Polyphonic Guitar to MIDI VST/AU "MIDI Guitar"- BETA TEST

[ariajazz]

Whoops, I didn't notice v0.7.5 is out. I'm downloading it and will be testing it tomorrow morning.

[sheldrakes]

Having the same problem with Sampletank it does not load

Try Starting the Standalone 32bit version of Midiguitar as Administrator (right click on Midiguitar and open as Administrator) then do a clear and a new scan and it should work.

[Maruuk]

The standalone sticks me with a 256 byte buffer and an annoying 5.8ms latency one way--probably double that in practical terms. Did I read somewheres that latency can be reduced when used as a VST w/a DAW?

[Cecelius2]

The standalone sticks me with a 256 byte buffer and an annoying 5.8ms latency one way--probably double that in practical terms. Did I read somewheres that latency can be reduced when used as a VST w/a DAW?

I too am interested in the latency of MG. I run a MOTU 828mkii hybrid and I can get my buffer samples down below the 128 of the MG, with both a 96 and 64 samples, but MG only recognizes or utilizes the 128 as its lowest setting. My latency at 128 is 2.9ms, but if MG would let me use my MOTU at 64, my latency would be reduced. Don't get me wrong, I am very pleased with the fast latency, but if MG can eventually work with a 64 buffer sample setting, that would be even nicer. Currently, if I set my MOTU down to 64 samples, MG comes up with a blank in the Audio Device buffer area (if I recall).

So, to the Developer I have a question: Is there a way that MG will eventually be able to use the faster buffer sample settings on some of the newer faster audio cards like RME and even my MOTU? Not complaining, just inquiring.

[nix808]

I've spent hours trying to convert guitar audio to MIDI,
currently I run an auto-tracking filter in a hexaphonic pickup-
just to let u know where I'm coming from.
The transient on guitar is not a steady tone.
It's like a percussive burst,
so any soft that uses real guitar strings has a limit on how minimal u can be with latency.
My 2c
IMO we can have usable latencies though

[Maruuk]

I understand there are limits, just that we need a dynamic partnership with the manufacturer to push those limits to the absolute minimum for a given set of hardware. Not some shelving deal set at an arbitrary number.

[nix808]

I can't exactly speak for JO,
but I think there is no reason the users can't dice with wrong notes versus minimal latency. In my experimental system, I can set lower latencies, but the pitch detection is less sure. So- they can decide if it's failsafe, or if some menu for the advanced can let you set latency in samples, thus allowing the user to eke out a couple ms.

I agree man, it would be cool if we could set it the point of wrong notes, and bring it in a fraction.

[Maruuk]

Yeah, for me it's unplayable at 256 samples/5.8ms which probably translates to 12ms or more round trip. Simply not musical, you lose all the feel. Somewhere between 64 and 128 would probably be acceptable.

[wasi]

Yeah, for me it's unplayable at 256 samples/5.8ms which probably translates to 12ms or more round trip. Simply not musical, you lose all the feel. Somewhere between 64 and 128 would probably be acceptable.

12ms latency is the equivalent of standing 4 meters from your speaker while you play. How is that not musical?

Sometimes it helps to put things into perspective. Anything up to about 15ms (5m) is what you'd often be dealing with on an open-air type stage. For anything significantly less than 6ms (2m) you'd probably have to bend over while playing (unless your speaker is on a stand) and you'd be missing part of the tonal spectrum to directivity.

Assuming you stand about 3m from your speaker cab, your real world playing 'latency' is about 9ms. At sea level, that is. You can lower latency by playing in a high pressure chamber or hot temperatures. The Himalayas have terrible latency.

[chucky5p]

Yeah, for me it's unplayable at 256 samples/5.8ms which probably translates to 12ms or more round trip. Simply not musical, you lose all the feel. Somewhere between 64 and 128 would probably be acceptable.

12ms latency is the equivalent of standing 4 meters from your speaker while you play. How is that not musical?

Sometimes it helps to put things into perspective. Anything up to about 15ms (5m) is what you'd often be dealing with on an open-air type stage. For anything significantly less than 6ms (2m) you'd probably have to bend over while playing (unless your speaker is on a stand) and you'd be missing part of the tonal spectrum to directivity.

Assuming you stand about 3m from your speaker cab, your real world playing 'latency' is about 9ms. At sea level, that is. You can lower latency by playing in a high pressure chamber or hot temperatures. The Himalayas have terrible latency.

What you're saying is all very true. However, for those who want to use MG in a live situation where they're already 3M/9ms away from their speaker cab, they don't want to have another extra 20~30ms added to that (from the audio-to-MIDI conversion). The total amount of latency then becomes a serious issue.

That's why any guitar-to-MIDI converter should have the smallest latency possible (with all other things being equal-i.e. tracking precision). I too hope that JO will find a way for us with low latency audio interface to use it to its full capabilities. In my particular case, my FastTrackUltra8R can do 64samples@96Khz, or for a total round trip latency of 3ms, and that's much better than the 8.4ms I'm "forced" to use because of this 128samples/44.1Khz limitation of current versions of MG.

Chuck

[bupper]

all this criticism for a little latency I believe is grossly exagerated. I must say that I'm nothing to do with jamorigin but a user as are you guys. This is a totaly software method of translating audio into midi in realtime, now obviously the MUST be a bit of latency in order for the calculations to take place. I too am a gigging musician & I feel your pain in having latency added to the already (dunno how many metres you are from the cabs) small latency from the cabs, lets face it, most musicians are rarely more than 2 metres away from their monitors while playing live, but the simple answer would be to use an ear monitor!!! problem solved. you would have the same latency as if you are standing 3 or 4 metres from the cab.

Now 96kHz would be nice as it would decrease the latency by half (while increasing the cpu load) but I believe that JamOrigin have created a wonderful product that is progressing all the time & is still not at even the v1.0, give it time to work

[Cecelius2]

Let me be clear. I am very pleased with MG's latency. I have 2.9ms as reported by MG's buffer size/latency window. I can easily play with it in my studio as it is right now. I also own two Axon's and a Roland; MG has replaced them right now. Neither my Axons nor my Roland track soft synths very well; the Axon does of course trigger its internal sound a little faster, but not external soft synths. MG is now my choice for triggering soft synths. Granted I have played midi guitar for two decades, have developed a very clean and pianoesque style of playing, and don't expect any midi guitar converter to track exactly like a midi keyboard; now that would be great, but they just don't --not yet.

My post about latency above was just an honest inquiry because I noticed that MG would not recognize my MOTU's 96 and 64 sample buffer settings. If JO stays just the way it is right now with 128 buffer as it's lowest setting and with 2.9ms on my setup, I am fine, happy, and exceptionally pleased with the latency of MG.

I am sure there is still some fine tuning to do escpecially with 6 note or full chords, but for $99 MG is a great value and has made midi guitar available to the masses. Thank you, thank you, thank you Jam Origin for this wonderful addition to the competition. Keep improving, keep fine tuning, and if you happen to get MG eventually to recognize 96 or 64 buffer samples that would be a bonus, but 128 is working very well for me.

[Maruuk]

Well then it simply ISN'T 12ms, the number in the window is always far less than the ACTUAL round trip latency. Whatever it is in the standalone (at the minimum 256) it's unplayable. Hopefully in SO2P I can get it way down. I have good hardware so there's reason for optimism.

[Guges]

Bottom line is, for real-time playability, it must be at least close to dedicated hardware with a hex pick-up. Years ago I had a Roland GR-1. Triggering was dependent upon the release, it was very responsive, and even handled pretty quick double-picking. That was twenty years ago, so MG must about match that, if it can't do better.

Sitting in front of my studio monitors, the triggered sound is a bit after the straight guitar sound. Now it could be my interface (Alesis USB mixer), or/and my VST (PianoOne) via MG, I don't know.

[CurryPaste]

MG already does latency at the level of hardware and puts out way less rubbish than the Roland GR55 with hex. MG is much easier to setup too. Using 44Khz with 128 buffers here in the standalone and its AWESOME even at fast playing.

Btw. this was recorded in version 0.5 where buffers were stuck at 256:


Thanks, happy customer here.