Waves released Vocal Bender

[simon.a.billington]

err, Waves Tunes Real-Time is for "real time t-pain" effect (how 2010s quaint).

And the point is that it DOES have latency, but the latency is fairly small (couple of samples) so it's "nearly realtime", and here's the catch - the latency is not reported.
If you do a loopback on this, or Antares Access, or Waves Tune Realtime, the recorded audio is always a couple of samples late. For true zero-latency plugins, it's never late.

Yes agreed. but it's somewhat of a grey area. A lot of ifs and buts that can change conditions from less than ideal to create an impact. But even true zero can never be true zero as calculations still need to be performed, even if it taps a nanosecond or two to produce it, it still a delay, just an unnoticeable one.

The only time it can be tru zero is if the audio is just passed through like a simple bypass.

No grey area i think, and not a lot of ifs.
Zero delay means that whatever process is taking place can take place in the same allocated Signal Vector size.
As in, if you push audio through and bounce it, it comes out 1:1 as if the plugin weren't there, if you do a loopback through a zero-latency DSP and without it, the looped back audio will be 1:1 on the same spot to the T, it will null-out to infinity.
If there's only slight delay that's unreported, parallel processing goes through the window. it would be extremely obvious in zero-latency compressors if they weren't actually zero latency.

How much time inside the signal vector it needs doesn't really matter - if it needs more time to process it, it causes latency.
That latency is either reported, or it's not reported.
In case of Autotune Access and Waves Vocal Bender, it's not reported, but if you do a loop-back of plugin, the resulted audio WILL be a couple of seconds behind, because there's no such thing as zero latency pitch-shifting. But it can be argued that couple of samples difference in timing of pitched material is not crucial for real-world applications.

That doesnt Mae sense because you can in theory do infinite loopbacks and still have no delay. Thats ALL calculations 1 to infinity done simultaneously without delay. The physics of this universe just doesn't work like that. There's always delay. Even in analogue electronics

[Ploki]

By loopback i mean means of checking whether the sound is actually delayed or not against a reference.
As i said, there's delay, but when its within signal vector, its considered zero delay because it does not ADD to the already existing delay.
In case of AutotuneAccess and Vocal Bender (also crispy tuner) it DOES ADD to existing delay, but it does not REPORT IT, making resulting signal a little late.

[simon.a.billington]

By loopback i mean means of checking whether the sound is actually delayed or not against a reference.
As i said, there's delay, but when its within signal vector, its considered zero delay because it does not ADD to the already existing delay.
In case of AutotuneAccess and Vocal Bender (also crispy tuner) it DOES ADD to existing delay, but it does not REPORT IT, making resulting signal a little late.

Okay, I understand you now. Put that one down to a little misunderstanding.

Pitch manipulation is really hard to do without at least some delay. Trying to run latency inducing plugins with compensation applied can increase the CPU load too. SO it could be that this was a decision based on keeping the CPU hit lower so more people can use it, especially under live conditions.

But hey, I'm completely speculating. What I do know is there is always a reason why a developer takes a particular approach.

[simon.a.billington]

Has anyone tried doing a psudeo double with this thing yet??