Polyphonic MIDI?!

[BertAnt]

When two tones are close together in frequency, they anyway can be easily distinguished without a PC. I saw it was done by a boy about 40 years ago... Of course his device had ZERO latency and worked accurately... It's the matter of the algorithm, not physics - just make it VERY simple... (to eliminate latency).

In the most expensive scenario you can hire, for example, a programmer who works/worked with EEG (brain rhythms), ECG (heart rhythm) and the like. Accurate-zero-latency-sound-recognition (making MIDI files) is not a hard task as it looks like. If you cannot resolve the problem the way you know - try another approach.

It doesn't matter if a programmer is a musician or not - all you need - give him a properly formulated task, which he could understand... Of course, you need to find a creative one... Why am I so sure? I worked on the similar project - it's all about the algorithm... Some programmers are not able to find a simple solution when they analyze complicated wave-forms and due to this they offer bulky unreliable solutions, others - can find the key-pattern and give you exactly what you need.

The sounds, which are between..., also can be recognized as a note (to the closest one) or, for example, as the guitar glide (depends on the settings). I just don't understand - why this is a problem...

Then try to understand it to see how difficult it is.

[globalisley]

I suppose arrogance comes with the territory when you deal with engineers and programmers. I’m guessing our computer whiz doesn’t have to play country style double stops with quarter bends at break neck speeds. If you can’t get the latency down to the same way my brain would process it then it won’t work so I certainly see the problem. If this problem was so easily solved the midi guitar wouldn’t be a dinosaur.
Enough of that. I'm actually writing because I’d like to be able to able to use transcription software to create lessons for students. I would think latency is less of an issue in this case. If I had to play slowly and stop every two measures it would still be faster than typing in G13 by hand so if you’re working on a polyphonic version I’m your first customer.

[joenoone]

I have a GI-20 for use with my 13 pin synth ready instruments- this take care of my needs for poly pitch to midi in the studio. However, I would love a mono in, polyphonic Sonuus G2M product. It would be portable (hopefully, still battery operated), and I could use any of my electrics to drive it. With a suitable laptop/netbook and a couple of vst's, I could play anywhere!

The GI-20 requires a GK style pickup (see: http://johnp.net/projects/guitar-synth/ ... nouts.html) which has a separate pickup for each string in it (and NO MIDI interface) which feeds 6 separate audio signals into the GI-20 or other GK compatable device which has 6 pitch-to-MIDI converters built in. The Sonuus device only has one pitch-to-midi converter and uses the output of the instrument's pickup.

No company that I've seen has built a MIDI device that can take an input signal of multiple frequencies and compute what are the primary notes from the rest of the signal - it's too complicated. For example, noise canceling headphones invert the sound it hears and sends it back which shows as you blend sounds stuff gets altered/lost. Put enough notes together and it's nothing but "pink noise" which would be difficult/impossible to pull out the primary frequencies from AND convert to MIDI quickly.

[rcf]

joenoone wrote:
No company that I've seen has built a MIDI device that can take an input signal of multiple frequencies and compute what are the primary notes from the rest of the signal - it's too complicated.

Not yet in hardware, as far as I know. Software wise however, Lateral Solutions & Melodyne have both successfully achieved polyphonic pitch recognition from monophonic sources. I have used the Lateral Solutions Software,and I can say categorically, that it does work, recognizing the individual notes in a chord from a monophonic input source ...so a hardware solution is certainly not impossible, as James from Sonuus hinted at in this very same thread:

We can't publicly talk about products that haven't been officially released, so we can't say if we will be releasing a polyphonic version or not. I can, however, say that our technology will allow this in some form.
There are two types of polyphony and they both have problems:
(1) Using a hex-pickup to separate the output of each string so they can be processed independently. This works very well, but the physical setup of the pickup can be tricky and of course you need the special pickup.
(2) Without a special pickup the audio can be processed and chords extracted.

As for the GI-20, sorry to be pedantic here

The GI-20 requires a GK style pickup (see: http://johnp.net/projects/guitar-synth/ ... nouts.html) which has a separate pickup for each string in it (and NO MIDI interface) which feeds 6 separate audio signals into the GI-20 or other GK compatable device which has 6 pitch-to-MIDI converters built in.

It doesn't necessarily require a GK style pickup. I don't have a GK style pickup to drive my GI-20; my Steve Ryder EM54 Synth Access Emando uses individual RMC piezo saddles to drive the GI-20 polyphonically ...and in this case it's not a hexaphonic pickup system, but a quadraphonic one

Steve Ryder EM54 Synth Access Emando

Steve Ryder EM54 Synth Access Emando.jpg (200.82 KiB) Viewed 252 times