Jump to content

Pros and Cons of Midi Concertinas


Recommended Posts

We're not talking about startup reed delay here, we're talking about total delay from controller switch closure to first sample out of the audio system.

When I create my sample sets for my iOS and web apps for concertina and several other instruments, I record and edit every single note on my own instruments individually, and then edit the samples to provide the most realistic playback experience.

It's quite easy to add additional startup time to the start of low reed samples to simulate startup delay, but that's more specific to the properties the instrument being modeled, not the end-to-end MIDI triggering and playback system, which is I think the primary topic here. 

You can construct the instrument samples any way you want to make them sound and feel more authentic when playing, either building-in reed startup latency or not, the key is you don't want to have excessive delay in the triggering and rendering system playing that sample.

Edited by Michael Eskin
Link to comment
Share on other sites

5 hours ago, Michael Eskin said:

reed startup latency

By "reed" do you mean a traditional acoustic vibrating reed? I don't think so because there aren't any in your digital instrument. Correct? Could you use the term "key" instead of "reed?" And is the latency the same for all keys?

 

Regards,

Tom

Link to comment
Share on other sites

I mean exactly what I wrote.


The attack transients are included in the per-note samples.

 

Each note on my instrument is sampled to an .wav file and played based on an incoming MIDI trigger for my apps that are MIDI enabled or a screen touch on the live play apps. 
 

As I recorded these samples over a decade ago, my memory of exactly the process I used is a bit fuzzy. Most likely if I followed the same practice as I do today when sampling new instruments, the samples start when the note waveform first is non-zero as displayed in Adobe Audition, includes the attack transient, and represents the sound recorded when I played each button on my instrument.

 

The post processing of the samples is mostly about balancing the volume across the entire instrument as well as micro-tuning of any notes that might not be perfectly in tune.

 

If you're asking do the apps or samples model the differential reed start latency, no, there is nothing specific in the code that does that, it's assuming that's inherent in the note samples attack transients.

 

My per-button concertina samples are used both in my iOS apps as well in my free web-based ABC Transcription Tools.  

 

 

 

Edited by Michael Eskin
Link to comment
Share on other sites

5 hours ago, Michael Eskin said:

The attack transients are included in the per-note samples.

That's all I wanted to hear. Thanks.

I had assumed that you store in memory only the developed steady state sound, but as I ponder, I can understand why it would require so much memory. The steady state sound, being independent of time, would require the information of only a Fourier series.

 

How much memory, by the way, per note? Also, how long a time span of the note is stored? Is the time span the same for each note? Then, when you keep playing the note for an extended time, how is the waveform continued beyond the time period you have stored?

 

Regards,

Tom

Link to comment
Share on other sites

Quote

 

The samples are quite long, a full bellows worth of each note at medium volume, about 6 seconds each, and they include whatever minor variations in volume my arm muscles might have created at the time I recorded them. Those minor imperfections I think add to the authenticity of the sound.  

 

There is no ability to extend the length beyond what's physically possible with the real instrument.  There is no looping of the samples.

 

if you need a note longer than what would be possible on the real instrument, you need to rearticulate the note, just as you would on a real instrument. 

There are limitations to this system:

1) While my iOS apps and ABC Transcription Tools will respond to velocity and volume messages by changing the playback volume, there is no change in timbre based on volume or bellows pressure modulation on sustained notes

2) No change in attack transient behavior or timbre based on velocity

But even with those limits, the samples have served me well over the last decade or so for multiple applications, but I'm not building applications for professional use, primarily mobile apps for silent practice and for use as a instrument in my ABC Transcription Tools.

Mobile examples from over a decade ago:
 

 

 

Use of the samples more recently as an instrument for my ABC Transcription Tools:
 


 

 

 

Edited by Michael Eskin
Link to comment
Share on other sites

I've built (and am learning to play!) a "midi bandoneon".

 

 

 

It uses keyboard switches and 3D printed buttons, a load cell/strain gauge (containing a Wheatstone bridge!) for "bellows", and is controlled by a Teensy (Arduino) - source code and details here. Obviously the keyboard can be configured however you like - I have the tango (bisonoric) layout, a few Peguri (unisonoric) variants, and two Wicki-Hayden layouts set up. 

 

It plugs into my phone to convert MIDI (over USB) to audio, and as of a week or so ago, it has a little audio amp/speakers, so is fully standalone (and can be pretty loud!):

 

image.thumb.jpeg.1cd87e3f74aa1c7bc5a3f547b0ff9293.jpeg

 

With the load cell, it feels very "expressive", and it feels more like a "real" instrument than you might expect. Maybe like a Clavinova/digital piano feels more real than it really is - even though it's never quite like even the most clunky real piano! I still sometimes find myself reaching for the "air button", or having a sense that I've been playing on the pull for too long and need to change direction before I run out of air! It would be interesting to hear what somebody who can play the thing properly could do with it!

 

Primarily, it's been a way for me to learn the keyboard whilst getting a real one made. It's also let me experiment with the exact keyboard layout (resulting in my making some adjustments to the order!). It also lets me practice silently. I don't know if I'll keep playing it when the real thing arrives - maybe, due to it's silent ability, and also because it's a bit smaller/more portable. 

  • Like 1
Link to comment
Share on other sites

37 minutes ago, RatFace said:

I've built (and am learning to play!) a "midi bandoneon".

Rat, impressive instrument, and your playing is very tasteful, pleasing. 

 

Am I correct to assume that the load cell has + and - values so that you get two different notes on the same key, depending whether you push or pull. 

 

Do the sampled sounds contain the start transient of the note, and are their durations the full sounding time you can get with one bellows direction?

 

Regards,

Tom

Link to comment
Share on other sites

2 minutes ago, ttonon said:

Rat, impressive instrument, and your playing is very tasteful, pleasing. 

 

Thanks!

 

2 minutes ago, ttonon said:

Am I correct to assume that the load cell has + and - values so that you get two different notes on the same key, depending whether you push or pull. 

 

Yes - the load cell/strain gauge is pretty well symmetric. I'm using a 3kg load cell. However... because the bandoneon technique is more about using gravity than push/pull, I'm only playing on the pull - because this fixed bellows system (it has a small amount of movement controlled through rubber mounting grommets). For a good bellows emulation, and bi-directional playing, something more like real bellows is needed, I think.

 

2 minutes ago, ttonon said:

Do the sampled sounds contain the start transient of the note, and are their durations the full sounding time you can get with one bellows direction?

 

I'm using a sampled bandoneon sound font by Jörg Bleymehl - it includes the starting transients, and then turns into a loopable clip. They're intended for score playback, so don't really account for the overlap region - where the octave around middle C is available on both sides, but sounds much more mellow on the left. I'm using some EQ processing to emulate that, but it's not ideal!

 

Link to comment
Share on other sites

On 5/15/2024 at 10:06 AM, Chris Ghent said:

My major issue with it is it cannot distinguish between a tipped button and a fully pushed button.  


I can perfectly relate to that indeed, and the same will be true re sudden changes of the bellows pressure, all the more in combination with treating the buttons differently 

 

there‘s so much more to the sound created and shaped by a vintage concertina - it could all be reproduced digitally, I‘m absolutely certain about that, but this would require the same level of sophistication that has gone into creating and permanently improving digital pianos and Hammonds (or other electro-mechanical instruments)

 

so what is discussed here at the moment would have to be only the very beginning of a digital future for our beloved instruments - and of course I would love to see, feel and hear the results some day

 

but still, a „real“ B3 - and Leslie cabinet - is and might always be preferred by the dedicated, and I on my part am perfectly happy each single time when I‘m given the opportunity to play a concert - or even good baby - grand

 

and whatever the outcome for the concertina, it could never fully replace the experience of playing our Lachenals, Wheatstones, Crabbs et al. I‘m sure

 

Edited by Wolf Molkentin
  • Like 1
Link to comment
Share on other sites

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...