advertisement


What does bit perfect mean?

J

JezmondTutu

pfm Member
So people hark on about the Wiim streamers being bit perfect.

Exactly what does that mean in reality, that it is pointless spending more than £100 on a streamer??
 
I take 'bit perfect' to mean that the series of sample values is carried from place to place without the sequence of bit/sample values being changed in any way. Its the basic requirement if you want to convey the info that defines the waveforms without loss or alteration. How well that can then be 'rendered' as analogue patterns to later stages is another matter. As is any problems with what may have happened between the source sounds in air and the series of bits being conveyed though the later stage.
 
I imagine it can mean a number of different things but the most common application in audiophile circles denotes a copy of a piece of music, usually from a CD to HDD or SSD, that is matched bit for bit. This will give you a perfect CD quality level copy (e.g. files types such as FLAC, WAV, AIFF etc), with a sampling and bit rate of 44.1KHz and 16 bit, i.e. the same quality as CD replay.

You can get much higher resolution music files that take those figures up to 196/24 but this would not be a bit perfect copy from CD.

The argument about how much difference a streamer makes in terms of overall sound quality is a perennial one here, seemingly taking over from where the debate about cables left off. The debate is not helped by the term 'streaming' potentially meaning a number of different things and the whole digital front end piece not being especially well understood or defined in the debate.

There are a number of things that will significantly change the quality of playback from a digital/streaming/music server front end; few of them are the actual data bits themselves; power supplies, jitter and noise all creep in a degrade quality. The better quality players, whether they are streaming from the internet or retrieving and playing from an SSD stored inside the machine, will all improve on those variables and, in general, sound better than the lower quality streamers.

This is the easiest thing in the world to test. Plug in your laptop (not a £100 streamer I grant you but probably about as poor as one), and compare it to a higher end offering and see what you think.

I have done this comparison many times including a week ago with a friend who had a £1000 Eversolo streamer that we compared to my (albeit significantly more expensive), LDMS music server. Streamed music from the internet via Qobuz into Roon into JRiver and then into his DAC all via my LDMS wiped the floor with the Eversolo playing it's chain of events. I also did the comparison by going from my Macbook pro, using an SSD containing bit perfect copies of my CD collection, playing via Roon (not into JRiver so using the Roon playback engine), and then moving to the LDMS. Again, the difference was night and day.

For clarity, the steps I'm referring to here are summarised as follows:

Music storage and retrieval - can be CD quality files stored on an external music server like Qobuz or Tidal; your streamer acts to identify these via a user interface and then retrieve the file, bringing it into your home. Alternatively this can be bit perfect copies you either create yourself from a CD archive and store on an SSD or files you have downloaded having puchased the music rights. Locally stored playback is noticeablly better than playback from a streaming internet source.

Music Curation/Interface - the user interface should be separated from the playback chain as in essence it should just delineate the way in which music is collected, curated and presented to you. The best and obviousy example of this is Roon, which is in my view excellent. I use Roon simply to navigate the music I have, stored for the most part locally (I have about 2TB of bit perfect music on my internal SSD drive in the LDMS music server), but you can also use Roon as a playback engine if you wish. The interface tells the server/streamer what you want to play and the OS on that device (often a version of Windows), goes and gets the file and makes it ready for play back.

Music Player/Renderer- the retrieval process presents the data to a music player which then renders it into a stream of data that can be passed to a DAC for conversion. Roon can do this, other streamers/servers will have alternatives built in. One of the common and well regarded ones is JRiver but there are many others. The player is just the software that makes the data into something a DAC can understand.

ALl of these steps have the opportunity for errors and noise to infiltrate meaning the bit perfect stream your started with, is no longer quite so bit perfect as it once was.
 
I think it has little meaning as we mostly use digital. In any case 99.99% of DACs convert to a 1 bit (or DSD if you like) stream to convert to analogue.
 
I’m not sure I know of any streamer which isn’t bit perfect. Being bit perfect is off-the-shelf stuff. As above, there is far more going on in a streamer than the straightforward conversion of ethernet packets/frames to a continuous bit stream without losing, adding or shuffling any, and the sonic differences amongst streamers lies here.
 
Thanks folks. I’ve certainly heard differences between both streamers and transports but many in the Wiim community seem to be blinkered in so far as their streamers are not perfect and therefore you cannot improve on that in terms of sound quality.
 
It is an all but meaningless term IMHO. It basically means a stream of data that was transmitted was received intact. That is obviously useful knowledge from an IT perspective, but it doesn’t suggest what was received was a ‘bit perfect’ copy of an original recording/master tape. I’m becoming more and more conscious of this since playing around with Tidal and Qobuz. Bit-perfect to what exactly? What format was the master created or digitised in? What has happened to it since, e.g. resampling, normalising etc? So many odd formats too, e.g. 24/44.1. No idea what that is for or where it ever existed in the wild! That’s before one gets to worrying about stuff like MQA, which still widely exists on Tidal, so is lossy CD resolution as far as I’m concerned as I haven’t a decoder.
 
... but used to be a 'thing' about 20-25 yrs ago when people were using cheap CD players as transports into 'better' DACs, vs the start of 'ripping' the data via a PC; and suddenly found things like the (iirc, Phillips 723... something like that) to output SPDIF data which had a relationship to what was on the disc best -described as ...tenuous.


Absolutely Not a thing these days, in fact once any kind of upsampling/DSP/Vol control/ redithering/Conversion to DSD/ low-bit conversion at stupid sampling rates is used (any of those choices, in several combinations of possible order!)

-being: the main benefit of modern dacs -

'bit -perfect' is out the window, by deliberate intent: because that way, even better results can be found.

13th Duke of Wymbourne said:
Darko would describe it as audiophile catnip
Click to expand...
I like that!
 
There are 3 aspects to think about.

1. Data. Transport bit perfection is good if you're happy with your DAC crunching all the numbers (for upsampling and in some cases volume control). Note some people use computer upsampling where they believe it will do a better job, in which case transport bit perfection isn't actually wanted. If you abandon bit perfection I suggest you be very sure you're making a good choice. As @Eoin mentions many DACs have several stages of internal processing, so computer upsampling is about replacing DAC processing with something better yes, but often this replaces a stage of DAC processing whilst other stages still happen.

2. Jitter. Async USB means no input jitter, that said USB circuits in the DAC may have their own inherent jitter albeit very low. S/PDIF inputs such as coax, TOSLINK and AES have timing combined with data so bear input jitter. TOSLINK is usually the worst for jitter. For this reason many DACs re-clock or re-sample input data using the local clock, thereby eliminating input jitter. Remember not even the local DAC circuits will be 100% jitter free, but local re-clocking makes it the lowest it can possibly be, potentially vanishingly low.

3. Electrical interference. Via wire connections, or simply radiated. A lower interference streamer is better. The interference may increase as data bit depth and sampling rate increases. Amongst the connection types, TOSLINK connection wins overall on this point. The bits/numbers won't change due to noise, this is more a problem for local DAC clock and analogue circuits.

Personally I use computer upsampling, Wi-Fi, a streamer on battery power and TOSLINK. This way there are no wire connections from streaming stuff to the signal path. I engage DAC local re-clocking so the input jitter is fixed. But as mentioned nothing's ever perfect and I don't claim such.

And what's audible is another bunfight.
 
Last edited by a moderator:
@darrenyeats All good stuff and all well argued, but the very reason many people choose not to use wifi (and therefore live with or deal with the noise issue by other means) is the noise generated by wifi receivers/transmitters themselves. Ditto TOSLINK optical-to-electrical conversion. While your logic appears impeccable, we have to ask ourselves why every single audiophile/music lover doesn't use wifi or TOSLINK as their connection of choice.

At your (3), people should experiment with wired and wireless to see which wins sonically with their specific streamer in their specific system.

Similarly, TOSLINK may win theoretically but there are a good few people out there who have rejected it on sound quality grounds. Something to do with bandwidth restrictions and/or something to do with the noise generated by the optical-to-electrical converter circuits.

If you've experimented with TOSLINK vs other methods, and with wifi vs wired, and found the very best sound quality your system can produce uses both then job done!
 
Yeah I have experimented. I've tried it several times with two different DACs. Also by toggling the inputs and confusing myself, and then picking which I like most without looking. I like TOSLINK. (Maybe the noise is more impactful than jitter levels, with my transport, a Touch.)

The new DAC allows me to toggle re-clocking. I did the same sort of blind test, only once though, and I picked re-clocking.

Someone else might prefer different, or maybe other equipment would lead to a different result for me. Also the Touch is outputting 24/192 (was 24/110.6) - this will generate more noise than 16/44, maybe 16/44 would lead to a different answer.
 
Despite what some others have said, bit perfect is important.

Simple example. You play a CD quality 44.1k/16b file on your windows PC but your DAC shows 48k/24 bit as the incoming rate.

That means Windows buggered around with your CD quality file and resampled it. Now depending on your point of view, resampling may or may not be beneficial BUT you dont want the Windows OS doing it as its resampling quality (compared to say sox or that within your DAC) will be poor.

Now if you DAC reports your incoming CD file as 44.1/16 then it means it hasnt been molested by some downstream process (i.e. is bit perfect) BUT that doesnt mean you are still bit perfect within the DAC.

Some DAC only support 48x rates (48,96,192 etc) and so will resample any 44.1x rate (44.1, 88.2 etc) to some 48x rate (maybe as high as 768k)

While the quality of this DAC resampling compared to say Windows will be much better, again some people prefer that the DAC doesnt do this.


Summary: bit perfect before your stream hits your DAC is important and within the DAC is personal choice.


NOTE: resampling is different from upsampling which is different from oversampling. Most DAC's (unless they are NOS) will oversample and so this cant be avoided but it isnt screwing with bit perfect.


Peter
 
Good point, you don't want your audio player to mess with the data unnecessarily. But once you do any filtering or oversampling then the bits don't perfectly match the original source and the same for digital volume control but, maybe, you shouldn't care at that point because the next step is to become analog, which doesn't have any bits.
 
13th Duke of Wymbourne said:
Good point, you don't want your audio player to mess with the data unnecessarily. But once you do any filtering or oversampling then the bits don't perfectly match the original source and the same for digital volume control but, maybe, you shouldn't care at that point because the next step is to become analog, which doesn't have any bits.
Click to expand...
Indeed and you want to reduce the number of times your raw input undergoes some transformation and so back to the OP's question about the WIIM... it doesnt change anything which is good but as you note (and I alluded to) all bets are off once you get a signal into the DAC.

Its nice to have DAC's with different filters even if you ultimately stick to a standard FIR. And if you have source files with differing sample rates (44.1x/48x) its nice to have a DAC with dual clocks cause that avoids a resample (but unless you go NOS you cant avoid an upsample).

Peter
 
ricko01 said:
Summary: bit perfect before your stream hits your DAC is important and within the DAC is personal choice.
Click to expand...

I agree partly, because you can upsample with a box or software before the data stream hits your DAC. I certainly do.
 
Jim Audiomisc said:
Where does that statistic come from?
Click to expand...
That’ll be from the ‘estimate’ I applied in my head in the 10 seconds I took to construct my post 😂.

What d’ya reckon yourself, Jim?

I’d thought that they pretty much all were single but excepting the early multi bit chips and the current low volume boutique modern multi bit DACs. 🫡
 
You must log in or register to reply here.


advertisement


Back
Top