Hi Nick. I love Bristol. It's not like anything else...has a great chunky feel to it...really dig it!
Do you think you might consider adding a clavinet in future? And possibly a wurlitzer e piano? Goodness knows how hard those are to program virtually.
Also I love the look of the Rhodes....it would be nice if it had a bit more bite if you don't mind me suggesting.....sorry I am sure it took a lot of work just to get it where it is....
I can run VST's under Studio to go but I'd much rather run Bristol.
Did you do all the graphics? They look ace! I really like this synth. It's different to what's around. Looking forward to future developments!
The Rhodes electric piano implemented by bristol is based on the FM algorithm (startBristol -dx, voices from 500 upwards), pretty similar to the Yamaha PF series from the 80's, the pf10 and pf15. These were both DX synthesisers released at a lower price with only preset piano voices. The voices could be improved but I would agree that the sound is rather thin as a Rhodes.
The original plan for the Rhodes was to implement a sampler (which is kind of specified in its capabilities however has never been implemented) and then feed in those samples from my own stage 73 m1. The problem with implementing this as a sampler is that the quality is typically dependent on the samples rather than the actual algorithms used to reproduce those samples, and bristol could not improve on the Linux Sampler results, something which does then beg the question as why it should be implemented.
Yes, bristol does do all the graphics. It does not use any of the typical GUI toolkits but implements its own bitmap manipulation toolkit called brighton with a thin shim onto X11 natively. The result is that it not as flexible as other interfaces however it was designed to look like the original equipment and has no additional dependencies other than a couple of X11 header files. That makes it far easier to install, and to maintain too.
I will give a bit of thought into your suggestion of how the Rhodes might be physically modelled rather than emulated with FM or sampled. The hardware was amazingly simple, not that it makes it easier to model - the oscillations of the tines was quite complex in how it changes with the strength of hammering, almost growling when struck hard.
The Rhodes emulator on Bristol rather an amazing acheivement in itself... And I can't imagine them being better than what they already are.
Changes to the instrument (at least I would like to see) are little things such as chorus control, tremolo control, and perhaps a ASDR built-in.
As great as the instrument sounds, sometimes I wish it would just fade, rather than hold the note I'm pressing.
Oh and as for the bristol soundbanks... I am working on them (slowly).
It's taking longer than I thought; partly due to Christmas over the horizon, but mostly due to the care I'm taking into it.
For each instrument I think I might only make about 10-20 patches, but I'm checking and re-checking that the patches are of the upmost quality and justify the best of what bristol can offer...
I hope you understand.
These are nice requests. I was intending to add a stereo chorus in since it was a very typical effect to space out the sound. It will require a few changes here and there, not least because I don't really want one on the FM emulation.
I think your report that the notes hold rather than fade is a bug. I will review the current memories and make some changes as they should actually fade out.
I checked out the envelope stuff and confirmed your report - many thanks for making it. I am probably not going to make general ADSR envelope parameters available however they were broken. The reason I do not want to make them available is that every voice has 6 envelopes due to the way the FM synth works, hence rather than allow the Rhodes to access them I would have to advise starting the -dx and going to memories 500 upwards. Will open a bug against it tomorrow and get a fix out, I think I know what it is.
The next release will incoporate a flanger in the Rhodes algorithm (not in the DX for other reasons), it should be capable of flanging chorus, tremelo (stereo panning in this case) and vibrato. All this should be in the next upload.
The issue with the decay should now be fixed, it went into 0.10.12 recently uploaded. The cause was a change to the sustain control to be linear rather than logarithmic. This gives better control but had the side effect that these voices barely decayed. The memories were reloaded into the DX to be reworked. Also resolved related issues with the decay on the FM harmonic operators.
Integrated a flanging chorus also. Both these should improve the results however if I had some time I would redo the memories again to emulate the dual hit of the hammer on the tine and add some distortion with high gain of the harmonic operators when at high velocity.
I've just downloaded the source... I think I'm the first person to do so...
I'll have to check it out in the morning, It's late here, and I'm tired.
Though I am excited to see what I hear tomorrow.
Yes, Well Done. It certainly has done the trick.
Took a small moment to figure out the tremolo/flange effect... but it's (as ever) considerably improving.
First thing I played was To Build A Home by Cinematic Orchestra... It works alot better now.
So the sustain controls are logarithmic now?
Not that I want to argue the issue, or question the programming. But wouldn't a Cosine fade be a more natural feature? Perhaps something for you to decide in your alpha stages for the next release...
Pleased things have improved. Here is the low down on the levels and changes:
The GUI controller called sustain just sends a linear value somewhere between 0 to 1.0, it's registered as a floating point number in the GUI however the MIDI library transfers this using coarse/fine resolution as a 14 bit continuous controller. The engine interprets the values depending on identifiers sent with them, and the envelope interpretted the sustain into a logarithmic scale. This put a lot of weight in the last 1/4 turn, and in a recent release I turned this into a linear controller as it feels better.
So, it is now linear a controller.
Fading between the stages is currently exponential, for each subsequent sample the current envelope gain level is multiplied by some value, something that takes place until the gain level reaches the target level. I don't see that changing soon however I would consider other options.
Ah, I get you now.
6 Months ago that wouldn't of made any sense to me...
Looks like my c++ classes are paying off! ;)
Log in to post a comment.