbruder
Common Wiggler
Posts: 231
Joined: Sun Oct 23, 2011 9:57 pm
Location: Chicago

Re: A Madlibs-like Modular Synthesizer?

Post by dragulasbruder » Sun Oct 18, 2020 2:47 am

I appreciate the nature of the post and everything, but why no DAWs? how is watching you noodle a modular on Zoom more engaging to kids than using a tool they can actually access? Other than acoustic instruments, it's almost certain these kids' first tools to make sound will be DAWs.

If you want to keep a "modular synth" philosophy, VCV rack is an excellent solution that students can replicate on their own time.

But if this is more for your own fun (a totally legit reason) then grabbing a couple of Distings couldn't hurt. VC sample selection for days. You could either use multiple units for different parts of speech, or arrange the samples into different voltage ranges and use one unit (might be tricky, but fun and edifying too)
Now offering sliding-scale SYNTHESIS and AUDIO PRODUCTION LESSONS during the pandemic--please PM for details!

WTB Wish List: https://www.modulargrid.net/e/racks/view/1224624

MY MUSIC: https://www.soundcloud.com/wish-fulfillment

User avatar
drxcm
Veteran Wiggler
Posts: 623
Joined: Fri Feb 07, 2014 10:31 pm

Re: A Madlibs-like Modular Synthesizer?

Post by drxcm » Sun Oct 18, 2020 3:15 am

tumblz wrote:
Thu Sep 10, 2020 4:49 pm
djdiscord wrote:
Thu Sep 10, 2020 4:03 pm
I'm not sure why people are saying the Assimil8or wouldn't work well here, I'm not sure that they understand the features well. You can very finely tune CV control over 8 different samples (if you merge the samples together as the same file) per sample channel/track.
While Assimil8or can technically get the job done, AFAIK it does not have CV control over per-channel sample selection (meaning only zones or chopping of single samples is possible). I'd be delighted to hear otherwise, of course. But I've read the manual and also the updates and I'm not seeing it.

For this particular project, I was making the assumption (based on his request that it be done "easily") that chopping samples in real-time context did not fit this use case. I think the idea is that this is all going to be done somewhat on-the-fly in front of the kids.

But now that we've introduced these two different ways of doing it (separate samples vs zone/chops of single samples), let's see what the OP thinks. They've been conspicuously quiet here. It would be nice to know what they are thinking, right?
Each channel of Assimil8or can have CV selection of 8 different samples (using zones). These can be triggered randomly instead of using CV, or alternatively addressed via CV .
Each sample could have multiple phrases, each CV selectable by using Sample Start position.
I'm thinking it can easily do the job, just requires a bit of setup.

Lets say you split the phrases into 4 seconds each, you could have 8 phrases per sample, times 8 zones = 64 phrases. And all you'd need to do is have a preset CV addressing sample start time, and keep the zones on randomise.

Then there are 8 channels of that.

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Sun Oct 18, 2020 2:01 pm

dragulasbruder wrote:
Sun Oct 18, 2020 2:47 am
I appreciate the nature of the post and everything, but why no DAWs? how is watching you noodle a modular on Zoom more engaging to kids than using a tool they can actually access? Other than acoustic instruments, it's almost certain these kids' first tools to make sound will be DAWs.

If you want to keep a "modular synth" philosophy, VCV rack is an excellent solution that students can replicate on their own time.
Regarding modular-vs-VCV, I think there are two points worth distinguishing here.

The first point is which one will better stimulate the curiosity of the kids. I think a tactile/haptic modular setup has more inherent appeal. VCV Rack is great once you get into it, but it's not immediately visually stimulating imo. I think those first moments of stimulating curiosity would be key in a middle school setting. And watching someone patch up a modular in person is going to be way more intriguing than watching someone do the same-old-same-old on a computer.

But the second point, which I think is ultimately much more important, is the general accessibility of modular for the kids. And for that, no question, the VCV Rack would be the way to go. But why not make both available, right? Start with the crazy Willy Wonka machine and then if one of the kids takes an interest, get them started with VCV rack.

A little side point here is that clearly modular-based demonstration would be highly tech-centric and not English-centric. You could more quickly and effectively convey the English concepts here with chalk and a board. But this is an interesting project and might make a lot more sense in the context of a science/tech class and maybe you could find another teacher to partner with. Otherwise, I agree that it seems a bit self-indulgent for a school activity.
Last edited by tumblz on Sun Oct 18, 2020 2:42 pm, edited 1 time in total.

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Sun Oct 18, 2020 2:20 pm

drxcm wrote:
Sun Oct 18, 2020 3:15 am
tumblz wrote:
Thu Sep 10, 2020 4:49 pm
djdiscord wrote:
Thu Sep 10, 2020 4:03 pm
I'm not sure why people are saying the Assimil8or wouldn't work well here, I'm not sure that they understand the features well. You can very finely tune CV control over 8 different samples (if you merge the samples together as the same file) per sample channel/track.
While Assimil8or can technically get the job done, AFAIK it does not have CV control over per-channel sample selection (meaning only zones or chopping of single samples is possible). I'd be delighted to hear otherwise, of course. But I've read the manual and also the updates and I'm not seeing it.

For this particular project, I was making the assumption (based on his request that it be done "easily") that chopping samples in real-time context did not fit this use case. I think the idea is that this is all going to be done somewhat on-the-fly in front of the kids.

But now that we've introduced these two different ways of doing it (separate samples vs zone/chops of single samples), let's see what the OP thinks. They've been conspicuously quiet here. It would be nice to know what they are thinking, right?
Each channel of Assimil8or can have CV selection of 8 different samples (using zones). These can be triggered randomly instead of using CV, or alternatively addressed via CV .
Each sample could have multiple phrases, each CV selectable by using Sample Start position.
I'm thinking it can easily do the job, just requires a bit of setup.

Lets say you split the phrases into 4 seconds each, you could have 8 phrases per sample, times 8 zones = 64 phrases. And all you'd need to do is have a preset CV addressing sample start time, and keep the zones on randomise.

Then there are 8 channels of that.
Worth noting that a "zone" is not a "sample", but a designated portion of a sample. You can turn multiple samples into a single sample with zones, but it requires more preparation than just loading up separate samples. It would also be impossible to do this quickly on the fly during a demonstration.

For this particular use case, what I think would make more sense for Assimil8or would be to record one sample on-the-fly into each channel and have all 8 triggers rigged up to an 8x switch that is receiving a random CV.

button/trigger --> random CV --> 8x switch --> 8 separate Assimil8or channels


SEE UPDATE BELOW!
Last edited by tumblz on Mon Oct 19, 2020 12:45 am, edited 1 time in total.

User avatar
drxcm
Veteran Wiggler
Posts: 623
Joined: Fri Feb 07, 2014 10:31 pm

Re: A Madlibs-like Modular Synthesizer?

Post by drxcm » Mon Oct 19, 2020 12:11 am

tumblz wrote:
Sun Oct 18, 2020 2:20 pm
drxcm wrote:
Sun Oct 18, 2020 3:15 am
tumblz wrote:
Thu Sep 10, 2020 4:49 pm
djdiscord wrote:
Thu Sep 10, 2020 4:03 pm
I'm not sure why people are saying the Assimil8or wouldn't work well here, I'm not sure that they understand the features well. You can very finely tune CV control over 8 different samples (if you merge the samples together as the same file) per sample channel/track.
While Assimil8or can technically get the job done, AFAIK it does not have CV control over per-channel sample selection (meaning only zones or chopping of single samples is possible). I'd be delighted to hear otherwise, of course. But I've read the manual and also the updates and I'm not seeing it.

For this particular project, I was making the assumption (based on his request that it be done "easily") that chopping samples in real-time context did not fit this use case. I think the idea is that this is all going to be done somewhat on-the-fly in front of the kids.

But now that we've introduced these two different ways of doing it (separate samples vs zone/chops of single samples), let's see what the OP thinks. They've been conspicuously quiet here. It would be nice to know what they are thinking, right?
Each channel of Assimil8or can have CV selection of 8 different samples (using zones). These can be triggered randomly instead of using CV, or alternatively addressed via CV .
Each sample could have multiple phrases, each CV selectable by using Sample Start position.
I'm thinking it can easily do the job, just requires a bit of setup.

Lets say you split the phrases into 4 seconds each, you could have 8 phrases per sample, times 8 zones = 64 phrases. And all you'd need to do is have a preset CV addressing sample start time, and keep the zones on randomise.

Then there are 8 channels of that.
Worth noting that a "zone" is not a "sample", but a designated portion of a sample. You can turn multiple samples into a single sample with zones, but it requires more preparation than just loading up separate samples. It would also be impossible to do this quickly on the fly during a demonstration.

For this particular use case, what I think would make more sense for Assimil8or would be to record one sample on-the-fly into each channel and have all 8 triggers rigged up to an 8x switch that is receiving a random CV.

button/trigger --> random CV --> 8x switch --> 8 separate Assimil8or channels
You are incorrect. You can load up to 8 separate samples per channel using the Zones function.
They are individual samples, not portions of a single sample.
Furthermore you can choose how the zone is selected. You can have each successive gate trigger advance to the next zone, randomly pick one of the zones, read a CV to choose the zone on gate trigger, or change the zone at any time using continuous CV.

Watch this video


tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Mon Oct 19, 2020 12:43 am

drxcm wrote:
Mon Oct 19, 2020 12:11 am
You are incorrect. You can load up to 8 separate samples per channel using the Zones function.
They are individual samples, not portions of a single sample.
Furthermore you can choose how the zone is selected. You can have each successive gate trigger advance to the next zone, randomly pick one of the zones, read a CV to choose the zone on gate trigger, or change the zone at any time using continuous CV.
I stand corrected! I think when I read the manual I didn't realize there were there two different types of sample handling within "Zones". I had even asked a question about CV selection in the Assimil8or feature request post (but didn't get an answer). This is great to know because it's really nice to be able to flip between samples in a variety of different use cases. Great news. Much appreciated.

User avatar
Ears
Wiggling with Experience
Posts: 274
Joined: Wed Apr 11, 2018 2:43 pm
Location: Chicago

Re: A Madlibs-like Modular Synthesizer?

Post by Ears » Mon Oct 19, 2020 11:37 am

wrngtrls wrote:
Sun Oct 18, 2020 2:31 am
Arneb wrote:
Mon Sep 07, 2020 6:48 pm
Hm, the Assimil8or doesn't look like the best choice for this to me. Seems to be rather resampling-focused. You'll want something which has CV over file choice... I'd probably get a bunch of TipTop Ones, or the VPME QD.

Then, you'd need a bunch of random CV sources for said CV over file choice inputs. DivKid RND STEP looks good. Or, if you want more immediate control, a bunch of offset sources. In that case my gut feeling tells me to go with faders over knobs for kids, e.g. NE Lapsus Os.

To get those single-word samples to sound like a sentence, you'll need to automatically trigger the second sample when the first has ended. So, a bunch of envelope followers with EOC outputs. You could rack Maths twice, or go for Doepfer A-135-4C into Circuit Abbey VERTY.

Finally, you need something to trigger the whole sentence chain. And you want it to be kid friendly. Let's face it, there is only one correct choice.
Whoa - I totally didn't realize people weren't ignoring this post!

This big honking button is amazing and perfect trigger event.

My novice wiggling is getting in the way of understanding all of what your envisioning here, but this is exactly why I asked for help.

I love the idea of a BHB triggering RND STEP triggering TIPTOP ONE just for "nouns" Then the same set up just for "verbs;" "adjectives" etc. It would make the system huge but instructional. The micro sd cards on the TTO look really helpful to switch samples with ease.

Why Maths twice?

Thanks for your suggestions!
BHB would make sense because it’s easy to hack. Coding sounds hard, but really in this case it’s pretty easy. You could easily take the existing code, add a few lines to pick a random sample each time (example code: https://github.com/theacodes/Winterbloo ... /random.py ), have the kids record samples to load on each button, and then line up three buttons - a subject, verb and object (for example). The kids would be able to push button one, two, three - and voila! a sentence. Then you could get fancy and add something to trigger the buttons in sequence demonstrating another concept - CV control.

Or you could be more literal with the madlibs concept and have a button loaded with the word type called for in actual madlibs, read the madlib and press the right button for the kind of word asked for.

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Mon Oct 19, 2020 3:10 pm

Ears wrote:
Mon Oct 19, 2020 11:37 am
BHB would make sense because it’s easy to hack. Coding sounds hard, but really in this case it’s pretty easy. You could easily take the existing code, add a few lines to pick a random sample each time (example code: https://github.com/theacodes/Winterbloo ... /random.py ), have the kids record samples to load on each button, and then line up three buttons - a subject, verb and object (for example). The kids would be able to push button one, two, three - and voila! a sentence. Then you could get fancy and add something to trigger the buttons in sequence demonstrating another concept - CV control.

Or you could be more literal with the madlibs concept and have a button loaded with the word type called for in actual madlibs, read the madlib and press the right button for the kind of word asked for.
Great suggestions! I'd say start with Ears' last suggestion of reading the madlibs and pressing the button with pre-loaded words as a "proof of concept".

Personally, I think the immediacy of recording the kids directly into a sampler would really elevate the immediacy and interactivity. But if that's not possible, get the samples loaded beforehand so you and the kids aren't futzing around for 30 minutes wasting valuable class time recording/transferring samples. Record the kids one day with a Zoom or whatever and then load the samples outside of class, prior to the day of the actual demo.

User avatar
Sync
Common Wiggler
Posts: 229
Joined: Sat Jul 29, 2017 5:40 am
Location: California

Re: A Madlibs-like Modular Synthesizer?

Post by Sync » Mon Oct 19, 2020 5:13 pm

Interesting idea. I've coded up purely text versions of random sentence generators many times on the computer, so I've learned a few things in doing that...

I always divided up the problem into to parts. Sentence structures, and the word lists. The sentence structures would define how to assemble the various word types into a sentence, and might include some hard coded parts, like "the", "like", "and" or "or" to combine two subphrases with a particular connecting word.

And each "word" in the sentence structure was essentially, the name of the word list to be used to select the random word for that position:

The noun verbed the adj noun advb.

This sort of thing is really easy to do in pure software, but doing it in modular hardware seems like it could be challenging.


It seems like the root of the problem is in designing how you select the samples. One option would be to look for a sampler that allows for CV selection of a fairly large set of samples, enough to contain ALL of your words in one "bank" selectable by CV. This could be tricky to calibrate as slight variations in voltages could cause it to pick a word from the wrong list. Radio Music is one budget sampler that can handle up to 75 individual samples in a bank, and you could have 16 banks of different combos for interesting variations. However, you'd like to have an End of Sample output in order to advance to the next word, which Radio Music doesn't have. And without that you'd have less control over the cadence of the sentence, each word would have to be the same duration.

But, here's a diagram of one way I think it might be made to work to generate entire random sentences that conform to a particular sentence structure. Not exactly madlibs, but that might be even easier, depending on how many "blanks" are in the sentence at hand:
SentenceGenerator.png
Note that the list-size sequencer needs to be synchronized to the sentence structure sequencer such that they'd probably be two tracks on the same sequencer, one to sequence the words in the sentence, and the corresponding one to provide the word list size for that step in the sequence.

And course there's many possible enhancements-- you would probably want a sequencer for the sentence structure that you can trigger to play through the sequence and stop until it's triggered again, so you don't get a zillion sentences all run together. There's where the BHB comes in.

And also, disclaimer-- calibration might be a challenge, if something is off it might generate all sorts of nonsense. Also, the larger the list of words is, the harder it will be to calibrate. And you probably would want a sequencer that is not quantized to notes, that can generate any voltage in between as the word list beginnings may not be on note boundaries.

A completely different approach could be to have a multi-bank sampler where your sentence sequencer picks the banks and each bank has a note list of a particular type. You still have the problem of picking the word in the list, where lists may be of different lengths. On the other hand, you could force all the lists to the same length by duplicating words if necessary, until all your lists are the same length which might make selection from an array of banks easier, though it could affect how often certain words pop up in an less than optimal way. I'm not sure of the best sampler for that-- the STS has the bank capability, but each bank only holds 10 samples, and it doesn't appear to support bank select by CV. I've no experience with any other samplers, but there's a bunch out there and other posters to this thread have mentioned some of them.

A row of budget samplers might work as well, but again you'd like to have end-of-sample output which neither the Play, Radio Music or TipTop One has. It doesn't look like the Disting MK4 does either, but maybe you could entice them to enhance it with that, they've been responsive to user requests.

One other thing I learned, is that fine tuning the word lists can make a world of difference in the results...
--

Sync

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Mon Oct 19, 2020 7:33 pm

Sync raises some great meta-level points, especially with regard to sequencing and sentence coherence.

With regard to sampler, the 4MS Stereo Triggered Sampler would be ideal. Most importantly, it has an end-of-sample gate output. It also has a record function. The downside is that it only has CV control over 20 samples at a time (2 banks of 10 samples). You can store up to 60 banks with 10 samples each in the sampler, but CV can only select from two banks at a time.

Image

Because of the STS' end-of-sample functionality, I think you could get away with just lining them up with a Mutable Stages.

Image

I think you can use Stages as either a sequential Sample & Hold (in a chain of end-of-sample gates) or manually triggering and CVing each segment of the "sentence". Here's a quick patch sketch...

Image
Last edited by tumblz on Tue Oct 20, 2020 1:03 pm, edited 3 times in total.

User avatar
megarat
Wiggling with Experience
Posts: 349
Joined: Fri Sep 20, 2019 4:09 pm
Location: Squirt Island, USA

Re: A Madlibs-like Modular Synthesizer?

Post by megarat » Tue Oct 20, 2020 1:05 am

While reading this fun thread, I’ve been thinking about how to detect the end-of-sample for a module that doesn’t have this ability. Comparator? Envelope follower? Gaps between words might be a fatal weakness to either.

tumblz
Common Wiggler
Posts: 52
Joined: Tue Dec 31, 2019 1:05 pm

Re: A Madlibs-like Modular Synthesizer?

Post by tumblz » Tue Oct 20, 2020 3:42 pm

megarat wrote:
Tue Oct 20, 2020 1:05 am
While reading this fun thread, I’ve been thinking about how to detect the end-of-sample for a module that doesn’t have this ability. Comparator? Envelope follower? Gaps between words might be a fatal weakness to either.
I can see two solutions for this problem.

1) Layer noise into each sample, just enough to keep an envelope follower from hitting 0, but not so much that it's distracting.

2) Layer a low-frequency or high frequency tone into each sample. An envelope follower will skip the gaps. Then use a high-pass or low-pass filter at the end of the chain to remove the unwanted frequency.

Post Reply

Return to “Eurorack Modules”