cute70.jpg
CuteStudio Ltd.

Getting the digital music out from SeeDeClip4 and into your HiFi

SeeDeClip4_LDR320.png


The idea of this page is to discuss the various ways of getting the music into your HiFi amplifier in the most accurate way possible.

There are a few easy ways to ruin the fidelity of the signal - as easy as lowering the volume control on the laptop - and the myriad of choices for DACs and USB/Optical mean that you may which to look for specific specifications: such as support of 88.2k, internal upsampling, rate conversion and optical or AES support.

The key to success is to be consistent in protecting the quality of the music signal. Too good is not a problem, but after the mastering damage done by the record company and our hopefully adequate repairs to that damage we don't want to rsik further degradation of a process that would be simple and accurate if we'd stuck to mild standardised compression, a ban on clipping and a 24bit datapath.

If we still fail to get a good waveform of our favourite music we can take comfort in the fact that inside the record company there rests a perfect 24bit unclipped master of what we wanted to hear, the information hasn't been lost for ever, it's just we cannot access it.

The DAC

Laptops and tablets have DACs built in, the quality of which can vary although some are very good. Some external DACs are driven straight from USB and a common DAC is the video surround sound decoder.

From computer to DAC

Ideally you'll be wanting an optical signal to the DAC, or professional AES connections. Optical has the advantage of isolation being complete and it always seems to work well.
Some laptops and PCs have an optical output built in so check for that first!

USBtoOptical.png   USBXMOS.png

Various adapters are available to go from USB to optical to drive the DAC. Make sure the one you buy can actually do 88.2k, many appear to skip over that rather important sampling rate. You'll want it to do 24/88.2 which should be quite attainable.

If your DAC will accept 88.2k at 24bit that's the best thing to feed it, the 88.2k gives it nice easy filtering and the 24bit ensures any levelling etc along the way shouldn't make any difference.

It's important to bear in mind that most DACs will be 18-20bit so 24bits may seem excessive but better too many bits than not enough.

If you don't have a separate DAC you may wish to consider a pro-audio one. They are often integrated into equalisers too - for instance an Ultracurve gives you both 64 band + digital parametric EQ, automatic room EQ plus an excellent DAC for a reasonable price. It also has a GUI menu so you can check bit depths, sample rates and the tonal balance. There are also various mods that can be done to enhance them.

ultracurve.png


Some advantages over domestic audio of boxes like this is for instance baffle step compensation can be dialed in with adjustable frequency, gain and Q/sharpness giving some real improvements right off the bat.

Bear in mind some of the connectors can differ from the inferior domestic audio ones.

Here's Thorsten Loesch's take on DACs which is well worth a read:
Thermionic Valve analogue stages for Digital Audio


Starting out: Bit depth

Lets first move rapidly to digital resolution: is 16bits enough?
The 16 bits at 44.1kHz was born in a time when many considered 20/48kHz was the minimum acceptable, but 30 years later we have what we have, indeed many HiFi buffs vigorously defend this format so it's here to stay in HiFi while the rest of the world moves to higher bits and sample rates.

We strongly recommend you read this excellent discussion of DACs, bit depths and sample rates:
  1. Mountains and Fog: the Sound of Digital Converters, Part 1 by Lynn Olson
  2. Mountains and Fog: the Sound of Digital Converters, Part 2 by Lynn Olson

The defence for the 16/44.1 format is:

"If perfectly mastered, dithered and upsampled on playback 16/44.1 is good enough for our ears"

Apart from various listening based studies contradicting this other more concrete problems are:

  • Many computer programs (including by necessity SeeDeClip4) digitally reduce the levels from the waveform and effectively re-master the track.
    I.e: By declipping the waveform (adding peaks onto the waveform) we suddenly have 17bits worth of waveform to output.

  • The modern 16bit dither is often 'shaped' too which is good for playback but plain old triangle dither is best for further processing.
    Dither is also an arguably poor substitute for the simple, immediate accuracy of 24bit.

    Fans of dither fail to account for its inherent statistical nature, for DC, slow and periodic waveforms it has good accuracy, but dither is unaware of musical content and those short transient sounds are still locked into the 16bit grid regardless of the lowest bit's dance with dither.

    By definition dither operates in a more timely manner at higher bit rates, so 16/88.2 is more accurate than 16/44.1, but less accurate than 24/44.1. The shaped dither is designed as a high frequency noise (rather than the traditional broadband noise) to allow higher effectiveness over time and a less instrusive noise.

    The very fact that the dither shape influences the sound of 16bit is the biggest clue that 16bits is not an adequate HiFi format. If one dither sound different than another who's to say which one is better? How do we know that's the best?

  • After pointing out the limitations of dither it's important to come back to the big picture and note that many CDs are mastered about 6dB louder than they should be so while that gives then at extra bit of detail, it alway throws away the most important bit at the top to clipping. Therefore our real problem is not usually one of detail but the other end - one of clipping. I.e. many 16/44.1 tracks are effectively 17 or 18bit tracks with the most important bit or two compressed and clipped out of the waveform.

    Whether this over-mastering is because the CD really needs 18bits and people are over-compensating - or whether is is just a race for max volume in the car we need to keep perspective on the big losses of information at the top bits as well as the tiny losses at the bottom that are prescribed the dither cure.

  • If you plug a DAC into a PC and don't set the volume on the PC to 100% you'll again get a digitally reduced waveform with loss of resolution.
    You can hear this for yourself if you are using a digital link, turn the PC level down, your amp up and listen.

There is also the question of 'why stick to 16bit?' which is difficult to answer. We now stream HD video on demand, have 1TB disks that are cheap, the FLAC system of open source lossless compression and can buy 24/96k DACs online for peanuts, yet serious HiFi men are still over-paying for slow, coarse equipment compatible with an obsolete 30 year old standard that wasn't good even back then.

Even SACD uses DSD system instead of the superior, simple, effective, 24bit PCM that over the past 30 years of the CD has become commonplace, cheap and accessible everywhere outside of the HiFi industry.

HD audio tracks: still victims of the loudness war

You can buy HD digital tracks today but many appear to still be clipped and some are said to be upsampled from 44.1 - which you can do yourself for free! If you are seeking original waveform improvements your best sources are:

  • Mastered For iTunes
  • Reputable places like Linn Records which give better (higher!) fidelity.
  • Finding older CDs with less 'loudness' or imports from Japan and mainland Europe.


The solution to loss of detail from levelling and processing is simple: Switch both SeeDeClip4 and the PC to output in 24bit. The next-best solution is 16bit with SeeDeClip4's HFII shaped dither.

Where to upsample

Non-upsampled 44.1k has a hard quality to the sound. Filtering directly from 44.1k must:

  • Restrict the bandwidth below 22.05kHz, 20kHz is often the stated perfect aim, losing 2.05k of bandwidth.
  • Anti-aliasing products will be present, disturbing the treble (distortion).
  • These filters will upset the phase of the treble and will often have ripples in their response (distortion).

Upsampling removes these errors, releasing a natural, focussed 'liquid' sound that is more accurate and easier to listen to.
Another point often missed is that upsampling reduces the step size between samples: so the frequencies to filter are not just higher but also quieter, from 88.2k to 22.05k the filter has an easy job and will need far fewer components.

A good place to upsample is on the PC/tablet (especially if you are stuck in 16bit). In Windows the sample rate and depth may be fixed in the 'properties' setting for the audio driver, so throw what you like at it from the app but it's going to be converted by the driver and output at the fixed rate set in the properties. For this reason I'd suggest setting this to 88.2k if most of your library is CD standard 44.1k, or upsampling externally.

Be aware that the sample rate your app outputs may not be the same as the one coming out of the USB and into your DAC, a pro-audio box like the Behringer Ultramatch is an invaluable tool as it will tell you the bitrate (and depth) coming out.

Your listening chain vs the Loudness War and Digital Music Streaming

Getting a good signal into your amplifier as it left the studio is as we have seen a difficult process, hindered by the insistence on using a low sample rate 16bit process. This process - possible because of it's shortcomings - is usually severely compressed, often clipped and relies on the delinquent dither method, such a shame when we have fine 19bit instant conversion DACs sitting there at home, waiting for a signal that rarely arrives in good shape..

The weakest link in the chain is the record company mastering and distribution pinch point.
digitalmastering.png

At least with SeeDeClip4 you can see the rating of the song to make your choice, see the waveform you are getting and start off your part of the chain with a better quality waveform than most of the HiFi world.

Switch your system to 24 bit to allow the detail and reconstruction to remain intact through levelling and your playing device's audio driver.
Upsample somewhere, just make sure the raw DAC doesn't have to convert 44.1k directly however you do it.
Your sonic mission is to ensure the entire digital audio path stays at a constant, high bit depth and sampling rate to preserve what detail and accuracy we have.

Summary


  1. Decide what's the best bit depth and sample rate you want to use.
    24/88.2 is recommended. Check sure all the devices in your system are compatible.
  2. Setup the app to generate the required bit depth and sample rate. In SeeDeClip4 this is in the Mastering section of Settings.
  3. Setup the audio link in Windows/Mac/etc to output the required bit depth and sample rate.
  4. Ideally verify the correct depth and rate is being delivered to the DAC.


Bear in mind that these will work perfectly with cheap USB cables and cheap optical links so concentrate on getting the system worked out before spending a lot on cables. More expensive cables won't make an difference but they are quite popular.
Note: Try never to kink or knot any optical cable, they rely internally on a gentle curve and no damage as they are light guides, not copper wires.


Copyright © 2007-2017, CuteStudio Ltd.
Page generated in 0.205s, Powered by Silk V1.2-0 from Cutestudio