[Rob Watts]

Thanks a lot for the taps clarification.
I watched with a pleasure the entire RMAF 2017 video, trying as a layman to understand (more or less) the principles. Only one thing left me clueless - the section with the "pre-ringing is not bad". Especially the statement "the more pre-ringing, the more accurate the reconstruction is." From my personal experience, whenever i had a choice of filters, the sharp brickwall filter was the least desirable one, being the most fatiguing. Also the Ayre white papers talks about pre-ringing as time-smear and slow roll-off filter improving transient response. I would greatly appreciate if you could drop a few more words about this part of your presentation.

Pre-ringing - a long tap length filter has much more pre-ringing than a shorter one - that's when you use a non bandwidth limited impulse. But sampling theory states you must use a bandwidth limited signal. When you use a bandwidth limited impulse (like you would find in practice on your recordings) then the longer the tap length, and the closer it gets to an ideal sinc function, then we actually get less ringing. You should never use the example of an illegal signal to make justifications as to whether a particular filter sounds good or not. And the pre-ringing is essential to reconstruct the transients properly:

So how does lots of pre-ringing end up creating no nett ringing? Each sample of data has it's own set of ringing, and they are arranged such that when summed within the filter, the nett ringing cancels, thus reconstructing the original bandwidth limited signal, with no changes at all.

The sharp brickwall filters people talk about are actually very limited tap length half band filters, and have poor aliasing abilities. The interpolation filters primary job is to remove aliasing from the infinite images centred at integer multiples of the sample rate - as the aliasing itself creates timing reconstruction problems. Half band filters are poor in this respect, as the attenuation at FS/2 is only -6dB. On the other hand, slow roll off filters can have better performance than this - and a slow roll off or an apodizing filter that is say -30 dB down at FS/2 will sound better than a half band pseudo brick wall filter. But a brick wall filter, with a sensible attenuation at FS/2, will easily beat a slow roll off filter....

 

[rayl]

How much easier would life be if we can get the world to move up to higher sampling rates... But it would really take a widely adopted standard to force a long upgrade cycle in the recording / post-production / mastering chains.

Unfortuantely, I suspect it takes more than showing the world the benefits -- it takes a successful marketing push towards a new standard (for some other marketable benefit) which will incidentally have a higher sampling rate.

 

[514077]

How much easier would life be if we can get the world to move up to higher sampling rates... But it would really take a widely adopted standard to force a long upgrade cycle in the recording / post-production / mastering chains.

Unfortuantely, I suspect it takes more than showing the world the benefits -- it takes a successful marketing push towards a new standard (for some other marketable benefit) which will incidentally have a higher sampling rate.

Not positive, but I believe a fair number of recordings are already done in higher sampling rates, and then brought down to 44.1k. Which is why some music is released to download sites at those rates. But, there are also those companies who take lower bitrates and upsample and resell as higher bitrates. I.E. I bought a recording of Rapsody In blue and other songs at DSD128, only to read the release notes afterword, finding it was originally mastered at 96k. (right eegit) I read the fine print, now and buy at that rate.
In short: With all the modern digital harmonizing, pitch correction, plastic stars singing about the most inane garbage, I don't think the average music customer will ever give a crap about high-res.

 

[castleofargh]

or maybe not everybody is concerned about making the perfect copy of a painting when they know they'll only ever be able to watch it through dirty colored windows anyway. *cries in transducer, ambient noise, acoustic issues, ears not getting younger...

 

[Triode User]

or maybe not everybody is concerned about making the perfect copy of a painting when they know they'll only ever be able to watch it through dirty colored windows anyway. *cries in transducer, ambient noise, acoustic issues, ears not getting younger...

What are you trying to say? That you would prefer just to select a DAC on whether it makes music you like listening to rather than it's accuracy? I hope you are referring to a level matched double blind test there.

 

[castleofargh]

I was trying to point at the idea of weakest link in the chain, while using the worst analogy possible because I'm an addict. crappy analogies work for nobody but I just can't seem to be able to help myself. don't blame me, it's a sickness.

 

[Rob Watts]

I was at a concert in Swansea last Saturday (Prokofiev 1, Shostakovich Cello, Rimsky-Korsakov Scheherazade by the St. Petersburg Symphony) and because I had bought the tickets rather late, was sat further back (16 row) than usual. So the sound was a little softer and darker than usual - also I don't know what the Brangwyn hall is like, as it's not a concert hall. But even though the sound was not at it's best, it was still fantastic - particularly in my emotional response to the music, with almost constant spine tingles. That's the reaction I want from reproduced audio.

Another aside - my youngest son has just started playing the drums, and his school kindly let him have a drum kit, which we installed yesterday. What shocked me was how very different drums sound unamplified, particularly the pitch and timbre of the bass drum - it's very, very different in real life to reproduced.

Sure we want to get at the weakest link in the chain - but I am not sure where that actually is - and that's why I have always wanted to go from mic to loudspeaker with my gear, so I can know exactly where the electronic weakest link actually is. With the Davina ADC project, I will actually find out, for certain, what the weakest link actually is.

 

[analogmusic]

Hi Rob

In terms of looking back and forth (how your WTA filter works), I do find a big difference between Mojo and Dave in this respect, Dave simply is able to reproduce more of the beginning and ending of a note

Not sure how this works exactly, maybe that is the secret sauce of Blu2 and million taps - with regards to being able to look into 1.4 seconds on music.

But what happens is a revert tail is actually longer than 1.4 seconds? does it get ignored by WTA?

 

[musickid]

i actually graduated from the University of Wales as a mature student. Swansea is okay but as you move towards Carmarthen and surrounding areas the beauty of Wales really opens up. unrelated but thought i'd add it anyway!!

can someone kindly point me towards both of the video presentations. i can only find one on you tube.

 

[Rob Watts]

Hi Rob

In terms of looking back and forth (how your WTA filter works), I do find a big difference between Mojo and Dave in this respect, Dave simply is able to reproduce more of the beginning and ending of a note

Not sure how this works exactly, maybe that is the secret sauce of Blu2 and million taps - with regards to being able to look into 1.4 seconds on music.

But what happens is a revert tail is actually longer than 1.4 seconds? does it get ignored by WTA?

Yes beyond 1.4 seconds the coefficients are forced to zero by the WTA windowing function. Of course, the coefficients from the ideal sinc function is now much smaller than 16 bits; that's how I know that the reconstruction is now better than 16 bit accuracy, because the coefficients that the M scaler is ignoring are much less than 16 bits.

 

[musickid]

to fill a gap in my knowledge once davina is released would any potential dave owner wanting to add an mscaler still go for blu2 as this also incorporates a cd player? for example even though davina has mscaler tech in it is its ADC section aimed primarily at commercial recording studios or those audiophile types recording in their own homes. or is davina designed to replace blu2 and if not where does it fit into the consumer audiophile market? finally once other manufacturers move to similar fpga approaches can we expect future dacs to catch up where we will have a market full of dacs performing at chord's level where this becomes the norm in years to come or will rob's implementations always have the upper hand.

 

[x RELIC x]

to fill a gap in my knowledge once davina is released would any potential dave owner wanting to add an mscaler still go for blu2 as this also incorporates a cd player? for example even though davina has mscaler tech in it is its ADC section aimed primarily at commercial recording studios or those audiophile types recording in their own homes. or is davina designed to replace blu2 and if not where does it fit into the consumer audiophile market? finally once other manufacturers move to similar fpga approaches can we expect future dacs to catch up where we will have a market full of dacs performing at chord's level where this becomes the norm in years to come or will rob's implementations always have the upper hand.

Considering that even though the Blu mk2 and Davina will both have M Scaler tech the two are completely different products that I suspect you'd want to purchase one that will actually fit your needs, ie., are you recording a lot of audio or playing it back?

There's already a handful of companies using FPGA chips in their DACs, but they aren't using the same approach as Rob and also don't use Pulse Array DACs, or have the same simple analogue stage (simple as in not a lot of components). An FPGA is nothing more than a blank slate to program how you wish so fundamentally it all comes down to the designer, not the use of an FPGA chip.

 

[Mython]

There's already a handful of companies using FPGA chips in their DACs, but they aren't using the same approach as Rob and also don't use Pulse Array DACs, or have the same simple analogue stage (simple as in not a lot of components). An FPGA is nothing more than a blank slate to program how you wish so fundamentally it all comes down to the designer, not the use of an FPGA chip.

For one moment, there, I was wondering if you were about to call Rob a 'blank slate', Relic!

 

[514077]

For one moment, there, I was wondering if you were about to call Rob a 'blank slate', Relic!

If ever my brain becomes a blank slate, I want Rob to design my new software.

 

[musickid]

but how many headfi'ers record music? its market is obviously the commercial sector?