What Ableton IS KEEPING SECRET about Live's audio engine...

[massenmedium]

A digital null test or comparison cannot seriously be argued with. That should be clear. And I'm not saying that supposed differences in the sound of digital software do exist. But...

Just because we don't necessarily know of a mechanism by which this might happen doesn't mean that there isn't one.

Science often proceeds by investigating observations which for a time may be dismissed as not happening by the orthodoxy of the day. This happens all the time.

For instance in this case while it's clear that if there is a real phenomenon it does not occur in the digital domain, what if something was happening in the analogue or acoustic space that was related somehow to what was taking place in the digital realm?

[fishmonkey]

you would be able to make a 1 bit recording that sounded good as long as the sample rate was very very high.

Not sure exactly what diskowipe is saying there but as regards the above, that's sort of what you can do. It might even be the future of digital recording.

http://www.acousticfingerstyle.com/1_Bi ... ording.htm

Sampling theory as generally interpreted supposedly says that sampling at a given rate gives an accurate picture of the frequency content present up to half the sampling rate. Which is kind of correct but actually there are a few limiting factors which aren't always taken into consideration, including the quality of the necessary band-limiting (to prevent aliasing), and also phase differences that occur at intervals smaller than the sampling rate.

that's a completely different kind of sampling, where the meaning of each single bit is relative to the previous one, and you sample at a high frequency.

in the standard kind of sampling we are talking about, each successive sample is independent from the previous sample. so at 1 bit resolution, no matter how fast you sample, you are only encoding a string of sound on/sound off...

[Khazul]

you would be able to make a 1 bit recording that sounded good as long as the sample rate was very very high.

Not sure exactly what diskowipe is saying there but as regards the above, that's sort of what you can do. It might even be the future of digital recording.

http://www.acousticfingerstyle.com/1_Bi ... ording.htm

Sampling theory as generally interpreted supposedly says that sampling at a given rate gives an accurate picture of the frequency content present up to half the sampling rate. Which is kind of correct but actually there are a few limiting factors which aren't always taken into consideration, including the quality of the necessary band-limiting (to prevent aliasing), and also phase differences that occur at intervals smaller than the sampling rate.

What is referred to as one bit sampling above is not directly comparable to 1 bit PCM sampling - regardless of bit rate. If you used 1 bit PCM sampling at an infinately high frequency, you are basically just running the audio through a schmitt trigger yielding say 1 when signal >= 0 and 0 when signal is < 0 which will sound horrible.

The one bit encoding described above is instead done in two phases:
- the first part of the process is based thresholding of the difference between the analog signal and the previous digital sample of the difference - ie it is efectively a sampled difference error (AKA Delta-Sigma) signal rather than sampling of absolute values as in PCM sampling.
Unlike PCM where instantaneous sample values have a value in their own right (ie the voltage level at the time of the sample), the instaneous value in Delta-Sigma encoding mean nothing without the surrounding values. The output stream is a continuously variable pulse width signal and not data - ie you cant directly store this signal.

- The second part is where the variable pulse stream is sampled at a fixed sampling frequency - usually 64 or 128 x normal PCM CD sampling rate. This is also the point at which a 1 bit samplng encoder is used to yield storable digital data - ie 8 samples can be stored in a byte and written to disc/memory etc.

Most people have seen this kind of error/difference processing in image printing to old black dot matrix printers. Simplistically its vaguely similar to part of the dithering process we might apply when trucating from say 24 bit to 16 bit, except here the dithering is ongoing in the processing+sampling chain.


(Edit: fishmonkey - seems we both posted at about the same time )

[massenmedium]

It is still 1-bit recording though. You are encoding a 1-bit data stream at very high sampling rates. I mean literally, in those terms.

[LoopStationZebra]

Not sure exactly what diskowipe is saying

discowipe isn't exactly sure what discowipe is saying.

[fishmonkey]

It is still 1-bit recording though. You are encoding a 1-bit data stream at very high sampling rates. I mean literally, in those terms.

yes, but it's a completely different way of storing and processing audio data, and it's totally different to what diskowipe is (incorrectly) talking about.

[3phase]

A digital null test or comparison cannot seriously be argued with.

Thats the point.. IT HAS TO BE ARGUED WITH..


when people do such tests on single audiofiles they are just worthless..its like the tricks of snakeoil sellers...

The use of an obvious and logic looking fact.. +2 -2 = 0..sure thats true.. but we deal here with 1 box of apples minus one box of pies... is this 0 aswell ???

look closer to your holy manual.. there they say... no !, between the lines they state that there are artefacts !!
but.. only little artefacts below -120 db

so its actually not 0..

so what is it than? and does it sound good when beeing summed up on 36 channels?

and -120 db on which frequency? any? or just at a friendly 650 hz? and at 8 k its a total different picture?

" hey man..its just one spike and its still -94... so it don matters anyway.."

so a maybe 30 db spike on a nasty freq? how does that might add up on 36 channels?

THis is all speculation but based on standard situations and experiances when gpoing deeper into sound problems and when starting to do some serious measurement arrangements...

In short..

Dont trust any tests you havent manipulated yourself !!!


beside that
maybe L8´s sound quality is really allwright and as good as any other daw.. but only if warping is off..
and aslong the mysterious brown henke dither turns that into an unreliable thing you defently shouldnt try to master with ableton live...

and leave mixes of sensetiv acoustic recordings to other daw´s that ensure bit transparency and no samplerate conversions on the files...


For electronic music and the recording of plug in synth sounds the sound quality of L8 is defenetly good enough and a little samplerate conversion here or there might even sound fatter and more smoshy... does wonders on nord modular recordings...

therefore we love our brown henke dither.. but not allways...

[massenmedium]

I say it can't be argued with on it's own terms

Forget calling it a null test, say it's a digital comparison. You are comparing two data streams, in this case the final output from audio software. Either they are exactly the same or they are not. I really don't think there's any room for doubt there in terms of the digital information.

You also need to acknowledge that the actual output of that information at the point of digital to analogue conversion is not really something that the audio software does.

I think if you deny these facts then any analysis that follows is likely to be faulty or fallacious.

So IF you are to still maintain that some phenomenon could be occurring that might mean the SAME information sounds different to a listener when being produced by different software, I think you at least have to concede that it would have to occur outside of the digital realm and probably in acoustic space.

..

But, as I said, knowledge does very often proceed from observation followed by investigation and some people still claim to be observing something.

Now of course there may well be nothing to see here, as most maintain, but actually I do think that the claim that this is a certainty, and that all explanations absolutely reside solely in the area of human perception, rests on the idea that because there is no known mechanism by which such an effect could occur, then there can be no such effect. I'm not sure if this is enough, especially in the absence of an adequate test of what is going on in the air at the point it reaches a pair of ears. There's a bit too much of an inference going on there to make a claim of certainty.

It's a question of epistemology. I have a theory about this, it's partly tongue-in-cheek, maybe science-fiction, but hopefully also serves to illustrate this point. And heck, there might even be something in it.

[3phase]

I say it can't be argued with on it's own terms

Forget calling it a null test, say it's a digital comparison. You are comparing two data streams, in this case the final output from audio software. Either they are exactly the same or they are not. I really don't think there's any room for doubt there in terms of the digital information.

You also need to acknowledge that the actual output of that information at the point of digital to analogue conversion is not really something that the audio software does.

sure ..and if ableton live would be a just a stereo audio editor that null test would proove perfect bit transparency.. but only.. if the null is a real null.. just dont hearing anything in the speakers is not enough..

problem is..

Ableton live is anything else than an stereo audioeditor.. saddly.. i would prever a little more audio edit functionality...

Ableton live is as other daw´s allmost unlimited in track count.. at least much more tracks than i ever will use..
So mixing is an issue..
it has fx send..groups..automation... plugin inserts aso..
lots of points something can go wrong...

a null test on a sinngle file dont says much about the audio quality of a daw..

it just says that its not total crap...

actually the earlier versions of live failed in that audio editor null test...

they managed to get this introduction test right by now,. so they are qualified for the real tests.. and enter the ring to compete with the other pro daw´s

so please stop talking about this stupid null tests..

Ableton live is defently far beyond the point where such a test holds valuable information..

they are not the inferior ugly sounding duckling stage tool anymore..

They are maybe not the best daw or ahead of the others.. but they compete in that lique now

as you all know.. because you are convinced allready.


You never will find digidesign or logic or steinberg talking about null tests..

because you cant take a professional daw serious that has to point out that a single file can pass it unharmed..

I would wish ableton would update there fact sheet and bring it more to an adult state that is not so offensive by sounding like cheap propaganda..

[shimmy]

.

[3phase]

[massenmedium]

a null test on a sinngle file dont says much about the audio quality of a daw..

Right, but isn't the discussion about supposed differences in the way the same information sounds coming from different programs?

Typically what people mean by a digital null test is a comparison of the final output from the DAW before it goes to the converters.

[3phase]

a null test on a sinngle file dont says much about the audio quality of a daw..

Even if the test is in the final output that is going to the converters? It's the last time it's digital before you hear it.

That's what people typically mean by a digital null test. Forget that it's a null test, call it a comparison.

you really dont understand it? its a null test because it has null value on a daw..a digital comparison of daw´s includes a whole lot of differnt measuremets and measuring scenarios..

sure you compare the digital output. but the digital output of what? a single file with unity gain? and they manage to be the same than as any other daw?

ok ..cool.. when i park my car it is as fast as any other parked car... and can compete with the fastes parked cars in the world...

and that is the unbendable truth !!!

[Thomas]

[massenmedium]

you really dont understand it? its a null test because it has null value on a daw..a digital comparison of daw´s includes a whole lot of differnt measuremets and measuring scenarios..

sure you compare the digital output. but the digital output of what? a single file with unity gain? and they manage to be the same than as any other daw?

Yes I understand it.

This is usually a question of the way different programs mix and output audio. That's tested using a comparison of the final digital output. Sometimes this is referred to or performed as a 'null test'.

Of course it isn't the whole story, but that's not what's usually being asked is it?