OT: good pitch shift plug-in?

[Dave Labrecque]

I'm working on a project which requires a pitch change plug-in as part of the emulation of a noisey guitar pedal board patch sound that I'm doing all in SAW (sans noise :)).

For a couple days I've been pulling my hair out (no mean feat insofar as there is so little) trying to track down a problem I've been hearing in the build mixes.

It's a 70-minute ambient guitar soundscape thing. The gtr signal is split into two paths, one adds nothing and the other adds the pitch change (+7 semitones). Each path then feeds a side of Bob's echo/delay module, wherein each is delayed by a slightly different amount.

The MT sounded great. The build mix sounded great at the start, but toward the end the echo-groove was all wrong. I kept taking stuff out of the picture until I finally pulled the pitch shift plug (SONY's) and, voila, everything fell back into sync. And so it appears that the pitch shift plug, even though it's set to "preserve duration", doesn't. Probably works on shorter stuff (like a typical song or solo), but over 70 minutes, it appears to change the length of the audio stream. D'oh! Why can't stuff just work like it's supposed to?

I tried Sonoris' PitchTime, but at this extreme setting, it breaks up the lower notes in an unpleasant way. Never found out if it would respect the time domain over 70 minutes. I tried Delaydots PitchWorks, which sounded okay, but suffered from the same time dilation issue.

Question 1: Is there a good-sounding pitch shift plug-in that someone could recommend, that wouldn't suffer from this problem?

Question 2: Is there any pitch shift plug-in that wouldn't necessarily suffer from this problem?

Question 3: Is there a work-around in SAW that anyone can think of to make the existing (SONY) plug work right? I've considered turning the bypass on and off every three minutes along the time line, but I'm not crazy about that idea. Quite tedious -- PLUS all that bypass automation on such a plug is asking for big-time sync trouble in my experience. :(

HELP! This project was due two days ago! :eek:

I know Bob's recommended the Anwida stuff recently. I hope the demo will allow me to check a 70-minute session! :o

EDIT: D'oh! No pitch shifter with the Anwida stuff!

[Bob L]

Dave, my guess is that somewhere along the timeline you may be bypassing the pitch shift plugin... that would cause the break in preserved length... it must continually process... even a stream of zeros during blank time...

Could this be what is going on?

Also... make sure the track with the pitch plugin does NOT have the Auto FX Bypass option turned on... right-click the track lablel... this option will be bypassing the plugin during silent sections... again... something you can not do with a time compression effect.

Bob L

[Dave Labrecque]

Dave, my guess is that somewhere along the timeline you may be bypassing the pitch shift plugin... that would cause the break in preserved length... it must continually process... even a stream of zeros during blank time...

Could this be what is going on?

Also... make sure the track with the pitch plugin does NOT have the Auto FX Bypass option turned on... right-click the track lablel... this option will be bypassing the plugin during silent sections... again... something you can not do with a time compression effect.

Bob L

Bob,

Thanks for your quick response.

Track is free of automation. It's a return so auto FX bypass is not even an option.

You don't think the plug could be accumulating a delta T over extended periods?

[Dave Labrecque]

Bob,

I just did a simple test. Clean session. Sony Pitch Shift plug-in on track 1, with 1-sec., 1 KHz regions at 0:00, 2:50, 5:00 and 10:00. I built a mix to track 2. The 1 KHz blips in the build were early by 0, .06, .07, and .09 seconds, respectively.

The Sonoris PitchTime came in at 0, .034, .028, and .033 sec. Doesn't appear to be cumulative on this one.

Delaydots PitchWorks came in at 0, .044, .047, and .067 sec.

I put a region out at the 30:00 mark. It came back 0.135 sec early via the SONY pitch shift plug. Go figure.

Do you expect there's no truly stable (within reason) pitch shift plug out there? :confused:

Hmmmm... I wonder if it's a SAW issue. I'm gonna try some stuff in Sound Forge and report back...

OK... I'm back. Looks like it's in the algo. In Sound Forge I got nearly identical results with the SONY plug. Actually, a tad better in SAW. :)

Whoda thunk it? Is there no such thing as a pitch change plug that will do this without drift? Crazy, man! :eek: How does anyone do pitch change over an extended (more than a couple minutes) without it messing with the tempo? :(

[Dave Labrecque]

My head's starting to hurt over this. If a plug-in can't do it? How can a gtr pedal do it? Unless the pitch change part of the patch is hardware clocked to the same source as the AD/DAs, I guess? :confused:

[mako]

Hello Dave - the smoothest pitch shifter I've tried is the one in Waves' Transform Bundle - it even does a good job on drum tracks in a full mix.

I don't know how it will go for your time duration though.

I'm usually not much of a Waves' fan but this does work well.

cheers

mako

[Dave Labrecque]

Hello Dave - the smoothest pitch shifter I've tried is the one in Waves' Transform Bundle - it even does a good job on drum tracks in a full mix.

I don't know how it will go for your time duration though.

I'm usually not much of a Waves' fan but this does work well.

cheers

mako

Thanks, Mako. If the demo allows an accurate test drive of it's capabilities in my situation, and if it passes the test, I'd be a happy man. Oh -- also if it doesn't break the bank. ;)

But none that I have tested so far seem to have a focus on the kind of time stability I require here. I guess I'm not optimistic. :(

[Bob L]

You may want to use the SS vari-pitch, then use the Sony or Sonoris Time Compression plug to stretch or shrink the length back... perhaps if the plug is only doing one of the functions, the sync accuracy will improve enough to get the job done.

Bob L

[Dave Labrecque]

You may want to use the SS vari-pitch, then use the Sony or Sonoris Time Compression plug to stretch or shrink the length back... perhaps if the plug is only doing one of the functions, the sync accuracy will improve enough to get the job done.

Bob L

I'd momentarily considered that option. Boy, I'm going up 7 semitones. I wonder what that'll sound like with the time compression. I guess that's what the SONY's trying to do all by itself, anyway, so it should be within the realm of possibility, eh?

I'll give it a shot.

Another idea I had (talk about painstaking) is to cut the gtr track up into 2-minute regions and spread them out over as many tracks -- on down the time line, patching the SONY pitch shift plug on each track. 35 or 'em. :eek: I suppose I'd have to automate the bypass to ON at the start of each region and do some SoftEdges to make the transitions decent. I guess the trial and error of dialing in the time compression for your idea is still more palatable, though.

I just can't believe there's no simple way to do this. Could it be that all the pitch shifting that's happening in software plugs across the globe is drifting out of sync?. Nutty. :(

[Tree Leopard]

Dave,

Why not take a two minute segment of the DI-ed guitar and copy the material down to a free track. Load in the Sonoris PS and select the track (i.e. click on the track number) and play it back. Look at the CPU load (might be 4-5% ) and listen. Remove the Sonoris and try and the Sony. Same again with Pitchworks. What happens?

After testing these one by one, if all these plugs work smoothly in track select mode (no other FX) then it's just a case of choosing which sounds more pleasing. To aviod any break-up or timing issues, just make build mixes of pitch FX only. By having a consolidated track you have the option adding other FX to that track or feeding it into a separate output etc.

As you say, working with shorter segments might be the way to go - but just with the pitch FX track(s).

Andre

[Scott Anthony]

Have you tried the free PSP Pianoverb? You could shorten the decay and probably tune it pretty close.

[Gerrysan]

Hi Dave

What if you split your 70 min across 10 tracks of 10 min segments then your pitch shift is only working for a shorter period of time and be less out of sync as not to notice.
Whatdoya think?

Cheers Gerrysan

[Dave Labrecque]

Dave,

Why not take a two minute segment of the DI-ed guitar and copy the material down to a free track. Load in the Sonoris PS and select the track (i.e. click on the track number) and play it back. Look at the CPU load (might be 4-5% ) and listen. Remove the Sonoris and try and the Sony. Same again with Pitchworks. What happens?

After testing these one by one, if all these plugs work smoothly in track select mode (no other FX) then it's just a case of choosing which sounds more pleasing. To aviod any break-up or timing issues, just make build mixes of pitch FX only. By having a consolidated track you have the option adding other FX to that track or feeding it into a separate output etc.

As you say, working with shorter segments might be the way to go - but just with the pitch FX track(s).

Andre

Andre,

I've already determined that the SONY is the best sounding. The problem is the 'drift' issue. And aside ffrom the painstaking aspect of 35 2-minute segments, there's the issue of marrying them into one smooth flow. I predict that will be a PITA, too. :(

[Dave Labrecque]

Have you tried the free PSP Pianoverb? You could shorten the decay and probably tune it pretty close.

Do you think this is plausible for +7 semitones of re-tuning?

[Dave Labrecque]

Hi Dave

What if you split your 70 min across 10 tracks of 10 min segments then your pitch shift is only working for a shorter period of time and be less out of sync as not to notice.
Whatdoya think?

Cheers Gerrysan

I've determined that I need to bring the segments closer to 2 minutes each to keep the groove in the pocket (how's that for a mixed metaphor?). That opens me up to the headaches detailed in my other posts, here. :(

[Dave Labrecque]

You may want to use the SS vari-pitch, then use the Sony or Sonoris Time Compression plug to stretch or shrink the length back... perhaps if the plug is only doing one of the functions, the sync accuracy will improve enough to get the job done.

Bob L

Bob,

There are only three 'streams' total: unpitched gtr, pitched gtr, and vocal. It occurrs to me that another approach would be to slightly re-pitch or time compress the other two against the SONY pitch-shift buildmix, rather than BIG SAWpitch/time compression on the one.

I'll report back...

[bcorkery]

Dave,

After you crack this nut, you're going to have to listen to some relaxing, meditative music. :)

[Dave Labrecque]

Dave,

After you crack this nut, you're going to have to listen to some relaxing, meditative music. :)

:)

Not to worry. At this point I'm already hearing it in my sleep. :p

[bit]

I've had this problem with the sony plugs too.
When I loop stuff I use the saw sequence varipitch function and pitch the tone back with the sony plug. Then I build to a new track and fine-times it after the mother track. This won't work in your situation as far as I understand. But I have a vage recolation that if the Sony pitch is plugged on an aux return, things behave differently.
If that doesn't work, I would just render the song part by part, 3 minutes at the time to get the project ready for the customer.

Bjørn

[Dave Labrecque]

I've had this problem with the sony plugs too.
When I loop stuff I use the saw sequence varipitch function and pitch the tone back with the sony plug. Then I build to a new track and fine-times it after the mother track. This won't work in your situation as far as I understand. But I have a vage recolation that if the Sony pitch is plugged on an aux return, things behave differently.
If that doesn't work, I would just render the song part by part, 3 minutes at the time to get the project ready for the customer.

Bjørn

Thanks, Bjørn. I'll have to try the aux return route sometime.

What I ended up doing was building the SONY-pitch-shifted gtr out to a dedicated track and used Bob's built-in vari-pitch to adjust the length as close as possible to the unpitched stuff.

One problem was that not only does the SONY plug (and maybe all such plugs) have an overall time shrinkage outcome, but there are "eddies" in this residual time compression such that the compression ratio modulates over time slightly. ARGH. I'm guessing that this happens because the algorithms are somewhat signal-content-dependent?

Also -- kind of doing a ruff average calculation (because the amount of time dilation, as I said, was variable), the amount of vari-pitch I needed to apply was about minus 0.01 percent! Sounds like it's minimal, but after 70 minutes that's almost half a second of slop -- no good on the gtr's cool echo groove.

Fortunately, the calc came out really close to 0.01 percent, which is the resolution of the tweakability I had available to me. A little more or a little less needed pitching, and I would've been hosed. :eek:

I'm still thinking there must be a more serious way to do this kind of critical pitch shifting in the DAW world. :confused:

[Bob L]

How about Melodyne?

Bob L

[Dave Labrecque]

How about Melodyne?

Bob L

Considered it. Didn't try it (lack of funds). Could be a solution, though, eh?

I'm not hip (what is hip?) -- does Melodyne work in SAW? Or is it stand-alone?

[Mark Stebbeds]

It's a 70-minute ambient guitar soundscape thing. The gtr signal is split into two paths, one adds nothing and the other adds the pitch change (+7 semitones). Each path then feeds a side of Bob's echo/delay module, wherein each is delayed by a slightly different amount.



I've been reading this thread with some interest, but didn't really focus on the detail because I was not intimate with the plug-ins you were using.

It's really not clear to me how you are using the pitch change. Are you creating another track that the pitch is shifted on, or in line as a plug in, and the undesireable aritifact is file length change, resulting in tempo/groove mess accentuated by the delays?

It would seem to me that a solution might be to start over, (or mute those tracks and start with the virgin gtr.).

Try sending to the delays first (not the pitch shifter), and set up your groove. You might consider 1/4 notes on one return, and triplets on the other. (just an opionion)

THEN, try to pitch up one of the delays, and add some atmosphere with reverb on the various delays and original track. I'm thinking some element of pre send where you hear more pitched up delay drenched in verb without a lot of orignal delayed signal.

Just a thought. Some great effects on analog consoles are created by setting up delays, or pitched delays, but taking them out of the stereo buss and using them only as sends to reverbs.

A very popular hardware box still used today is the AMS 1580.(I think that's the model number. ) Two channels of delay with pitch shift. Think like that instead of pitch shifting the entire track and then adding delays later.

My. $.02

Mark

[Tree Leopard]

Dave -This thread has been very interesting for me too. If anything, you've highlighted one of the (many) limitations of native DAWs for intensive FX processing. So this has actually been very helpful.

Top-notch pitch shifting software - like Serato Pitch 'n Time or Prosoniq Time Factory - is not exactly cheap and for large files you need a fair amount of processing power & RAM for smooth operation. The trade off is that the results are usually excellent. Likewise the sound warping ideas using convolution I tossed into the previous harmonizer thread are really "write to disk" processes. (I haven't played with Melodyne enough to know what it's really capable of, but for the price it seems very good indeed.)

But for dynamic "live" processing - which you're after - it just makes me think how sensible it would be to invest in a good quality outboard (as Mark is suggesting). For my own sound design work something like a SPX2000 is now looking like an attractive option - as much as I'd love an Eventide of one flavour or another.

Anyway, good luck with the rest of the project!

Andre

[MMP]

I'm not hip (what is hip?) -- does Melodyne work in SAW? Or is it stand-alone?

The answer is yes.

It works as a stand alone, but has an ability to interface to SAW through the "Melodyne Bridge", which syncs the two applications and allows the audio to be passed back & forth. Melodyne is quite solid as a stand alone, but using the bridge mode is a little tricky, at least on my system. For instance, if I forget and try to control Melodyne using its own transport instead of SAW's in bridge mode, Melodyne crashes. Little things like that "train" you fast.

Regards,

MM

[Mark Stebbeds]

But for dynamic "live" processing - which you're after - it just makes me think how sensible it would be to invest in a good quality outboard (as Mark is suggesting). For my own sound design work something like a SPX2000 is now looking like an attractive option - as much as I'd love an Eventide of one flavour or another.



Outboard would certainly work, but what I was suggesting was to think about using plug-ins like the outboard device works.

Meaning shift the pitch of the delays instead of pitch shifting an entire track and adding delays to that. It should solve/mask the timing and groove issues.

Mark

[Dave Labrecque]

I've been reading this thread with some interest, but didn't really focus on the detail because I was not intimate with the plug-ins you were using.

It's really not clear to me how you are using the pitch change. Are you creating another track that the pitch is shifted on, or in line as a plug in, and the undesireable aritifact is file length change, resulting in tempo/groove mess accentuated by the delays?

It would seem to me that a solution might be to start over, (or mute those tracks and start with the virgin gtr.).

Try sending to the delays first (not the pitch shifter), and set up your groove. You might consider 1/4 notes on one return, and triplets on the other. (just an opionion)

THEN, try to pitch up one of the delays, and add some atmosphere with reverb on the various delays and original track. I'm thinking some element of pre send where you hear more pitched up delay drenched in verb without a lot of orignal delayed signal.

Just a thought. Some great effects on analog consoles are created by setting up delays, or pitched delays, but taking them out of the stereo buss and using them only as sends to reverbs.

A very popular hardware box still used today is the AMS 1580.(I think that's the model number. ) Two channels of delay with pitch shift. Think like that instead of pitch shifting the entire track and then adding delays later.

My. $.02

Mark

Hi Mark,

Thanks for the input.

I had the DI gtr track split into two streams, one going to the delay and the other getting pitched, then going to the delay. The problem is that as long as this all stays in the digital domain, the pitch shift plug time compresses ever so slightly (about .01 percent), but enough that after a few minutes, the slap groove is off (dry vs. pitched/delayed signal).

Unless there's a plug that doesn't time compress (Melodyne?), I don't see a way around this without going out to analog. Maybe I shoulda done that.

Next time! :)

[Dave Labrecque]

Dave -This thread has been very interesting for me too. If anything, you've highlighted one of the (many) limitations of native DAWs for intensive FX processing. So this has actually been very helpful.

Top-notch pitch shifting software - like Serato Pitch 'n Time or Prosoniq Time Factory - is not exactly cheap and for large files you need a fair amount of processing power & RAM for smooth operation. The trade off is that the results are usually excellent. Likewise the sound warping ideas using convolution I tossed into the previous harmonizer thread are really "write to disk" processes. (I haven't played with Melodyne enough to know what it's really capable of, but for the price it seems very good indeed.)

But for dynamic "live" processing - which you're after - it just makes me think how sensible it would be to invest in a good quality outboard (as Mark is suggesting). For my own sound design work something like a SPX2000 is now looking like an attractive option - as much as I'd love an Eventide of one flavour or another.

Anyway, good luck with the rest of the project!

Andre

Thanks Tree. Ya know -- I never did try Eventide pitch shift convolution IRs in SIR for this. I wonder how that would've worked out...

[Dave Labrecque]

The answer is yes.

It works as a stand alone, but has an ability to interface to SAW through the "Melodyne Bridge", which syncs the two applications and allows the audio to be passed back & forth. Melodyne is quite solid as a stand alone, but using the bridge mode is a little tricky, at least on my system. For instance, if I forget and try to control Melodyne using its own transport instead of SAW's in bridge mode, Melodyne crashes. Little things like that "train" you fast.

Regards,

MM

:) Be trained or "train wreck". The choice is yours. :p

[Mark Stebbeds]

I had the DI gtr track split into two streams, one going to the delay and the other getting pitched, then going to the delay. The problem is that as long as this all stays in the digital domain, the pitch shift plug time compresses ever so slightly (about .01 percent), but enough that after a few minutes, the slap groove is off (dry vs. pitched/delayed signal).



Maybe I didn't explain correctly. I was only using the hardware piece as a suggestion on how to duplicate with software.

My suggestion is to pitch shift the delays instead of adding delays to the pitch shifted track, which isn't working for you. That's how we would have done it in the good ole days.

Mark

[studio-c]

I'm usually not much of a Waves' fan but this does work well.

cheers

mako
Not to hijack, but the guy down the hall just lost 2 ILOK dongles. He's out about 15 grand, and gets to spend next week begging and pleading for re-authorization (probably from some really jaded customer service person who's heard it all before).

Okay, back to your thread. BTW I just bought the Sonoris PitchTime. Is that back on topic? :)

Scott

[DominicPerry]

I get some pretty bad glitching with Sonoris Pitch Time. The processor doesn't seem overloaded, just lots of stuttering. What am I doing wrong (apart from using a singer whose out of tune in the first place)?

Dominic

[MMP]

Are you using UAD-1 plugins on the same track?

I have a problem with all the Sonoris plugins when they are on the same track as UAD-1 plugins.

It's a mystery...

MM

[DominicPerry]

No, not using UAD but thinking about it, maybe I had SIR running. I'll check. Thanks.

Dominic

[Bob L]

Many plugins that change buffer sizes do not like other plugs that do also... and many times, certain plugins will not tolerate buffer size changes at all... so you just have to test and figure what will work together.

The Pitch/Time plug should not cause any glitching when used on its own... and if you are compressing time... then you should definitely use it by itself... buildmix to a new file and then use that file in complex mix environments with other finicky plugins.

Bob L

[Tree Leopard]

Maybe I didn't explain correctly. I was only using the hardware piece as a suggestion on how to duplicate with software.

My suggestion is to pitch shift the delays instead of adding delays to the pitch shifted track, which isn't working for you. That's how we would have done it in the good ole days.

MarkMark - were you referring to the AMS-DMX 1580s? Like this (http://cgi.ebay.com/Overhaul-service-and-calibrate-your-AMS-DMX-15-80S_W0QQitemZ270017716991QQihZ017QQcategoryZ47125Q QcmdZViewItem) one at eSpend? This AMS has a harmonizer function, in the form of tuned delays. OK you'll get a bit of noise, but it's still a great leap forward over Zoom and Boss (Roland) pedals and mini racks.

For anyone here who's curious, have a look at what the SPX2000 (http://www.yamaha.com/yamahavgn/CDA/ContentDetail/ModelSeriesDetail/0,,CNTID%253D43007%2526CTID%253D560494,00.html) offers. It's funny - add the cost of equivalent "carefully modelled" plugins vs. purchasing one of these, it begins to make sense to just buy the box. Another fave is the Sony V77 (from a decade ago). Once you started playing with the presets you could extract some excellent and unusual FX from this unit.

Muse Receptor is about the only hybrid VST + external hardware option out there that will support a wide range of plugins, but I'm not sure about it yet. Seems better suited to VSTi.

Just tossing in some alternatives here.

Andre

[Mark Stebbeds]

Mark - were you referring to the AMS-DMX 1580s? Like this (http://cgi.ebay.com/Overhaul-service-and-calibrate-your-AMS-DMX-15-80S_W0QQitemZ270017716991QQihZ017QQcategoryZ47125Q QcmdZViewItem) one at eSpend? This AMS has a harmonizer function, in the form of tuned delays. OK you'll get a bit of noise, but it's still a great leap forward over Zoom and Boss (Roland) pedals and mini racks.



Yep, that's the one. The noise is negligible, although with digital recording and plug-ins, we have become hyper-sensitive to noise and hiss.

I think the advertised price in that ad is for servicing/bring to spec the 1580 or the AMS RMX 16 reverb, the box that set bar for non-linear reverb.

Mark

[Dave Labrecque]

Maybe I didn't explain correctly. I was only using the hardware piece as a suggestion on how to duplicate with software.

My suggestion is to pitch shift the delays instead of adding delays to the pitch shifted track, which isn't working for you. That's how we would have done it in the good ole days.

Mark

Mark,

I understood that. But the thing is... it wouldn't matter. The unwanted time compression of the pitch shift would happen either before or after a delay of the signal. Dontcha think? I could try an experiment, I suppose, but I can't see the plug-in suddenly not time compressing just because it came after a delay instead of before it.

[Mark Stebbeds]

The unwanted time compression of the pitch shift would happen either before or after a delay of the signal. Dontcha think?

No, but I see where you are going with this. I think it's worth an experiment if you have the time.
.
When you add a digital delay effect, there should not be an issue of time compression as time moves along the timeline. In other words, a 200ms delay should be exactly the same at the end of the track as it is at the beginning.
Each note played is delayed 200 ms.

Just shift the pitch of that. Where I may be wrong in my thinking is assuming that each new note triggers a new starting point as in a analog delay with the pitch shifted.

Perhaps that's an idea. Automate muting the send in between notes as the guitar pauses, or if you're triggering the pitch off of the delays, as each delay sequence fades out into the track.

Mark

[Dave Labrecque]

Where I may be wrong in my thinking is assuming that each new note triggers a new starting point as in a analog delay with the pitch shifted.

I think you are right in that this is where you are wrong. :)

With these all-digital all-the-time processes, there's no starting and stopping of notes as far as they're concerned. It's all one uninterrupted digital stream of stuff, whether or not a note is playing, whether or not the delay is before the pitch shift.


Perhaps that's an idea. Automate muting the send in between notes as the guitar pauses, or if you're triggering the pitch off of the delays, as each delay sequence fades out into the track.

Mark

Two reasons I don't think this would work:

1 Muting doesn't change the continuous nature of the data flow, so the pitch shifter wouldn't ever get a chance to "start over" and thus minimize it's inherent time compression over long periods. (I could be wrong about this -- maybe muting does interrupt the processing?)

2 There really aren't too many silent points where the muting could come into play; it's a pretty continous performance. Then again, I haven't gone back and listened to it with this purpose in mind, so I could be full of sh*t here.

It'd be interesting to see if the muting interrupts the processing; and if the pitch shifter would "know" to reinitialize it's processing with the next piece of audio that comes down the line. Much to my disappointment, I ain't got no time to check it out, though. :(

[Mark Stebbeds]

1 Muting doesn't change the continuous nature of the data flow, so the pitch shifter wouldn't ever get a chance to "start over" and thus minimize it's inherent time compression over long periods. (I could be wrong about this -- maybe muting does interrupt the processing?)



It seems to me that no processing can take place until there is an input, and then it would trigger.

Even if it was too much effort to mute/unmute too many times, I think the timing issue that is messing up the groove would be FAR less noticable if you were pitch shifting the delays, which should remain constant.

Mark

[Dave Labrecque]

It seems to me that no processing can take place until there is an input, and then it would trigger.

That could be. I'd be interested to know the truth about that. I guess it would come down to whether or not muting actually creates no "input" (rather than a stream of zeros, say) in SAW.


even if it was too much effort to mute/unmute too many times, I think the timing issue that is messing up the groove would be FAR less noticable if you were pitch shifting the delays, which should remain constant.

Mark

And I think the artifactual (yeah, I made that up) time compression would be identical regardless of pitch shifting then delaying, or delaying then pitch shifting. Why is it that you think the order of these processes would alter the way the pitch shift algorithm works on the signal? I'm thinking that the signal on the OUT side of the pitch shifter is time compressed compared to the signal on the IN side of the pitch shifter, regardless of what processing may pre- or proceed the effect.

[Mark Stebbeds]

Why is it that you think the order of these processes would alter the way the pitch shift algorithm works on the signal? I'm thinking that the signal on the OUT side of the pitch shifter is time compressed compared to the signal on the IN side of the pitch shifter, regardless of what processing may pre- or proceed the effect.


Because the delays are consistant and set the rhythmic pattern you are saying is getting messed up. The original gtr is never sent to the pitch shifter.

I'm not sure we are communicating on this one, and I'm not sure you can set up SS in the way I'm thinking....but there probably is a workaround.

This is how I would set it up on an analog console or in PT. ...

Gtr track-->aux send 1 to delay-->delay on aux return 1-->aux send 2 to pitch shift-->pitch shifter on aux return 2. The original track is not sent to the pitch shifter at all.

or alternately, put the pitch shift on the insert of aux return 1 (delay)

All three channels could send to a reverb on aux 3 and returning wherever.

Mark

[Jay Q]

Because the delays are consistant and set the rhythmic pattern you are saying is getting messed up. The original gtr is never sent to the pitch shifter.Mark, I think I understand what you're saying. However, if what Dave is saying is true, then it wouldn't make any difference where in the chain you put the plugin or what signal it receives. It depends on the specific plug, but many do operate as Dave suggested, i.e., once they're triggered, they essentially loop until they're shut down by stopping playback. You can think of it as a cycle which is unaffected by input or lack thereof, so even if there's no input, the plugin marches along its cycle, not "caring" whether it's receiving a signal. Hence, if it's off by .03% or whatever, it'll continue to drift further from the intended value as time progresses no matter where it's patched. I've experienced this kind of thing with a number of plugs and have tried "tricking" them by doing numerous things including what you've suggested but to no avail.

Jay

[UpTilDawn]

I think Jay is right here. I've seen the time compression plug in SF change the timing between two tracks when processed independently but to the same settings and the same original track length (that's processed within SF and not as a realtime plug within Saw).

You may be able to process the delays properly by putting the plug first, but the time compressor will do whatever it wants regardless of the delay timing.

Seems to me that processing the pitch shift in smaller segments after the fact is the only hope for getting this accomplished.

What I think,

DanT

[Mark Stebbeds]

Mark, I think I understand what you're saying. However, if what Dave is saying is true, then it wouldn't make any difference where in the chain you put the plugin or what signal it receives. It depends on the specific plug, but many do operate as Dave suggested, i.e., once they're triggered, they essentially loop until they're shut down by stopping playback.

I can't understand how any plug-in can process anything that is not on it's input, let alone output audio BEFORE what is on the input, unless it is "looking ahead" which is a bad idea in this senario.

I just did a test of the Sony Pitch Shift plug-in that started Dave's issues, and made a very interesting discovery.

When processing a 12 1/2 minute file in Sound Forge using the same 7 semi tone settings as Dave ( a musical fifth, quite a big bump), and preserve duration option checked, the results were SAMPLE ACCURATE over 12 1/2 minutes.

Not knowing which "mode" Dave was using, I experimented with several "lower quality, but faster" modes. The worse case senario I could find was .0000653% of time EXPANSION, certainly negligable over 12 1/2 minutes, but many other modes were also sample accurate.

So my testing indicates that the Sony Pitch shifting plug in is sample accurate over long time periods. I did not try a 70 minute segment as Dave is using, and to be fair, my source sample rate was 22.025k on a speech file.

So it seems to me something else is going on in real time, although my testing methods weren't perfect.

So Sound Forge may be a solution.

This thread has piqued my interest.

Mark

[Mark Stebbeds]

The gtr signal is split into two paths, one adds nothing and the other adds the pitch change (+7 semitones). Each path then feeds a side of Bob's echo/delay module, wherein each is delayed by a slightly different amount.



Expain this routing in more detail, specifically the "split into two paths" and "each path feeds" something.

Mark

[Jay Q]

I can't understand how any plug-in can process anything that is not on it's inputZeros can be and are processed -- of course, nothing audible is produced. Like I said, think of it as a loop which is set in motion when playback starts. For example, I use a Leslie simulator plug where the first attack of the Rhodes sounds different depending on where the cursor is when I start playback because the signal is "landing" in a different part of the loop. The Leslie keeps swirling at an even rate regardless of whether there's any input.


When processing a 12 1/2 minute file in Sound Forge using the same 7 semi tone settings as Dave ( a musical fifth, quite a big bump), and preserve duration option checked, the results were SAMPLE ACCURATE over 12 1/2 minutes.That is interesting. Hopefully, Dave'll chime in.

Jay

[Mark Stebbeds]

For example, I use a Leslie simulator plug where the first attack of the Rhodes sounds different depending on where the cursor is when I start playback because the signal is "landing" in a different part of the loop. The Leslie keeps swirling at an even rate regardless of whether there's any input.

Well, of course. A flanger does the same thing. But the resultant real time processing in Dave's senario is shifting time when it shouldn't be.



When processing a 12 1/2 minute file in Sound Forge using the same 7 semi tone settings as Dave ( a musical fifth, quite a big bump), and preserve duration option checked, the results were SAMPLE ACCURATE over 12 1/2 minutes.


That is interesting. Hopefully, Dave'll chime in.

It's very interesting.

Mark

[UpTilDawn]

As I pointed out, I have experienced a similar discrepency using this same plugin in Sound Forge, where two tracks from a single session, processed with the same settings over a (length of time I don't recall) period of time and originally the same length ended up with beats out of sync... Imagine two overhead mic channels processed to produce a slower tempo, and set to maintain pitch and duration ending up with flammed beats.

I think that the plugin itself may cause some other than expected changes to the content.

I may try a long segment of drum overheads tonight to check the results again since I have stopped processing these problem areas in this way because of the results I had gotten in the past, so it's been a long time ago.

DanT

[Jay Q]

But the resultant real time processing in Dave's senario is shifting time when it shouldn't be.Right.... So, what's the point again? ;) Oh, that the plug's location in the chain doesn't necessarily have anything to do with that time shift. I guess we won't know unless Dave tests it exactly as you did.

Can't recall if it's been mentioned already, but it's certainly the case that individual processors will behave differently when performing the kind of number-crunching required in Dave's case, so the problem may be the result of some limitation of the code and/or his processor.

Jay

[Mark Stebbeds]

Right.... So, what's the point again? ;) Oh, that the plug's location in the chain doesn't necessarily have anything to do with that time shift. I guess we won't know unless Dave tests it exactly as you did.



No, that's not the point. The point is it shouldn't be doing it at all.:)

The point of the location in the chain was one of perception, and a workaround.

Mark

[Jay Q]

No, that's not the point. The point is it shouldn't be doing it at all.:)

The point of the location in the chain was one of perception, and a workaround.Guess we were making different points then. :)

Jay

[Dave Labrecque]

Sorry I've been gone, gents. Two-day shoot spaced over three days. I got home at 5 this morning and slept six hours. Been playing catch-up all day here at the studio. Nice to see the discussion has been moving forward in my absence. :)

Two things:

1) Wow. Mark says he had sample-accurate performance of the plug in Sound Forge, while I had noticeable (and problematic) inadvertent time compression happening in SAW. I need to do an apples-to-apples test, but could it be that SAW's code is somehow getting in the way of the Sony plug's proper performance in this area? I hope I can find some time to test this, but I'm gone all next week (Heading out of Phoenix at 5 AM Satuday. That's a two-hour drive. Plus terrorist-induced early-arrival time. Ouch.), so I don't know when I might get to it. FYI, the algo I'm using in the Sony pitch shift plug is "A11. Solo Instruments 2". Feel free to experiment amongst yourselves...

2) Here's the work-around I used to make it work to my satisfaction (have I posted this yet?):

A Built the Sony pitch-shifted gtr track to a new track
B Determined that the overall time compression, though variable over time, averaged very close to 0.01%.
C Vari-speeded the gtr region -0.01%
D Sounds good to me.

Gee, I guess I coulda tried putting a vari-speed type plug-in after the pitch shift plug and set it to -0.01% for realtime delivery. Maybe the Sony again without "preserve duration" enabled? :p

[Mark Stebbeds]

1) Wow. Mark says he had sample-accurate performance of the plug in Sound Forge, while I had noticeable (and problematic) inadvertent time compression happening in SAW. I need to do an apples-to-apples test, but could it be that SAW's code is somehow getting in the way of the Sony plug's proper performance in this area? I hope I can find some time to test this, but I'm gone all next week (Heading out of Phoenix at 5 AM Satuday. That's a two-hour drive. Plus terrorist-induced early-arrival time. Ouch.), so I don't know when I might get to it. FYI, the algo I'm using in the Sony pitch shift plug is "A11. Solo Instruments 2". Feel free to experiment amongst yourselves...


Good to know the mode you were using is "A11", as there are many. I was using the default "A01", and randomly selected a few others. If I find the time, I'll check a little more, but I am also planning to try and enjoy the last official weekend of summer. Have a nice holiday.

Mark