Realistic fast tremolo MIDI bass part programming

[Djabthrash]

Hi there,

i'm one of those people using a VSTi (Trilian) for bass, and really happy with it, except when it comes to super fast tremolo parts (very common in extreme genres such as brutal death metal), where it sounds weird/robotic/unrealistic/unpleasant/annoying.

I'm not gonna buy a bass and i want to get those fast parts to sound decently realistic. The other parts sound awesome and realistic, and people know how to program MIDI drums to sound realistic even for extreme parts, so i guess there should be a way to do the same with fast tremolo bass parts.

This is an issue that has been mentioned in many past bass VSTi MIDI programming threads, and yet nobody seemed to be able to solve it.

For now i do the following (in addition to the usual humanization (random timing and velocity changes with my DAW) i do to all my MIDI bass parts) but i guess it's kinda wrong or not enough at least :

-emphasize (i.e increase velocity) on the first note (out of 4 quarter notes) of each series of quarter notes
-for the other 3 notes of each 4 quarter notes group, i do the same kind of velocity changes i would do for a drum roll

which in the end resembles the following, velocity-wise :
100 / 80 / 82 / 78

Any ideas ?

Here are mine :

-use way lower velocity for tremolo riffs than for normal riffs ? (not sure about that, because i can think of say finger players where the tremolo part would have less energy, but there are also pick players who pick the shit out of their strings during fast parts (Fear Factory ?), and i'm using a finger played but very "rattly" patch ) -> just tried it and it doesn't solve the issue, it just does what expected, i.e changing the playing style from soft/round/fingerish to harder/clankier/sharper/pickier
-try using a different cutoff/legato/sustain mode and/or articulation in Trilian ? -> maybe Staccato articulation would work (idea taken from MetallyGuitarded in a previous another bass programming thread) ? -> tried it, didn't seem to fix the issue (it seems to really come down to MIDI programming (velocity, timing, note length))
-the most extreme and cumbersome : playing it with a guitar, and then convert the DI into MIDI using a Melodyne/VariAudio kind of plugin

HELP

example here (fast tremolo part during the blast beat part) :

drums + bass :

http://dl.dropbox.com/u/15008135/Rea...ot02_wBass.mp3

bass alone :

http://dl.dropbox.com/u/15008135/Rea...2_BassOnly.mp3

NB : i think the bass sounds fine during the non tremolo part, and this is the kind of tone i like (Slipknot, Cannibal Corpse, Pantera, Madball, Earth Crisis), namely some kind of Ampeg tone with rattly strings

I guess it comes down to "think of how a real bass player does it and it how it would translate it termes of MIDI information", but i have a hard time figuring this out...

Pics :

my MIDI score (zoom in if needed) :



my Trilian settings :








EDIT : found out i was using the "sequential" Round Robin mode in Trilian, which contribute to this unnatural/robotic feeling on tremolos (but not as much as no Round Robin at all trust me). I tried switching to the "random" Round Robin mode (i guess the "random full cycle" mode is less random than the "random" mode, yet more random than the "sequential" mode... There is no precise explanation in the Trilian user manual btw), and it seems a tad better, but not there yet IMO :



Here is how it sounds like now (to sum up : maybe a tad better (placebo effect ?), not worse i think, but not quite there yet) :

http://dl.dropbox.com/u/15008135/Rea...y_RandomRR.mp3

I think for now my Trilian settings are fine and it comes down to improving the MIDI programming, but i'm still clueless and need some help.

Couldn't find a version of that part/song ("I Cum Blood" by Cannibal Corpse) played by Alex Webster himself with the bass loud enough, so here is an example played by another dude who sounds a bit like Webster (same kind of finger playing), for reference :

You Tube

I like this video cuz' the tremolo is real but sounds weird/a bit annoying somehow, so i guess this weird/annoying character is not due to MIDI bass programming alone

EDIT 2 : Actually i just noticed that this guy is playing series of triplets (using his first three fingers : 1/2/3 1/2/3 1/2/3...) with the last note ringing a bit more than just a 1/3th note and not real tremolo, but i'm not sure if that's the way Webster does it, and if that's how finger tremolos are played, or if there are "real" finger tremolos on bass.

Posted a comment on this guy's video, i hope he'll answer back :

"Little question about your technique while playing the first riffs : are you playing "real" tremolo as in "using your four fingers in a 1/2/3/4 1/2/3/4 1/2/3/4 fashion, with each finger playing a quarter note (16th i mean) ? Or are you playing kind of "triplets" with three fingers (1/2/3 1/2/3 1/2/3) and leaving the last note of each series of 3 ring a little longer ?

Or maybe if you're doing real tremolo using the four fingers, the 4th hit of each series (which i guess is played with the pinkie) is softer sounding the first 3 hits ?

I'm trying to program that bass part in MIDI and have it decent and realistic sounding so that's why i'm asking those weird questions (i sent you a PM with an audio example btw)"

If someone has a solo-ed tremolo bass part i'm very interested to hear it for reference.

EDIT 3 : Listening to some solo-ed exported bass tracks i made on another song and realizing there are drop-outs, which seems like the VSTi can't reach all samples properly while exporting in speed time (=faster than normal 1x speed) offline, so i'm thinking it might also cause the tremolo part (which should resort to a lot of different samples/Round Robins being triggered) to fuck up...

I'm gonna do some "proper" exports (1x speed and/or online) and see if it makes any difference.

[abt]

I'd like to know this as well. Looking at the video I'm not sure that the fingering is 1,2,3 but probably 1,3,2,1,3,2. I have know idea, maybe he'll answer.

From playing classical guitar for a long time we always play tremolo in the order p a m i (thumb, ring, middle, index). This seems to be the most natural way the fingers tend to move. We're always working on the "a" finger because it is the weakest, getting that stronger makes for a good tremolo sound. If he is playing this way then he would start on the index finger to compensate for not using the thumb. I terms of midi velocity the index finger is the strongest then the ring finger is played, it's the weakest, then the middle finger not as strong as the index.

Having said all that I'm not sure that Trillian can do it. IMO it's probably more to do with the length of the sample, how fast the picking is, the length of the pick attack, picking a string while it's already in motion, where the finger is on the string. These all must have an impact.

[Backe]

sounds like someone's close miking a fart with a 57, so you're perfectly on the money i would say! cause that's what would happen with a real bass too

[Spectivum]

Originally Posted by Djabthrash

not sure if that's the way Webster does it

from memories at gigs he seemed to use all 4 fingers at fast parts

[abt]

Originally Posted by Backe

sounds like someone's close miking a fart with a 57, so you're perfectly on the money i would say! cause that's what would happen with a real bass too

[Djabthrash]

Originally Posted by abt

I'd like to know this as well. Looking at the video I'm not sure that the fingering is 1,2,3 but probably 1,3,2,1,3,2. I have know idea, maybe he'll answer.

From playing classical guitar for a long time we always play tremolo in the order p a m i (thumb, ring, middle, index). This seems to be the most natural way the fingers tend to move. We're always working on the "a" finger because it is the weakest, getting that stronger makes for a good tremolo sound. If he is playing this way then he would start on the index finger to compensate for not using the thumb. I terms of midi velocity the index finger is the strongest then the ring finger is played, it's the weakest, then the middle finger not as strong as the index.

Having said all that I'm not sure that Trillian can do it. IMO it's probably more to do with the length of the sample, how fast the picking is, the length of the pick attack, picking a string while it's already in motion, where the finger is on the string. These all must have an impact.

Wow thanks for that insight ! That's exactly the kind of information i was looking for.

I had no idea classical guitar players were using their thumb for tremolo btw...

The first part of your post can be translated into MIDI programming, while the second is more a matter of soundsource/samples being played (and note length also), so i guess i'm gonna improve the MIDI programming and then compare different soundsources ("normal" (full range, including sustain) / staccato / true staccato).

Anybody has an idea of what would be the difference between staccato and true staccato in Trilian ? No explanation on this in the user manual AFAIK.

[Djabthrash]

Originally Posted by Backe

sounds like someone's close miking a fart with a 57, so you're perfectly on the money i would say! cause that's what would happen with a real bass too

How serious are you about this (the "it sounds like the real deal" part i mean, not the fart joke ) ?

[Djabthrash]

Originally Posted by Spectivum

from memories at gigs he seemed to use all 4 fingers at fast parts

Ok that's what i was thinking...

Now i need to know if he's using his fingers like this 1/2/3/4 (1 being index, 2 being middle finger, 3 being ring, 4 being pinkie) or in a different manner like the classical guitar players do for instance.

I also need to know the power of each finger to translate it into velocities.
If it's 1/2/3/4 i guess it would translate into something like :
100 / 82 / 79 / 70 (i assume 1 is way stronger and triggering the note so is played harder anyway, and then the following fingers get weaker and weaker)

[Djabthrash]

Listening to some solo-ed exported bass tracks i made on another song and realizing there are drop-outs, which seems like the VSTi can't reach all samples properly while exporting in speed time (=faster than normal 1x speed) offline, so i'm thinking it might also cause the tremolo part (which should resort to a lot of different samples/Round Robins being triggered) to fuck up...

I'm gonna do some "proper" exports (1x speed and/or online) and see if it makes any difference.

[Djabthrash]

Originally Posted by abt

picking a string while it's already in motion

This is key also IMO.

IIRR there is a mode in Trilian that gives that feel, i'm gonna look it up.

[Djabthrash]

Found this thread where various bass players expose their various techniques for tremolo. They don't use the same number of fingers/fingers/finger sequences :

http://www.metalguitarist.org/forum/...-question.html

I think what makes sense to me right now is : (those are hypothetical, not golden rules)

-with this kind of hard/rattly finger style, when you play tremolo i guess you play quite softer/rounder than when playing normal speed notes.
-> less velocity (to reach that softer/rounder tone) during tremolo parts than during normal parts

-you use some kind of accent on the first finger/note of each sequence of four 16th notes
-> more velocity on the first note of each sequence of four 16th notes

-you have some weaker fingers (i.e you can't play as hard with them as with your stronger (often leading) finger(s))) and stronger fingers
-> write an average drop of velocity for 1 or 2 fingers of each sequence (and they can be at exact same post/note from one sequence to another, providing your sequences are 4 notes sequences (ex: 1/4/3/2 1/4/3/2 1/4/3/2) and not longer (ex: 1/4/3/2 1/2/3/4 1/4/3/2 1/2/3/4))

-when moving your fretting hand to another fret/string, you change your velocity a bit as well (like you would do when playing regular speed single notes while changing the fret each time (which makes one single note for each fret/time))
-> write a slight change of velocity for the whole sequence from one sequence to another, when you're switching to a new fret

-do some slight random velocity humanization on the whole tremolo part, and of course don't use the same random humanization "seed" (pattern ?) for the whole part if that part is very long, better split the part in sub-parts and use one different random humanization "seed" for each sub-part (keeping almost identical parameters though : velocity % / timing %)

As a whole, on a medium-length finger tremolo part, I'd write it the following way :

-write the part with 4th notes (not tremolo) with the velocity values / changes (for humanization) you'd have for a normal part (a slow part that is played quite hard to get that rattly/hard sound)
-drop the velocity of the whole part to have it sound softer/rounder, resulting in having the "leading" notes (i.e the first note of each sequence) written only
-add the other fingers on the first sequence, at a lower velocity than your leading finger, and then change up the velocities in between those fingers, considering some will be identified as "weak fingers" and will have a real velocity drop
-copy paste those remaining/following fingers (i.e all the fingers of each sequence, except for the leading finger) to the next sequences, adjust their piano notes if needed
-select all the remaining fingers of the whole tremolo part, and slightly random humanize them (using the same seed)

[Djabthrash]

Found this big bass guitar forum, gonna do some research and maybe ask there :

http://www.talkbass.com/forum/f45/

http://www.talkbass.com/forum/f21/de...-metal-378173/

[Spectivum]

Originally Posted by Djabthrash

Ok that's what i was thinking...

Now i need to know if he's using his fingers like this 1/2/3/4 (1 being index, 2 being middle finger, 3 being ring, 4 being pinkie) or in a different manner like the classical guitar players do for instance.

I also need to know the power of each finger to translate it into velocities.
If it's 1/2/3/4 i guess it would translate into something like :
100 / 82 / 79 / 70 (i assume 1 is way stronger and triggering the note so is played harder anyway, and then the following fingers get weaker and weaker)

the other way round, start with pinky and go in order to index. just try tapping on your desk and you will see it is much easier

[abt]

Originally Posted by Djabthrash

Ok that's what i was thinking...

Now i need to know if he's using his fingers like this 1/2/3/4 (1 being index, 2 being middle finger, 3 being ring, 4 being pinkie) or in a different manner like the classical guitar players do for instance.

I also need to know the power of each finger to translate it into velocities.
If it's 1/2/3/4 i guess it would translate into something like :
100 / 82 / 79 / 70 (i assume 1 is way stronger and triggering the note so is played harder anyway, and then the following fingers get weaker and weaker)

He could use the pinky but in my experience it's not that common, they tend to use 1-2-1-2 or 1-3-2. I'm not sure how many bass players use the thumb as they tend to rest it somewhere. The pinky is pretty useless in comparison to the other fingers. Also it's length means that in order to have it even reach the strings you have to change the hand position to one that's not very conducive to good tone.

Originally Posted by Spectivum

the other way round, start with pinky and go in order to index. just try tapping on your desk and you will see it is much easier

Yes, this order is correct but you don't lead off with the weakest finger because you want the strength on the beat. so instead of say 3-2-1-3-2-1 you'd start with the first finger 1-3-2-1-3-2 once you get going it feels the same but sounds different.

[jarenare]

The problem that I find every time I try programing fast picking bass, are the lack of multiple samples on the libraries. 3 or 4 samples of every note isn't enough for these parts. If you use round robin, the ear will note the repeating cycle of samples, and using random will cause that sometimes the same sample will be played 2, 3 or 4 times in a row...

[Djabthrash]

Originally Posted by abt

He could use the pinky but in my experience it's not that common, they tend to use 1-2-1-2 or 1-3-2. I'm not sure how many bass players use the thumb as they tend to rest it somewhere. The pinky is pretty useless in comparison to the other fingers. Also it's length means that in order to have it even reach the strings you have to change the hand position to one that's not very conducive to good tone.



Yes, this order is correct but you don't lead off with the weakest finger because you want the strength on the beat. so instead of say 3-2-1-3-2-1 you'd start with the first finger 1-3-2-1-3-2 once you get going it feels the same but sounds different.

Thanks a lot for the info dude ! This alongside the following thread is helping me a lot when it comes to understand what's going on during those tremolos :

http://www.talkbass.com/forum/f21/de...-metal-378173/

[Djabthrash]

Originally Posted by jarenare

The problem that I find every time I try programing fast picking bass, are the lack of multiple samples on the libraries. 3 or 4 samples of every note isn't enough for these parts. If you use round robin, the ear will note the repeating cycle of samples, and using random will cause that sometimes the same sample will be played 2, 3 or 4 times in a row...

So do you imply that "random" would use less samples for the same note/velocity than the "sequential" mode ? What about the "random full cycle" mode ?

Round robin mode aside, i guess humanizing the MIDI part manually to have slightly different velocities for each note helps to get rid of that weird robotic feeling.

[smy1]

I actually agree with the "mic a fart with an sm57 comment". Both the bass in your programmed examples and in the "I cum blood" video sound horrible.

I'm aware that it is a stylistic device and that people actually play like this, but it really doesn't work for me. You'd generate much more low end power/tightness by not playing tremolo/highspeed bass. But obviously that's not what you are going for

With regards to your question: I have no answer other than that your examples will sound indistinguishable from real bass in a full mix.

[Djabthrash]

Originally Posted by smy1

I actually agree with the "mic a fart with an sm57 comment". Both the bass in your programmed examples and in the "I cum blood" video sound horrible.

I'm aware that it is a stylistic device and that people actually play like this, but it really doesn't work for me. You'd generate much more low end power/tightness by not playing tremolo/highspeed bass. But obviously that's not what you are going for

With regards to your question: I have no answer other than that your examples will sound indistinguishable from real bass in a full mix.

I agree that there are some parts where playing slower would benefit, but there are some parts where tremolo fits IMO and i like how it sounds in Cannibal Corpse and some other death metal/thrash bands, so for now my goal is to have it sound like the real stuff, as annoying as the real stuff can sound to some people
I'd never put tremolo bass all over the place myself, and can't think of bands doing it actually...

About your answer to my question : i think people who are used to that kind of bass parts and don't mind the real deal can still tell and find the programmed part a bit annoying (me included), even in a full-mix (not to mention that there can be some parts where you can hear the bass more than say during a "full-on assault" part where the bass would be more buried and undistinguishable).

[Djabthrash]

Originally Posted by smy1

With regards to your question: I have no answer other than that your examples will sound indistinguishable from real bass in a full mix.

Another example of mine with programmed tremolo finger bass, in a busy full-mix this time :

http://dl.dropbox.com/u/15008135/oth...h_20120227.mp3

From 0"13 to 0"22 (and 04"00 to 04"07 also) there is a part with 16th notes tremolo bass and guitar at the same time, does this part sounds shocking/odd (besides the shitty guitar playing i mean ) to you guys ?

Also from 3"22 to 3"43 there are some short bursts of synced (well supposed to but i fucked up the guitar part big time ) machine gun bass/1 guitar/kick drum.

I wanna know if it sounds any different/worse than a real bass, or if it sounds like the real deal.

[Djabthrash]

Also found this interesting post (in this very interesting thread mentioned before : http://www.talkbass.com/forum/f21/de...-metal-378173/) that supports some stuff some of you guys previously said :

"Old 11-23-2007, 02:23 PM
LUMPYKILLS LUMPYKILLS is offline

the secret is to accent the first note heavily. start slow then get faster.

so if lets say you play ring, middle, index, or PMI,

play

P M I P M I P M I P M I P

the triplet sound comes when you play P M I P M I P M I

slow down and get the accents right, then it'll fall into place and work even when you get to mind blistering speeds"

Note that this is a series of 16th notes using 3 fingers played in the same order, so the accent is switching from one finger to the next when switching from one sequence of 4 16th notes to the next.

[jarenare]

Originally Posted by Djabthrash

So do you imply that "random" would use less samples for the same note/velocity than the "sequential" mode ? What about the "random full cycle" mode ?

I guess the number of samples used are the same, no matter what mode are used. But the ear are very sensitive detecting this (at least mine )

If random full cycle works like I think, you can still have the same sample two times in a row. For example, if we use 4 samples, we could get something like [1,4,2,3][3,4,2,1].

[Leon--]

No irony but with the statement "I'm not gonna buy a bass and i want to get those fast parts to sound decently realistic." the answer is already given,there's no way to make it sound as a real bass
"people know how to program MIDI drums to sound realistic even for extreme parts, so i guess there should be a way to do the same with fast tremolo bass parts."
That's because everyone is using triggers and the audience is used to that sound,so people add the same samples both on a real drummer and a drum machine.Also many extreme metal recordings are recording with an E-kit
So the "fake" cymbals are common in the audience.
If you want a r e a l bass guitar sound that will make your mix sound huge then a real bass guitar is needed.
Cheers

[Djabthrash]

This one is a good real bass tremolo playing example :

You Tube