Continued discussion about Warp Modes with unwarped Clips

[Timur]

I fork this from a Tips & Tricks thread that I don't want to get bloated. Nevertheless I want to start another discussion on how different Warp Modes affect clips when clip tempo is set equal to global tempo based on my very own test with Live 7 beta. I have found kind of an "issue" or "bug" regarding Autowarp, so anyone who want to dive into this can find my post in the Live 7 beta forum, but ignoring that I'd like to repeat the following statement.

As long as you don't use Autowarp and your clip's tempo is the same as global tempo the output of all Warp modes except Complex remains the same bit by bit!

To verify this statement I took several different samples (including a complete 2:30 song by System of a Down) and rendered them unwarped and warped in all Warp Modes. Result is the above statement, the rendered files are exactly the same when doing a comparison on a byte-basis. No Inversion tests needed actually, but sure you can put an Utility effect with Inversion on them and will hear silentium.

Autowarp is a different beast that leaves to many different results. And since I consider the inconsistent behavior of Autowarp more of a bug I can well imagine that Autowarp is to be blamed for all the different results that older discussions came up with. Although it is also possible that the internals were simply changed with Live 7, but that is exactly the point where we can start discussing how the current (to be) product works.

[Timur]

These are the statements so far:

if i recall correctly from prior threads, repitch is actually safer than beats at same tempo (because there are some rare instances in which beats causes problems if two tempos are not exactly the same (beyond the decimal places shown by live)).

I remember it as the opposite. A potential 1-2 sample offset due to jitter or something, in pitch mode...

if i recall correctly from prior threads, repitch is actually safer than beats at same tempo (because there are some rare instances in which beats causes problems if two tempos are not exactly the same (beyond the decimal places shown by live)).

actually you are right but you got it the wrong way round!

it's repitch that can waver over a period of time - Ingo from Abe clarified that a year or two ago

I thought the same - it seems to make the most sense to use repitch as there is never any kind of warping, only altering pitch, but beats is the one that does nothing until it detects that there is a difference between clip and master tempo

the problem is the variations beyond 2nd decimal place as you pointed out, but using Angstrom's method here if you type in the value manually you are avoiding this

Thanks Angstrom - nice post!

Again, no offense, but there has been quite a bit more experimenting done by many people over a very long time...

I think in your specific case you are probably right, and the potential errors in the other three modes can be traced to decimal rounding problems if tempos are not set exactly.

By toggling the warp button you're probably eliminating that problem, although it would depend on how Ableton applies the tempo to the clip field. It still may be safer to actually type in both values.

However I would still recommend to any and everyone - to use beats mode as your default warp mode. That is confirmed by Ingo @ Ableton,as forge said.

[Timur]

when you load it unwarped and warp manually or disable/re-enable Warp of an autowarped clip it will be set to the value of the global tempo with only one warp-marker set at the beginning of the clip.

I'd have to check on this, but I think this is only true of a clip which doesn't have warp marker info saved in an .asd file yet.

Because there are definitely clips on which turning on/off warp always retains the set warp markers

Normaly if you change anything about warp markers Live asks you if you want to keep the tempo of the clip in a popup window. If you says "Yes" then it deletes all warp markers except the first at the beginning and sets the Sequence/Clip tempo to the value of the global tempo. But I have to admit I did not try to save .asd files yet.

[Timur]

I would love to get hand on some samples that older tests have found to be produce different results with different warp modes. PM me or post a link here so that we can try them with Live 7.

[longjohns]

Some remarks concerning unwarped playing
for warpmodes when the file tempo equals song tempo:

Unfortunately there is currently no indicator in the gui when
file tempo or song tempo are cut of after the second decimal.
So they might well differ a small amount, even though they
seem to be equal. This is what i assume to
be the case shown in the video.

In fact, even for such small differences in the tempo
this will not immediatly lead to warping for most warp modes,
they propably will make a different decision as to what grain
is played at some points but for most of the time will just
play unwarped with a constant sample offset resulting
from such decision.

There seems to be a quirk in the "tones" warper that
causes such an offset quite from the beginning.

"Repitch" starts pitching the sample to accomodate the
small tempo difference, and in effect diverts
more and more from the unwarped sample.

"Complex" mode is different from all others in that it
does no "grained" playing of the existing material.
It performs a time base analysis and transformation
of the input material. Furthermore "complex" also
does the resampling necessary for
sample rate conversion. In effect it diverts in almost all
cases from the "unwarped" sample.

Independed of the above i experienced single sample
offsets (from jitter) between warped and unwarped after subsequent
starting and stopping with "Shift+Space".

For the time being (meaning as long as there are no
envelopes for unwarped clips, no "clip tempo to song tempo" button)
i suggest the workaround:
- use more or less integral clip tempo
(equaling song tempo, naturally)
- use beats mode

I hope this was helpful,

regards, ingo

[Timur]

I knew this post, so let's reply to it:

Unfortunately there is currently no indicator in the gui when
file tempo or song tempo are cut of after the second decimal.
So they might well differ a small amount, even though they
seem to be equal.
...
or the time being (meaning as long as there are no
envelopes for unwarped clips, no "clip tempo to song tempo" button)

But in fact there is such a button: Warp
If you disable warp and start a whole new Warp for a clip (which will delete all it's warp-markers and changes you made to the clip's Warp settings) it will set the clip exactly to song-tempo (global tempo).

Independed of the above i experienced single sample
offsets (from jitter) between warped and unwarped after subsequent
starting and stopping with "Shift+Space".

I just did the following (Autowarp was disabled!): Played 10 tracks of the same 00:22:50 clip looping with two tracks using each warp mode except Complex plus unwarped. I kept pushing shift-space time after time, not a single jitter audible. Then I put Inversion Utilities on half of the tracks and shifted them around between random tracks with different Warp modes on them, no sound audible except for a short crackle whenever the Utility was moved over to a different track.

Here an list of the track-setup:

1. Unwarped
2. Beats
3. Tones
4. Textures
5. Repitch
6. Unwarped
7. Beats
8. Tones
9. Textures
10. Repitch

[Timur]

I enabled some warp-makers and use the SAVE option. As soon as you disable and re-enable Warp the aforementioned popup will ask if you want to preserve the tempo. Say "Yes" and your clip will be at song-tempo!

And I tried Live 6.10 and found out that it works exactly the same as Live 7 as far as the popup is concerned (only that I cannot use the SAVE button there because I only use a Demo version). Even putting up a couple of tracks with different warp modes and running them in parallel with Inversion Utilities results in silence. I cannot render with my Demo version but I suspect the results will be on par with my Live 7 results.

[dj superflat]

i'm not sure your test really captures everything. e.g., does the same apply where you have short loops that are getting retriggered all the time? i thought, from ingo's reply, that's where some of the potential issues arose.

sides which, why are discussing this? even if modes other than beats are only rarely a problem (leaving aside complex, which often is), why not still just use beats as default?

[Timur]

i'm not sure your test really captures everything. e.g., does the same apply where you have short loops that are getting retriggered all the time? i thought, from ingo's reply, that's where some of the potential issues arose.

That can easily be tested. How short do the loops have to be and how often retriggered? I just created loops of 0.0.2 length and copied them 30 times on 5 tracks (one unwarped, the other each with a different Warp mode except Compex). Same results, same rendered files, same silence when inverted.

sides which, why are discussing this? even if modes other than beats are only rarely a problem (leaving aside complex, which often is), why not still just use beats as default?

Because I want to understand what alters my sound and what not, and if the former discussions have led to wrong results and people thinking that any other mode than Beats is to be avoided as Default then we should know about it. Some people might not want to use Beats as default because they need another mode more often for their warping. Take a look at the confusion in the above quotes "Was it Beats or was it Repitch that could be used for unaltered warping?"

Another reason is that if I understood the mechanics right and if I am right about the Warp modes (which is not the point, you can gladly prove me wrong) then Autowarp is faulty to some extend, because I can trace all differences back to using Autowarp in my own tests. If so then again I want to know what alters my sound and which tools to use safely.

And this is quite important: when using Autowarp then all Warp modes may alter the sample including Beats!

[Tone Deft]

Timur - what are you on about? that's a lot to read, digest and address... can you boil all this down to a few sentences?

[dj superflat]

i too can't follow most of what you're concerned about, particularly re auto-warp (my observations are based on comments in the old thread i direct you to). but, so far as i can tell, beats and repitch generally will be fine, tones may have somewhat more issues, complex clearly does. i think the latter was the main problem, because many people equated complex with best (when it often isn't). once that was cleared up, most people were fine.

[Timur]

but, so far as i can tell, beats and repitch generally will be fine, tones may have somewhat more issues, complex clearly does. i think the latter was the main problem, because many people equated complex with best (when it often isn't). once that was cleared up, most people were fine.

Well, this is actually what I'm concerned about. There is no "best" in Beats and there is no "sub-par" in any of the other Warp Modes and there is a "always different" in Complex, at least as far as my own tests go.

There seems to be wrong folklore going around how to use Warp when you want to loop clips without actually altering/warping them. And there seems to be a lack of true understanding/knowledge (including my own) what is going on "under the hood" (as to quote Ableton's CEO) when discussions about the sound-engine and its respective quality come up. Also I seem to have found several sound-altering bugs/quirks in only a few days time that have not been discussed on this board before, while things like this unwarped looping have been discussed "ad nausae" and still seem to lead to no real conclusion. And even the conclusions that have been found may have been build on false assumptions.

That doesn't mean that I'm the "know it all", far from it. But we need to know how things work to be able to find bugs and discuss quality and enhancements. And that's why I start discussing things in order to share what knowledge is around. It's been a pain to search through pages of archive posts just to find out that the wisdom therein does not apply to my own experience. So either I'm doing something wrong or the past conclusions were wrong (or Ableton changed something about the sound-engine since then).


PS: That being said I am still hoping/waiting for someone who could answer my specific questions about Latency Compensation.

[Timur]

Timur - what are you on about? that's a lot to read, digest and address... can you boil all this down to a few sentences?

I can try!

- All warp modes produce the same output when running clips at song-tempo. Autowarp is inconsistent at best, probably even broken. I suspect this has been the case for a looong time without anyone noticing.

- Either Ableton has changed something about the sound-engine/warp modes or all past discussions/assumptions about this matter were wrong or my tests are all wrong (or using wrong samples).

- Most discussions about quality of Ableton's sound-engine lack knowledge of what's really happening there, where degradation/alteration happens. So either we need to find the real sources of alteration or should cease to discuss this matter all along (not even starting talking about the "advantages" and "brighter sound" of 64-bit summing and such).

Ah, and the reason I have time to dig so deeply into all this is because my Lady is out of house for several days and I'm working to wards mastering my tools at the moment before starting to make music again. Once I start making music I don't want to be bothered with all this anymore, but want it to work properly and grand!

[leisuremuffin]

where's rihky and douche-von-picklester?



.lm.

[Tone Deft]

- All warp modes produce the same output when running clips at song-tempo. Autowarp is inconsistent at best, probably even broken. I suspect this has been the case for a looong time without anyone noticing.

OK, two ideas in one bullet point, what? first point, yes. second point, what are you talking about?

- Either Ableton has changed something about the sound-engine/warp modes or all past discussions/assumptions about this matter were wrong or my tests are all wrong (or using wrong samples).

what are you talking about? change from what to what? the warp modes are different.

- Most discussions about quality of Ableton's sound-engine lack knowledge of what's really happening there, where degradation/alteration happens. So either we need to find the real sources of alteration or should cease to discuss this matter all along (not even starting talking about the "advantages" and "brighter sound" of 64-bit summing and such).

whatever. warping analysis can be done with simple phase inversion with the utility plug-in.

Ah, and the reason I have time to dig so deeply into all this is because my Lady is out of house for several days and I'm working to wards mastering my tools at the moment before starting to make music again. Once I start making music I don't want to be bothered with all this anymore, but want it to work properly and grand!

NO, spend the time making music, or at least making sense.
FORGET 64 bit summing analysis until the Abes put out the white paper on it, I didn't think you had the beta anyway.