steven taylor wood

professional portfolio github

[steven-taylor-wood]

IS53043C_update_12

Write up done!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! At 3:27 on a Friday night!!!!!!!!!

Just a couple things I need to take care of now like the video demonstration, readme file, and tidying up the non-presentation view of the patch so it’s up to scratch with other Max for Live devices.

Well done Steve

[steven-taylor-wood]

IS53043C_update_11

Finalised the UI, really happy with the way it looks now.

Added some nice buttons using the “live.” objects within Max.

As well as a button at the top that flashes when the Markov object bangs, so you can see if the device is functioning while it’s bypassed.

Also labelled everything in the plugin, so users can read what everything does in the help window in Ableton.

Still really behind on the write up - been a rough week.

Presentation is tomorrow! Feeling prepared but very nervous.

[steven-taylor-wood]

IS53043C_update_10

presentation mode!

added a few quality of life features:

every time the user changes the order of the chain it rebuilds the model, so the flow of audio doesn’t stop

using live.thisdevice to essentially loadbang components when the olugin is turned on/ loaded

added nice live.numbox and live.buttons for the UI in presentation mode

need to add feature where when i clear the model it bypasses the plugin, so audio doesn’t stop while markov model isn’t trained

and still need to sort out the plugin working on multiple tracks 

good night x

[steven-taylor-wood]

IS53043C_update_9

had a tutorial with atau today,

he pointed out some extra features/ quality of life things that would be good to implement.

firstly was a bypass mode so you can hear the original midi input without having to turn off the device in ableton, was an easy fix with a couple of gswitches.

secondly was the patch’s ability to be active in multiple tracks at the same time, currently its a bit buggy - he suggested using # to assign individual keys to things, but haven’t really figured out what those things are.

last was an ability to store the midi that is coming out of the markov into sort of presets - potentially through writing this data to a midi file

to be honest, i think just setting up a new midi lane in ableton and recording/ resampling the midi from the markov track is a lot more straightforward and intuitive - so i might just leave as is

write-up is currently at 2200 words minus my bibliography

lots of work ahead of me

[steven-taylor-wood]

IS53043C_update_8

most of the development in max is meeting my expectations of a minimal viable project and I’m pretty happy with it to be honest, potentially I could mess around with the transition matrix to fine tune its output and I definitely need to tweak the UI and get a nice preview mode setup, but. that shouldn’t take long

can now devote some time to the write up, its due in less than a month!

on about 2.6k words right now

goodnight

[steven-taylor-wood]

IS53043C_update_7

Added functionality for transport detection from Live using the plugsync~ object which detects every time playback is stopped/ started in Live, as well as a lot of other useful info -- seems to be a pretty decent solution for now

I’d say with this fix that most of the core functionality of the patch is there, I can now perhaps start to think about UI design and user testing

[steven-taylor-wood]

IS53043C_update_6

ok so i think i’ve managed to figure out a solution already! using borax!!! also i wonder what borax means

time delta between note-on values is used to set the timing of the metro object, which then bangs the markov objects, whereas previously the markov objects were being banged every time a new note came in from live

the sound coming out of it is so much more coherent its crazy

[steven-taylor-wood]

IS53043C_update_5

patch now detects multiple midi notes coming through at once

still unhappy with the way notes are being triggered, need to look into detecting note onset values and work it into the model somehow

also showed the patch to a producer friend today, they noticed the same issue

[steven-taylor-wood]

IS53043C_update_4

I have to go to work soon 

Added functionality to read and interpret velocity information from Live using a similar process to how it reads pitch info.

Also refined how the patch interprets the MIDI data, instead of using the awkward method with the comment box have opted to let the input come through live, and after listening you build the Markov model and hear the result, I think it sounds a lot more coherent this way :-) .

Started playing around with how the patch reacts to a drum sequence in Live that has multiple hits at once, right now it doesn’t sound right because the MIDI values aren't being grouped together -- I’ll be working to fix this next probably using a timer and zl.group to gather the incoming data.

[steven-taylor-wood]

IS53043C_update_3

Reworked how the patch reads MIDI information from Live, opting to use the midiin & midiparse objects instead of notein, which allows for more control over the MIDI data coming from Live.

Also started experimenting with the Borax object, which outputs data regarding note-on and note-off values, which can then be interpreted by the Markov model.

https://docs.cycling74.com/max5/refpages/max-ref/borax.html

[steven-taylor-wood]

IS53043C_update_2

It’s finally making coherent sound!

Implemented a rudimentary way of reading the MIDI from Live, this still needs some major work doing to it, such that it can just instantly capture the MIDI information straight from the clip in Live.

Started testing the patch using a sliced breakbeat in a drum rack, outputting MIDI info to max (as seen above). I found that setting the order of the Markov Chain to the same value as the number of slices yielded the most coherent result...

𝑃[𝑠𝑡+1|𝑠𝑡,…,𝑠2,𝑠1] = 𝑃[𝑠𝑡+1|𝑠𝑡,𝑠𝑡−1,…,𝑠𝑡−(𝑛−1)]

where n = number of slices.

Next up is adding functionality to allow for multiple MIDI notes coming through simultaneously, like in a more conventional drum rack.

[steven-taylor-wood]

IS53043C_update_1

With most of my annoying deadlines out of the way, I can now start dedicating some much-needed attention to this project. Finally started building the bloody thing; on a Saturday evening no less 😎

Settled on building it within Max For Live instead of Max standalone, the workflow seems a lot easier when dealing with incoming MIDI data from Ableton.

I started with implementing basic MIDI in functionality - my next job is finding out a way to store the MIDI data in a way ml.markov can interpret it, maybe through a coll or message object.

Hoping to have a somewhat viable project within the next 2 weeks and be back on schedule.