i2M for violin

[strashilol]

I'm noticing that some notes just aren't being played through the i2m. For example, when I play an arpeggio over the first five strings on the guitar notes toward the end of the arpeggio are not recognized. On violin I'm noticing the same thing when I play at higher speeds. If I play at a slow speed it seems to recognize things ok. I've also noticed that the i2m won't play exactly what I'm playing - sharp or flat notes.

I'm gonna bet 20 bucks (euros, actually ) that you have different pitch bend settings in the i2m editor and the MIDI sampler you're using.

Having been developed for keyboards, MIDI format is essentially staccato-oriented. You press a key, and a MIDI note is created, as simple as that. Guitar, wind and bowed instruments are, however, capable of producing true legato, and MIDI conversion hardware/software tries to emulate it by using pitch bends.

What happens when you're bowing is as follows. Once your bow touches the string, a corresponding MIDI note is created; then, while you move your left hand along the string, no new notes are "issued", but instead the original one is being "altered" with pitch bend events. The same thing would happen on a synth if you pressed the key corresponding to the initial note of a progression and then "played" all of the successive notes not by hitting their respective keys, but by manipulating the pitch bend wheel.

The problem is, pitch bend events are "relative", i.e. they depend on range settings.

So what you should do is make sure that pitch bend range is the same in the sampler you're using and the i2m editor. I'm not sure about Komplete 7, since i can only afford a free Kontakt Player , but for i2m you can use the clock-like thingie in the "pitch mode" column.

In case Komplete has the same interface as Kontakt:
- open the instrument you're using in edit mode;
- locate the "Source" section;
- click the "Mod" button there to expand the list of modulators;
- the horizontal modulation intensity slider on the pitchbend->pitch modulator controls the pitch bend range.

Another thing you should do is set up your sampler to recognise i2m's breath controller events, otherwise your violin will sound very flat. In "guitar" and "wind" modes (if you've installed the latest firmware that is, otherwise only in the "wind" mode) i2m issues breath controller events (CC2, control change 2 = breath control) when the volume of your playing changes; however, your sampler doesn't know about it yet. I don't have Kontakt installed right now, but the procedure mirrors the aforementioned one: open the instrument in edit mode, go to the "Volume" section (i think... or is it called "Amplifier"?) and set CC2 events as the source of volume modulation.

Hope this helps.

[tklenke]

I checked the settings and adjusted them as you suggested and am still having the same problems. I've played around with the settings a lot and haven't had any luck. Thanks for the suggestions though.

James, any word from support about the file I sent in?

[strashilol]

I checked the settings and adjusted them as you suggested and am still having the same problems. I've played around with the settings a lot and haven't had any luck. Thanks for the suggestions though.

Please confirm that:
1) you're having the same problems when utilizing your i2m in chromatic mode;
2) there are no such problems when playing a synthesizer through Komplete (in case you have a synth of course and thus can afford such a test).

[james]

We've had a look at the violin samples that tklenke sent us...

The missing notes (i.e., notes just not being played by the i2M) is a side-effect of having to be robust against other cases where we don't want to incorrectly detect new notes that don't exist.

For example, on a bass guitar a long sustained note can evolve over a long period of time with different overtones appearing and disappearing and we don't want to jump between octaves while this happens.

Similarly, when two notes are playing together, the beating can cause fast amplitude variations which must be ignored to avoid repeated triggering on new notes which don't exist.

In the example recording, both of these things are present, and sensibly so because it's a violin. The transitions between notes are not obvious and each note blends slowly into the other (much like the example above with the evolving bass tone). But it is very difficult to distinguish between when this is real (e.g., a violin) or not real (see examples above). Moreover, because the i2M is designed for ultra-low latency we can't simply wait for 50ms or more to see what happens, we need to react faster than this. So note changes such as this can be detected, but only if the change between notes happens quickly (we specifically ignore things that happen slowly since it's less often "real").

Another thing is that when a note is missed, it is because some remnant of the previous note is still present. The i2M is designed to latch onto a note until it fully disappears (so making it robust against accidental noise, and to allow notes to sustain for a very long time). But when there is some energy from the original note still present, it will try to ignore new notes unless they are "obviously" new. I'm guessing there is some resonance going on where the old note is still ringing. You get the same thing on guitar, where carefully damping all non-playing strings ensures there are no remnants left as you switch notes, particularly as you play across the strings. Note that such remnants are not a problem if the new note is cleanly detected by an obvious transient (i.e., a pluck) in the signal.

I think the solution is to ensure there is a more obvious transition between notes, and there are three aspects to this (ensuring any one of these should be enough; ideally all three should occur):

(1) Play in a staccato fashion so there is an obvious "off" point between notes (i.e., an audible gap).

(2) Have a more obvious transient when a new note starts -- maybe not possible on a bowed instrument. Obviously simpler on a guitar that is plucked. However, if the "off" mentioned in (1) is done, the new bowed note will still look like a transient and you get the benefit of this (faster, more reliable detection).

(3) Ensure the last note is fully damped before playing the new note. This allows the new note to be tracked even if there is no visible transition (i.e., in the envelope) between the notes.

[VmusicV]

I'm a violin player interested in this as well..... I know the old Zeta midi violin was very "touchy", every note had to be exact and perfect for it to work. The slightest bit of bow noise, wolf tone, or overtone messes up the analog to midi

Why select the i2M over the G2M? the range of notes is similar -- of course the violin is bowed not picked/plucked

I would hapily forego double stops (2 notes at once) - in exchange for single notes that really worked

Thanks
Alex

[james]

Why select the i2M over the G2M? the range of notes is similar -- of course the violin is bowed not picked/plucked

The i2M has more configuration options than the G2M. In particular, you can set different pitch-bend ranges on the i2M, whereas the G2M is fixed at +/-2 semitones (to match the most common guitar bend range). The i2M can also be firmware upgraded, so it gives some future-proofness if we add anything specific to help with violin MIDI conversion.

If using hardware MIDI gear, the G2M may work better though, and can sometimes feel faster since there is no computer involved.

[PaulK]

I just wanted to inform violinists into electronica, that there is an interesting bow out, the k-bow. See it here; http://www.keithmcmillen.com/k-bow/overview