Home Digital Drive

Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it.

Re: I thought this was "Mythbustered" already??

"thought that the difference between synchronous and asynchronous was that:

-synchronous used a single (master) clock and either the receiver or the transmitter "slaved" to the master
-asynchronous is when receiver and master have independent clock sources, and an ASRC (asynchronous sample rate converter) was used in a PLL (Phase locked loop)."

That's partially right. Synchronous in this context is the input data being linked to the output data, and both the input and output streams share a common clock.

"So, this means that 44.1khz to 192khz can be synchronous OR asynchronous just as 44.1khz to 88.2khz can be."

If the stream is first 320x oversampled and then 147x downsampled, it can become a "synchronous" conversion. But aside from DSD-based playback, I don't know of a player or DAC that actually does this.


"When resampling to even integer numbers, only 1/2 of you new sample set is "generated", where with non-even integer numbers the ENTIRE sample set is generated."

When the conversion is to an integer multiplied oversample rate, the new sample set is coupled to the raw data. The values are convolved with the FIR function to determine the interpolated values.

When the conversion is to an non-integer oversample rate, an extremely high oversample rate is initially used, to create data whose samples are close enough together for the output to grab the nearest sample. But there is both intrinsic jitter due to the samples not being infinite, and a noise component due to the jitter between the input and output clocks resulting in *amplitude* errors.

"I'm pretty sure this is right. My Behringer DCX2496 has a CS8420 ACSR in a PLL - it resamples everything from 44.1 to 88.2 to 96khz Is it operating synchronously for 48khz but asynchronously for 44.1 and 88.2? Nope. It's asynchronous for ALL sample rates because it's an ASCR in a PLL."

That's how most ASRCs work. This is why going 88.2 kHz with an ASRC may not necessarily do what a lot of people would expect. If it somehow switches to a true synchronous mode, where the ASRC becomes dormant, only then would 88.2 become the better option.

"If it had a wordclock output and I could slave my soundcard to it, then it would be synchronous for all sample rates."

If it goes by the DSD frequency, true. But only in the context of frequencies of either multiples of 44.1 or 48. Otherwise it cannot be made synchronous by a single clock.

"In any case, I think it is just sufficient to say 'resampling to non-integer multiples' as opposed to 'asynchronous versus synchronous'."

Maybe, but if the 320/147 algorithm is used, the noise from "independent clock jitter" would go away. So I mean "asynchronous", in regard to the sonic artifacts it would cause. Even in the case of "integer multiples". The problem is the clocks being independent.

"Myself, I'm leaning towards NON-oversampling after what I've heard it do for PC audio, even using the famous Secret Rabbit Code upsampler. Oversampling is another ballpark and does (AKAIK) use even multiples of the sample rate."

Non-OS works great with small-scale music. If the music gets complex, the modulation components take over, and the sound becomes "congested". I favor a "time-resolute" digital filter, such as the Lanczos3 convolution function.


This post is made possible by the generous support of people like you and our sponsors:
  Kimber Kable  


Follow Ups Full Thread
Follow Ups

FAQ

Post a Message!

Forgot Password?
Moniker (Username):
Password (Optional):
  Remember my Moniker & Password  (What's this?)    Eat Me
E-Mail (Optional):
Subject:
Message:   (Posts are subject to Content Rules)
Optional Link URL:
Optional Link Title:
Optional Image URL:
Upload Image:
E-mail Replies:  Automagically notify you when someone responds.