You are here

Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

For all things relating to guitars, basses, amps, pedals & accessories.

Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Jay Menon » Sun Apr 12, 2020 4:16 pm

I was under the impression that (unlike the old Axon) the Roland Guitar Synths (GR30, G33 etc) have to listen to one cycle of a note before recognising its pitch - and therefore the lower notes have an integral 'delay' to them before 'sounding'.

While trying to play piano sounds with my old Godin ACS I found this rather distracting (perhaps no more so than my own sloppy playing).

Now everyone says the Fishman Triple Play tracks much faster. How can this be? I thought the delay happens at the guitar synth end, rather than at the guitar pickup.

So apart from the fact that it's wireless, are there any other benefits to the FTP as compared with a GK3 or the piezo saddles that come on a Godin type guitar?
Jay Menon
Frequent Poster
Posts: 845
Joined: Sun Sep 30, 2007 12:00 am
Location: Norfolk

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Sam Spoons » Sun Apr 12, 2020 11:13 pm

I seem to remember reading that it was only ¼ of a cycle* but whatever it is all notes will have a delay/latency and the lower the note the longer it will be. Roland VG and SY devices avoid this by passing the audio (with some processing) to fill in the gap between the note being played and the synth's pitch engine taking over. Maybe the Fishman uses a similar trick?

* But I guess that is plus processing latency.....
User avatar
Sam Spoons
Jedi Poster
Posts: 13365
Joined: Thu Jan 23, 2003 1:00 am
Location: Manchester UK
Finally taking this recording lark seriously (and recording my Gypsy Jazz CD)........

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Wonks » Mon Apr 13, 2020 10:07 am

That Fishman site is awful. I want to read about what's going on, not look at countless videos. And the text repeats the same info on numerous pages. I really can't be bothered with it.

I know I'm starting to sound like Arpangel here, but it doesn't inspire me with any confidence in the product.
User avatar
Wonks
Jedi Poster
Posts: 10889
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby adrian_k » Mon Apr 13, 2020 2:10 pm

+1
Too many sites now are heavy on media and light on things like specs.
User avatar
adrian_k
Frequent Poster
Posts: 1029
Joined: Thu Jan 30, 2003 1:00 am
Location: Gloucestershire
getting better all the time..

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby MOF » Mon Apr 13, 2020 3:43 pm

I seem to remember reading that it was only ¼ of a cycle

I would have thought it can only be as little as half a cycle since the negative part of the waveform will be identical but out of phase with the positive side.
A guitarist friend of mine is interested in this kind of technology and is convinced just from watching the videos. After all if you see someone riffing on the lower strings and hearing a synth sound in synch’ with the playing then what’s not to like and trust you’ll have similar results.
I think it’s unlikely you’ll be given sufficient technical information because it’s almost certainly part of their intellectual property.
MOF
Frequent Poster
Posts: 1201
Joined: Thu Mar 06, 2003 1:00 am
Location: United Kingdom

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Sam Spoons » Mon Apr 13, 2020 4:57 pm

You could reconstruct a symmetrical wave from ¼ cycle, I think that was the theory behind it. However you do it (and by the time you've added in some processing latency) the low notes are going to sound laggy.
User avatar
Sam Spoons
Jedi Poster
Posts: 13365
Joined: Thu Jan 23, 2003 1:00 am
Location: Manchester UK
Finally taking this recording lark seriously (and recording my Gypsy Jazz CD)........

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Jay Menon » Mon Apr 13, 2020 5:16 pm

Low E string on a guitar: 82Hz
Or 82 cycles / second
So for low E One cycle: 1/82 of a second

1/4 cycle: 1/328 of a second

From my memory of when I had my Godin ACS and GR20 there was a surely lot more delay than that on the low notes...
Jay Menon
Frequent Poster
Posts: 845
Joined: Sun Sep 30, 2007 12:00 am
Location: Norfolk

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Wonks » Mon Apr 13, 2020 6:13 pm

You'd need more than 1/4 cycle, as most guitar waveforms are asymmetrical, so the peak isn't right in the middle as with a sine wave. 1/4 would work with a plain sine wave, but guitar signals aren't simple sine waves.

It's not made easier to detect fundamental frequencies with the GK -style pickups as they are placed right near the bridge, where the fundamentals have almost no amplitude and the signal is almost all upper harmonics. This gives a very odd output compared to the 'normal' one from the pickups and it's easy to pick up a higher frequency than the actual one unless you wait for a full cycle and use amplitude to determine the real peaks (as opposed to the intermediate peaks).

I used this simple harmonic waveform generator (https://meettechniek.info/additional/additive-synthesis.html) for an example showing how lack of a strong fundamental and high amplitude harmonics distorts the waveform (harmonic levels above the fundamental are purely random and not supposed to be an actual example of a GK pickup output).

Image
If you went from the largest +ve peak to the largest -ve peak, you'd definitely get the wrong frequency. And if you went from the first positive peak to the first minimum point, you'd also get the wrong frequency.

Obviously you can use hardware low-pass filters to remove some of the upper harmonics to make life easier, but the low string response will still be mainly driven by the need to wait to get workable information, which can be a complete cycle.
User avatar
Wonks
Jedi Poster
Posts: 10889
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Sam Spoons » Mon Apr 13, 2020 6:15 pm

This is interesting, https://www.joness.com/gr300/MIDI_SPEED.htm it suggests that the fastest synth is the Roland GR 300 which had a delay on the low E of 5.8 ms which is less than half a cycle (a full cycle of the low E is 12.2 ms). But bear in mind it is not a digital synth.
User avatar
Sam Spoons
Jedi Poster
Posts: 13365
Joined: Thu Jan 23, 2003 1:00 am
Location: Manchester UK
Finally taking this recording lark seriously (and recording my Gypsy Jazz CD)........

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Wonks » Mon Apr 13, 2020 7:02 pm

The GR-300 doesn't do MIDI and is purely analogue. So it should be near-instantaneous. Any delay must be down to the trigger for the envelope filters.
User avatar
Wonks
Jedi Poster
Posts: 10889
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby arkieboy » Mon Apr 13, 2020 7:56 pm

(The hex pickups are placed close to the bridge to avoid crosstalk, which turns out to be a bigger problem than not having as much clean fundamental - the algorithms implemented in music technology products don't normally have the kind of computing grunt to match the pitch extraction capabilities of melodyne so they need a single pitch to work).

Roland pitch-to-midi is very much vanilla AFAIR - there's not been much advance for a while - although my SY1000 is noticeably more stable and consistent than my VG99. Received wisdom is that having the internal conversion inside the box with the sound generation unit means that there is some secret sauce going on in the GR units from the GR50 onwards, possibly triggering atonal attack portions before the pitched portion of the note. But I haven't trawled the Roland patents so this is only hearsay.

Andreas Seleazy (sp?) is the engineer behind the old Axon and the modern Triple Play units. He's developed some (supposedly) AI algorithms to extract the pitch more quickly than the standard Roland algorithms - you'd assume he's using some of the higher frequency components to guess the fundamental. The Axon AX100MkII and the subsequent Triple Play are the fastest hardware units for pitch to midi.

And of course Wonks is completely correct with the GR300 which uses 'phase locked loops' (whatever they are) to sync the oscillator frequency with that of the guitar string.

The VG series and the latest SY units from Roland - and almost certainly the EHX units too - use an alternative approach, which is to restructure the harmonic content of the guitar string to get the sound you want. The SY1000 can render pretty much all of the basic waveforms, including an already detuned sawtooth and supersaw waveform -directly from a guitar string to pretty high fidelity and with practically instantaneous response. The combination of this, the internal virtual analogue for when you must have precise control over the envelope, and midi output via USB to run MainStage is for me a killer combination - my 12u guitar synth rack with my Axon, Supernova, EMU5000 and JV880 doesn't hold a light to it.

Finally Roland, EHX and the like are also playing with real-time polyphonic pitch extraction with implementations in the SY1 and SY300 - just plug a guitar into the jack input and it can 'resynthesise' 2 or 3 strings worth of audio at once, but not to the clarity of the SY1000.

Poly pitch detection is also implemented in Jam Origin's Midi Guitar product, which is probably the fastest pitch to midi implementation and works well with regular guitar chords. I have it on good authority that 6 MG instances in MainStage listening to the =individual strings= sent from a SY1000 is really very good indeed and better than the Triple Play. But a MacBook Pro + SY1K + MG license is a good deal more expensive than a triple play unit!
User avatar
arkieboy
Regular
Posts: 441
Joined: Thu Nov 07, 2002 1:00 am
Location: Oxfordish

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby MOF » Mon Apr 13, 2020 10:06 pm

That Fishman site is awful. I want to read about what's going on, not look at countless videos. And the text repeats the same info on numerous pages. I really can't be bothered with it.

I know I'm starting to sound like Arpangel here, but it doesn't inspire me with any confidence in the product.

I think some people are over thinking it, just because you don't know how it works doesn't mean you shouldn't trust it.
I don't understand how a CPU operates so many times per second, but it does, my iMac has a 4.2Ghz CPU, my Atari 1040 CPU operated at a mere 8Mhz, even that blows my mind, just counting up to one million would take many hours, yet in one second it dealt with 8 million instructions (that might not be entirely correct, it might not be one instruction for every CPU cycle but you get my drift).
So all I'd say to you is don't mistrust the Fishman triple play, either be convinced by their videos or go and demo one when you can and find out for yourself.
Incidentally there are software only versions to try before you buy such as https://migic.com/downloads and
https://www.jamorigin.com/products/
MOF
Frequent Poster
Posts: 1201
Joined: Thu Mar 06, 2003 1:00 am
Location: United Kingdom

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Jay Menon » Tue Apr 14, 2020 9:51 am

For me it’s always been about playing piano/keyboard accompaniments to my vocals.

I also wonder if it might be worth changing the fifth and sixth strings to half the gauge and tuning them an octave higher - to reduce latency.
For example
6th String: 024
5th string: 018

Then assign a separate ‘sound’ to those two strings transposed an octave lower...?

Is the GR 55 any faster in terms of its tracking than the GR 33? I’m not in the least interested in the COSM amp modelling parts of the 55...
Jay Menon
Frequent Poster
Posts: 845
Joined: Sun Sep 30, 2007 12:00 am
Location: Norfolk

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Sam Spoons » Tue Apr 14, 2020 10:00 am

Have you tried the Electro Harmonix '9 series' https://www.ehx.com/browse/9-series the emulations are very convincing and latency is negligible to my ears/fingers?
User avatar
Sam Spoons
Jedi Poster
Posts: 13365
Joined: Thu Jan 23, 2003 1:00 am
Location: Manchester UK
Finally taking this recording lark seriously (and recording my Gypsy Jazz CD)........

Re: Fishman Triple Play vs GK3 & RMC Piezo in terms of Note Latency

Postby Jay Menon » Tue Apr 14, 2020 10:09 am

Thanks Sam - Looks interesting, but I’m not seeing any ordinary acoustic piano sounds in their range...?
Jay Menon
Frequent Poster
Posts: 845
Joined: Sun Sep 30, 2007 12:00 am
Location: Norfolk

Next