LIVE dithering

[Michael-SW]

I thought that when rendering you could only choose 24 or 16...i'm not in front of it now, so I may be way off.

In the preferences, when you set the bit depth, that is for recording correct? My understanding was that all calculations in the audio engine took place at 32 bit float...

)

Well, my Live 4 can only render in 16 or 24 bit, but seeing the discussion above I guess 5 can do 32 bit too?

You set recording bit depth in the prefs, bit you also set the bit rate/freqeuency (44.1 kHz etc) there.

[headquest]

Well, my Live 4 can only render in 16 or 24 bit, but seeing the discussion above I guess 5 can do 32 bit too?

Yes.

[djsynchro]

You don't really need to render to 32-bit unless you have 32-bit files in your project

Yes you do the internal audio path (VSTs, summing, processing) is 32 bit.

[huffcw]

Here's a question - if your audio card is 24-bit (which most are) and everything in Live is 32-bit - the signal is converting to a lower bit rate on output - correct?

I think someone else mentioned it earlier...isn't there a need for something to smooth the 32-bit to 24-bit conversion in real time during playback (outside of actually rendering a file to disk)?

[TheAnimal]

Another question: when rendering @ 32 bit does that mean that Live is storing in floating point format?

[headquest]

OT but related, I see that Ableton Live has now been added to the infamous Sample Rate Conversion tests analysis site here:

http://src.infinitewave.ca/

Interesting (if controversial ) reading/comparissons!

[lunabass]

I think I'm understanding these graphs correctly. The ideal filter is a straight cutoff at 22.05k (the nyquist frequency) with the green line the actual filter curve?
The Ableton graphs look quite bad. In fact it's suggesting that it's allowing frequencies through above the Nyquist frequency that would produce aliasing frequencies and this = very bad.
Have I misunderstood the graph?

[lunabass]

By the way I was talking about the Pass band graph.

[rad30n]

hi, does anyone know the dithering process that occurs in LIVE when bouncing down a session to 24 or 16 bit mix and within the session itself, is there quatization happening?

it is interesting how live can handle different bit depth files within a session that is 32 bit which leads me to believe that there is some sort of quantization happening somewhere but theres no mention in the manual.

is dithering and quatization even happening on bounces and in sessions?


thanks

there such no thing as dither this is hocus pocus small boy penis people say make sounds big yes rutter no so long thin in irregular passage i feel today alcohol smell father in prison.

[huffcw]

I think I'm understanding these graphs correctly. The ideal filter is a straight cutoff at 22.05k (the nyquist frequency) with the green line the actual filter curve?
The Ableton graphs look quite bad. In fact it's suggesting that it's allowing frequencies through above the Nyquist frequency that would produce aliasing frequencies and this = very bad.
Have I misunderstood the graph?

Yes it does = very bad - and, if true, may contribute to people perceiving the sound of Live as inferior to other DAWs. Ableton should be focusing on fixing the sound quality - a top priority for the next release I would think. Although I think it would only come into play if the signal in Live is at a very high frequency.

[lunabass]

Although I think it would only come into play if the signal in Live is at a very high frequency.

As far as I understand it thats not necessarily true but I'm happy to be corrected
Alias frequencies are introduced into the audible band and can be worked out in the following way:
Sample Rate (44.1k) - the frequency input = alias frquency.
So IF for example a frequency of 30kHz (inaudible to us) was allowed to get past the anti-alias filter and the sample rate is 44.1k then 44.1kHz -30kHz= 14.1kHz which is well within our hearing range. This would obviously be very bad.

Note that I'm not saying the anti alias filter in Live does let 30k through it's simply an example.

Of course there could be something else that fixes this problem in Live that I'm not aware of but notice how Pro Tools etc ensure that their anti alias filter doesn't let anything past 22k through.

[headquest]

Yes it does = very bad - and, if true, may contribute to people perceiving the sound of Live as inferior to other DAWs. Ableton should be focusing on fixing the sound quality - a top priority for the next release I would think. Although I think it would only come into play if the signal in Live is at a very high frequency.

Well first of all, I have Live 5, Sonar 5 PE and Audition 2, and I really don't think that Live is "inferior" at all, even though Sonar has its much trumpeted 64-bit mix engine.

Regarding the charts, as I understand them, Ableton - and most other popular sequencers, such as Sonar, Logic and Cubase - is not particularly good at sample rate conversions. Nor for that matter are some of the specialised tools - if I understand the graphs correctly, Soundforge doesn't appear to work at all!! Generally though specialised tools such as Adobe Audition are better at sample rate conversion than sequencers are. So far, not exactly shocking news, then.

From a practical point of view though, I think it might be wise to record at 44.1kHz rather than 96 kHz (which I do anyway), in which case sample rate conversion simply isn't an issue anyway, assuming you plan to export at 44.1.

[Michael-SW]

If someone missed, the topic switched from bit depth conversions (dithering etc) to sample rate conversions.

Not the same thing...

[headquest]

Yes, that's why I prefixed my post "OT".

Although sample rate reductions and bit rate redutions are at least tentatively linked, if only because they are often done simultaneously (as in Live).

[raapie]

32 float is only useful when recording WITH effects. no soundcard is capable to playback 32 float, only 24 bit, so... it's a bit silly.

32 float also truncates into 24 bit. so 24 bit in general is a lot better I guess because the results are far more predictable.

but in the end: let your ears be the guide. if you can't hear dithering yourself, stop wondering about it and start makin' some noise