Previous Thread
Next Thread
Print Thread
Rate Thread
Page 5 of 7 1 2 3 4 5 6 7
Re: Ethernet vs. Digital Audio Out
BrenR #144688 01/04/07 09:18 PM
Joined: Dec 2006
Posts: 4
P
newbie
Offline
newbie
P
Joined: Dec 2006
Posts: 4
Quote:

I introduced jitter by hand into some WAV files (at a much higher rate than would normally appear) and let everyone give a listen to see if they could hear it. Consensus was that no one could, not even after burning to a CD then bringing it into their home theatre.




There's a lot of dissinformation regarding audio in general floating around and I'm more than willing to believe jitter is one of them. My issues with PC sound cards has more to do with their mixing or resampling etc than jitter.

That being said, I do have to ask how exactly you inserted jitter into a wav file. From what I've read, jitter is a syncronization issue on the digital to analog conversion end. If you take the digital stream and convert it directly to analog you get sound. If however, some of that data that is being converted is not received and converted with the intended timing, you get still get analog sound, but it will be off from what is is supposed to be. This should have nothing to do with the wav file file.

If you take a VOIP call, jitter there gets introduced because not all packets necessarilly get received in correctly timed intervals and you end up with a robotic sounding connection.

With an spdif connection the timing issues should be small, perhaps undetectable, but it would be in the timing of the conversion of the data of the wav file, not the data in the wave file.

Re: Ethernet vs. Digital Audio Out
packetloss #144689 01/04/07 11:15 PM
Joined: Dec 2003
Posts: 3,602
B
connoisseur
Offline
connoisseur
B
Joined: Dec 2003
Posts: 3,602
Quote:

I do have to ask how exactly you inserted jitter into a wav file.


Simply by changing at random, single samples. Selecting a single sample and entering RND (65536) into the numeric equation for each channel. That's how jitter will be received by the other end. Unlike SMPTE audio in broadcast or audio-video interleaving, you've got nothing else to go out of sync with with straight audio. You'll lose two binary words (16 bits per channel) and at the very worst at the other end, it'll completely randomly make up the sample. Best case scenario (and even the cheapest CD players do this), it will interpolate the signal if it can't read it.

It's not like you'll lose one bit in the stream and every future bit will be ROLed (rotated left) to fill the "hole"... if that happened, the rest of the track would play as static.

Bren R.

Re: Ethernet vs. Digital Audio Out
packetloss #144690 01/04/07 11:20 PM
Joined: Dec 2003
Posts: 3,602
B
connoisseur
Offline
connoisseur
B
Joined: Dec 2003
Posts: 3,602
Quote:

If you take a VOIP call, jitter there gets introduced because not all packets necessarilly get received in correctly timed intervals and you end up with a robotic sounding connection.


VOIP works via packets over UDP, and a traceroute of even a call from me to my next door neighbour will show the data packets hopping around the province a few times before ending up at their end. It's a much different animal than digital audio within a building along a wire or fibreoptic cable.

If I was sending my audio stream of Fresh Fruit for Rotting Vegetables over "teh Intarweb" to Toronto and back before it made it to my speakers, I'd probably get some of the same issues.

Bren R.

Re: Ethernet vs. Digital Audio Out
BrenR #144691 01/05/07 01:34 AM
Joined: Dec 2006
Posts: 24
P
hobbyist
Offline
hobbyist
P
Joined: Dec 2006
Posts: 24
Quote:

Simply by changing at random, single samples. Selecting a single sample and entering RND (65536) into the numeric equation for each channel. That's how jitter will be received by the other end. Bren R.





Randomly changing parts of the data stream isn't the same thing as jitter (Although if you damaged every single sample and no one could tell then they aren't going to notice jitter either). That would be simulating read errors and lost data. Jitter is defined as a time based instability. This instability is going to occur on every word from every single sample and the shift is going to be random (sometimes faster sometimes slower.) Your not losing and interpolating the data, your just not reconstructing the sound waves with the same timing (which will equate to shape) they were encoded with. An extreme example of this affect would be to take a turntable that allows fine tuning of the rotation speed and to keep making random small adjustments to the speed.

This is the reason all DAC's require an oscillator crystal. The crystal establishes the time sync with the audio samples for reproduction of the analog sound wave from the digital sample. When transfering the digital samples over spdif, the clock sync is not transmitted and that is where the jitter is introduced.

What really matters of course is if you can you hear the difference. These fluctuations are probably so small that you wouldn't notice, not to mention since the timing instability is random it's not like you would hear the same imperfections (assuming you heared any) if you played the music over and over.

Re: Ethernet vs. Digital Audio Out
packetlosss #144692 01/05/07 01:58 AM
Joined: Aug 2006
Posts: 213
local
Offline
local
Joined: Aug 2006
Posts: 213
Hello

I agree with 2 points recently made.

VIOP can not be compared to jitter. Its my understanding UDP is a stateless protocol and there is no packet delivery confirmation transmitted back to the originator. damaged packets are rerequested, hence sometimes not arriving in time and utilized out of order. the issue mentioned regarding VOIP could be due to the compression used, or other reasons. Thats not to say that jitter couldnt occur in the dac process of encoding the audio data, but the fidelity is so low, i cant see it mattering. I believe phones lines used to have a tap/load_coil/I_forgot_the_technical_term on them to keep the bandwidth to 3khz, and I am guessing a phone line sounds better than VOIP.

Jitter is not randomly changing bits. What I read agrees with what was posted about how it is related to the timing of the bits, not the data within the bits. I dont have a better understanding of the transmission of digital audio though. I also agree that you cant distribute wav files to test jitter. each individial setup that plays the wav file would be effected by jitter differently. hadnt really thought about the randomness of the same machine playing the same track over and over.

with that all said, I am learning as I go and find this an interesting topic.

Randy


Axiom M80, Ep600, Qs8, VP150, Crown XLI 2500 , DBX Driverack PA2, Focusrite Scarlet 2i4
Re: Ethernet vs. Digital Audio Out
packetlosss #144693 01/05/07 09:35 AM
Joined: Dec 2003
Posts: 3,602
B
connoisseur
Offline
connoisseur
B
Joined: Dec 2003
Posts: 3,602
Quote:

Jitter is defined as a time based instability. This instability is going to occur on every word from every single sample and the shift is going to be random (sometimes faster sometimes slower.)


Please, explain further... you're suggesting that it would make it halfway between beats of a crystal oscillator, and that would cause the bits in the stream to shift left or right?

I'm kind of confused on what exactly you believe happens... besides a vague "time instability".

Bren R.

Re: Ethernet vs. Digital Audio Out
BrenR #144694 01/05/07 03:11 PM
Joined: Dec 2006
Posts: 4
P
newbie
Offline
newbie
P
Joined: Dec 2006
Posts: 4
jitter
A flicker or fluctuation in a transmission signal or display image. The term is used in several ways, but it always refers to some offset of time and space from the norm. For example, in a network transmission, jitter would be a bit arriving either ahead or behind a standard clock cycle or, more generally, the variable arrival of packets. In computer graphics, to "jitter a pixel" means to place it off side of its normal placement by some random amount in order to achieve a more natural antialiasing effect.

For clock jitter, there are three commonly used metrics: absolute jitter, period jitter, and cycle to cycle jitter.

Absolute jitter is the absolute difference in the position of a clock's edge from where it would ideally be if the clock's frequency was perfectly constant. The absolute jitter metric is important in systems where a large number of clock sources are trying to pass data to one another (eg. SONET).

Period jitter (aka cycle jitter) is the difference between any one clock period and the ideal clock period. Accordingly, it can be thought of as the discrete-time derivative of absolute jitter. Period jitter tends to be important in synchronous circuitry like digital state machines where the error-free operation of the circuitry is limitted by the shortest possible clock period, and the performance of the circuitry is limitted by the average clock period. Hence, synchronous circuitry benefits from minimizing period jitter, so that the shortest clock period approaches the average clock period.



To really explain what is happening, we need to take a look at what digital audio really is. Analog sound is a wave. I'll assume everyone has seen graphs of sound waves and knows that what one looks like. Eseentially a sound wave can be thought of as a signal which OVER TIME can vary continuously in amplitude.

To get a digital representation of this signal samples are done at different intervals along the wave. For CD audio this sampling is done 44,100 samples a second with 16 bits of resolution. DVD audio is between 96,000 and 192,000 with up to 24 bits of resolution. What this means is we are taking timed samplings of this original analog sound wave and making a digital approximation of it. There is of course much discussion with regards to how much sampling is needed to make the interpolation of the original analog sound indestiguishable from the original but that has nothing to do with jitter.

What a DAC does is it takes this digital rendering of a sound wave and it reassembles the analog sound wave. To do this properly it needs to have a clock sync so it can establish the correct interval of the samples.

Think of it like a plotter. The timing is the rate at which the paper is moving under the pen. If the paper isn't moving at a constant and determined speed, you are not going to reproduce a proper graph. It will either end up drawing the wave over a longer duration of time than the orignal or a shorter duration.

This is the time instability which causes jitter. Without a proper clock that sound wave can't be reproduced with accuracy. This is why all DACs and CD players have an oscillator crystal.

My VOIP example was a rather extreme example of the timing issue. Jitter for VOIP represents the same problem, it just happens for a different reason and on a different scale (large chunks of the wave are transmitted - so it's not individual samples that are out of time sync). Some packets make it there fast, some slow, some don't make it and have to be resent, but essentially the sound can't be reproduced with proper timed accuracy.

Last edited by packetloss; 01/05/07 03:22 PM.
Re: Ethernet vs. Digital Audio Out
packetloss #144695 01/05/07 08:10 PM
Joined: Dec 2003
Posts: 3,602
B
connoisseur
Offline
connoisseur
B
Joined: Dec 2003
Posts: 3,602
I had to go back and reread what the original question surrounding this was.

And to be clear, jitter in regards to networking - I'm not really clear on, besides a basic understanding of transport protocols, once the IT bar gets raised beyond basic networks, I spread Chee-zee Poofs around my desk and club the IT guy that gets closest and force him to do my bidding.

Digital audio I've been working with for a while. From the stone age to now.

Quote:

What a DAC does is it takes this digital rendering of a sound wave and it reassembles the analog sound wave. To do this properly it needs to have a clock sync so it can establish the correct interval of the samples.


Absolutely correct.

Quote:

Think of it like a plotter. The timing is the rate at which the paper is moving under the pen. If the paper isn't moving at a constant and determined speed, you are not going to reproduce a proper graph. It will either end up drawing the wave over a longer duration of time than the orignal or a shorter duration.


Which is why digital audio is reframed every frame to the internal oscillator. Again, the part I'm unclear on is how the jitter is supposed to let the stream run long or short with regards to time. If you could shed some light on that - the actual practicality of what occurs to disrupt the waveform - do you suggest that some bits pass other bits on the wire (or in the fibreoptic bundle) to arrive at the receiver before others? That some travel slower than others and cause a traffic jam on the wire? That the sending unit somehow loses time and starts firing the info at a non-regular rhythm? Or that both are like drummers and each ones' oscillator is off-beat with the other, and one is receiving on the upbeat while the other is sending on the downbeat?

In home audio, there are sometimes issues with digital audio, my DVD player transport is sometimes slow in switching over chapter lines, and the data becomes corrupted. My receiver will switch from Dolby Digital or DTS ES 6.1 or whatever it's decoding and revert to the standard DSP for the DVD connection (Neo 6 Cinema in my case) when it loses any part of the data. This happens maybe once a movie and only happens with this one DVD player. It's a bit of a PoS (curse you Toshiba!)

And on the other side of things, something I go back to a lot. What do the pros use? Whatever the guy actually cutting the movie, TV show or sporting event is using is probably more than good enough for the home user. And what is that standard? AES/EBU, SPDIF's big brother and nearly identical except for voltage (AES3 is 3-10v, I think SPDIF is <1V), connector (balanced or unbalanced XLR or BNC vs "RCA"/whatever that little toslink connector is called) and some of the data transfer.

It's really a great system, well designed, flexible and well thought out by the Audio Engineering Society and European Broadcasting Union.

Bren R.

Re: Ethernet vs. Digital Audio Out
BrenR #144696 01/05/07 08:21 PM
Joined: Aug 2006
Posts: 213
local
Offline
local
Joined: Aug 2006
Posts: 213
Quote:


Digital audio I've been working with for a while. From the Stone age to now




circa 1992 my stone age
and another shot showing CS800 amp at bottom, used to get drunk and crank it, was on middle floor of apartment building, neighbors never smiled at me

I could do a now photo , but it would be so boring.

sorry to drift off topic, but when you mentioned stone(d) age, somehow it rang the little bell in my head


Axiom M80, Ep600, Qs8, VP150, Crown XLI 2500 , DBX Driverack PA2, Focusrite Scarlet 2i4
Re: Ethernet vs. Digital Audio Out
Jim_Perkins #144697 01/05/07 10:59 PM
Joined: Dec 2003
Posts: 3,602
B
connoisseur
Offline
connoisseur
B
Joined: Dec 2003
Posts: 3,602
That a Roland stage piano?

Bren R.

Page 5 of 7 1 2 3 4 5 6 7

Moderated by  alan, Amie, Andrew, axiomadmin, Brent, Debbie, Ian, Jc 

Link Copied to Clipboard

Need Help Graphic

Forum Statistics
Forums16
Topics24,940
Posts442,457
Members15,616
Most Online2,082
Jan 22nd, 2020
Top Posters
Ken.C 18,044
pmbuko 16,441
SirQuack 13,840
CV 12,077
MarkSJohnson 11,458
Who's Online Now
0 members (), 145 guests, and 4 robots.
Key: Admin, Global Mod, Mod
Newsletter Signup
Powered by UBB.threads™ PHP Forum Software 7.7.4