Is WaveLab 9 based on a 32-bit or 64-bit float engine? Looking for an engine that can capture, store, and deliver 32-bit integer (clean) files, such as AIFF32 and WAV32. A 32-bit floating point engine will tend to lose some of the top 8 bits. Tks.

The 32 Float engine is in bit depth really good, in 44.1 Khz more then enough for audio. The 64 bit engine is supposed to be better, but it will be only very slightly. For me i would prefer to stay with 32 bit Float.

For those who say higher bit rates and higher sample rates are pointless - I suggest a little humility here. Can you make “Wavelab”? Can you make “Cubase”? Can you make Weiss digital processing equipment? Very few people in the world can do these things… despite that, do you really think these people are “stupid”? That they somehow don’t know what they’re doing and just add pointlessness to everything they do? Or are they more likely striving for the absolute pinnacle of excellence - even if it’s only a 1% improvement? Anyway - with that in mind, I have to say - I have done extensive testing with both higher bit-rates, and higher sample-rates (including 192khz, DSD, etc). While I feel “recording” at higher rates offers VERY little benefit (but “possibly” some), I have come to believe that some digital “processing” can benefit from these higher rates as well. In the context of a software like Wavelab/Cubase/ProTools/etc, that “processing” occurs MANY MANY times throughout a mix… every single fader deviation on each and every track from 0db? Digital Processing. Every single plugin? Digital processing. Every summing occurrence? Every envelope? Digital processing. Work at low rates if you want, but don’t get too arrogant about it. There are others in the world who may disagree with you, and they are not just the “stupid” bottom-of-the-barrel type people. Anyway… just some thoughts on the subject. We all need to trust our instincts and our hearts when making creative, critical decisions… just as the engineers who make our favorite tools do. Maybe put at least a tiny bit of trust in some of the people who make these incredible tools that we use every day. -Todd

Yes, to agree with Toader in a shorter manner: Just because you can’t hear it, doesn’t mean it’s not beneficial to the end result, and now the end user result after lossy encoding. 44.1k is technically enough for the human ear, but I personally feel that working at 96k allows my digital and analog tools to perform better and ultimately achieve a better end result. Then I use quality SRC (Saracon) to reduce to 44.1k and/or 48k as needed. I would welcome a 64-bit audio engine from WaveLab but I don’t think WaveLab is behind in adding this. REAPER has a 64-bit audio engine but since RX6 is still 32-bit float and I heavily use RX6 as REAPER’s external editor, I still save processed audio files as 32-bit float rather than 64-bit.

This has nothing to do with arogance, but with science versus marketing. Try some articles by Dan Lavry (not the average audio nono), in which he suggests, backed by data, that 192 kHz is possibly worse than 96 kHz - especially in AD converters. I’m not even going into the massive dynamic range 32bit float has, and to which 64bit float adds nothing useful - simple maths. Marketing has its own rules though: ‘If everyone else has it, so must we’, and ‘If 96k is so much better than 44.1, can we go and double that?’.

Although Dan Lavry discusses the optimal sample rate for “recording”, he doesn’t speak much about “processing” at higher rates - but even he recommends 88k or 96k. Here is a quote from one of his papers: At 60 KHz sampling rate, the contribution of AD and DA to any attenuation in the audible range is negligible. Although 60 KHz would be closer to the ideal; given the existing standards, 88.2 KHz and 96 KHz are closest to the optimal sample rate.. http://www.lavryengineering.com/pdfs/lavry-white-paper-the_optimal_sample_rate_for_quality_audio.pdf Anyway, regarding sample rates, I know what I hear. Regarding 64-bit processing, the designers have decided to add it for some reason. I’m assuming they’re striving for excellence. I look forward to testing to see if I can actually tell a difference

My guess is that 64 bit float processing is more about systems compatibility and future proofing than “sound”. It’s unlikely to make a difference between whether the record goes platinum or not. I work at 96kHz now mainly because it’s typically the deliverable for MFiT and for some processing (eg DMG EQ which does not apparently upsample internally) there is a case to make that it also “sounds better”. I do accept that a lot depends on your SRC in terms of the subjective test of what it sounds like at 44.1kHz.

the only thing that is relevant to safety is monitoring volume, this has NOTHING to do with dynamic range or 64bit accuracy

The interest in 64-bit float has nothing to do with “headroom” or “dynamic range”… Pros: No need to convert between 32-bit and 64-bit float: some plugins require 64-bit internally for their computations. Working natively in 64-bit avoids this conversion, giving a small performance gain and no precision loss between successive 64-bit plugins. Better precision when mixing audio signals, explained in detail at the end of this message. If audio devices ever move beyond 24-bit precision, 64-bit float will become necessary, since 32-bit float only really offers 24-bit precision. IIR filters (biquads, shelving/peaking EQ), where the output feeds back into the computation sample by sample, accumulate errors over time, and this effect is worse at low frequencies. Long convolution or reverb tails, where truncation noise can build up over thousands or even millions of samples. 64-bit float also delays a technical issue called “denormals”, extremely tiny values very close to zero. These can sometimes make a CPU slow down sharply if a plugin doesn’t handle them properly. Because 64-bit float has more precision, it takes much longer for values to shrink down to that problematic range, so this kind of slowdown is far less likely to happen.. Cons: It requires more memory, which can hurt performance since more data has to move around. But once a sophisticated plugin is involved, that plugin is likely to become the real bottleneck rather than the memory overhead. So this is really only a “relative” con. 64-bit CPU instructions run just as fast as 32-bit ones, since CPUs today are natively 64-bit. However, many instructions are actually faster in 32-bit float, because the CPU can process two of them in parallel (via SIMD) in the time it takes to process a single 64-bit instruction. Now, here’s an explanation of 32-bit vs. 64-bit float in the context of mixing. While 32-bit float effectively offers 24-bit precision, 64-bit float offers 53-bit precision, a dramatic increase. I can illustrate this with simple grade-school math (this is only an analogy of what actually happens): Suppose samples can only take values like 0, 1, 2, 3, 4, 5, and so on. Start with a sample whose value is “3.” Apply an audio gain of “divide by 2.” The result is “1.5,” but this value isn’t allowed, so it must be rounded, let’s say it becomes “1.” Later, apply a gain of “multiply by 2.” The sample becomes “2.” Result: we started with “3” and ended up with “2,” even though the two gains should have exactly canceled each other out. When this kind of loss happens repeatedly, as in complex mixing, the errors accumulate. The effect isn’t dramatic, since some errors are randomly offset by others (rounding up vs. down), but this compensation is itself a form of noise, a kind of “digital fog.” 64-bit float processing pushes this digital fog well below the 24-bit domain, resulting in a cleaner signal by the end of the audio chain. The difference between 32-bit and 64-bit, then, comes down to “audio definition”, assuming your ears are sensitive enough to notice. But that’s a separate discussion!

Thanks for the great explanation PG. I wonder if this is why years ago people thought that SoundBlade by Sonic Solutions had a better playback engine sound? I had a feeling that the reason would be more about the bigger picture. Just because you can’t hear something directly doesn’t mean it won’t have a bigger affect down the line adding some “digital fog” as you say.

Hi Philippe very good explanation. When you go for the highest quality, you have to go to 64 b. We work with acoustical music and there is it a must to try to reach the highest result. Wavelab could be a very high quality DAW

Steinberg Forums

32-bit or 64-bit?

WaveLab

Rat November 19, 2017, 7:02am 27

Thank you for the explanation PG!