Need a summarized lesson on digital audio and DACs

[jsr]

I've never kept up with technology well. I went from CDs to adopting MP3s without really looking at the details in between. Most of my MP3s were ripped from CDs via iTunes nearly a decade ago with the iTunes default settings, which I found out are far from ideal (heavily compressed and losing a lot of detail). The other MP3s are from friends who are of the kind that are fine with tiny ear bud earphones from their iPods (i.e. don't really care too much about audio quality).

Recently, I started looking at external DACs and am quite confused/overwhelmed with all the compression formats, ripping bit rates and DAC sampling rates. I'm fine with the concepts of ADC and DAC conversion. What I need a summarized lesson on are the various formats of digital audio and how the available audio devices such as the external DACs available relate/interact with them. From web reading, MP3s seem to be the poorest digital format losing the most of the original lossless content. Then there's AAC, which I've read is supposed to be an improvement, although the conversion bit rates are the same when ripping from a CD. And there's MP4 (M4a and M4p) which are lossless. I don't know the difference between M4a and M4p. I've seen them used interchangeably to describe the Apple format. Then there's FLAC, a form of lossless digital audio.

What prompted some of this confusion was when I started looking at external DACs. In a review of a NuForce DAC, the reviewer said the USB port supports 24bit 96kHz sampling and that his USB content was sampled at 88.1kHz. I thought the conversion (ripping) from CDs to create MP3s was described in bit rate (320kbps, etc.) and couldn't find anywhere mentioning the sampling rate of ripped content from CDs so I don't understand how the DAC's USB port's sampling rate is incompatible with the USB (MP3?) content. Or is he referring to lossless stuff only? This initial confusion grew to my thinking about all this format and hardware stuff we have today and am now completely lost. As you can probably tell, I'm not even sure exactly what to ask as you usually need to know enough to ask intelligible questions and here, I'm still a bit too lost. So I'll start with these questions and hope to clarify further.

I've read individual articles online about a specific topic (i.e. CD ripping bit rate or DAC sampling rate, etc.), but nothing pulling it all together. My apologies for the broadness of this question. I hope you knowledgeable folks out there can take some time to put together some explanations that pull all this stuff together to help a dumba$$ like me catch up to the times.

Thanks.

[greywarden]

Re: Need a summarized lesson on digital audio and DACs

There was time that everyday for a month probably that DACs were a hot topic around here. I think we all decided that a DAC with a well-designed power supply and 2 seperate crystals, one for 44.1, 88.2, 176kHz, and one for 48, 96, 192Khz, would be ideal. There are several different great DAC chips from Sabre, Analog Devices, Wolfson, etc. and that jitter is hardly measureable, much less audible. Just look at the specs, there are DACs that are 0.01% THD and some that are 0.001% THD in the same price range.

When I rip CDs, I rip them to 44.1kHz/16-bit, becuase any higher is absolutely worthless. Some people rip them to FLAC, which makes them somewhat smaller files, and others with iThingys rip them to AAC or w/e file type iThingys use.

You can, however, from various websites download high-resolution music, up to 192kHz/24-bit and most of them are actually from the masters. My advise to you is build a NAS (Network Attached Storage) computer, Mini-ITX board, CPU, 4GB RAM, etc, if you don't already have a computer that can handle that sort of storage on your network already, and select a reasonably-priced DAC or one of the higher-end sound cards, like the ASUS Xonar series.

[fastbike1]

Re: Need a summarized lesson on digital audio and DACs

Likely to be lots of controversy.

Lets start at the beginning. Where are you playing this music? At the PC or do you want to get to your stereo/HT? The whole bitrate/format stuff is actually pretty simple (controversy aside), once you know what your end goal is.

[bohdan]

Re: Need a summarized lesson on digital audio and DACs

Hi jsr,

I would strongly recommend for you to take a look at Ultimate Equalizer V4 http://www.bodziosoftware.com.au/

You will find there a lot of information on how to put together a fairly complex (or a very simple, if you do not need all components ) home entertainment system with full-digital DSP path for audio server playback.

http://www.bodziosoftware.com.au/Signal_Routing.JPG

http://www.bodziosoftware.com.au/Ultimate%20Equalizer%20V4%20in%20SPDIF.pdf

The above two links I supplied above will only give a small amount of total information available for you, it’s best if you visit the website http://www.bodziosoftware.com.au/ and browse to your heart content.

Let me know if you have more questions.

Best Regards,
Bohdan

[jsr]

Re: Need a summarized lesson on digital audio and DACs

Thanks guys. I'll start with some of those links when I get a chance to browse them. I don't want to get into any controversial debates about what's better. My goal here is to know what the various formats are, what the various ripping rates are from CDs, and what the sampling rates are and what they all mean both individually and together. Like the example I mentioned above, the reviewer said the DAC's USB channel supported 24-bit 96kHz sampling so his USB music wouldn't work. I'm confused with this specific scenario because if his USB tracks are ripped from CDs (into some form of compressed format), I don't understand what 44.1kpbs, 320kpbs, etc. have to do with the DAC's sampling rate of 96kHz. Or is he only referring to the DAC's USB port's capability of decoding the lossless content? Are those figures (kpbs bit rate vs kHz sampling rate) for different things (i.e. bit rate for compressed formats and sampling rates for lossless)?

I don't use my PC for music/HT. I use either CDs or a USB. The USBs hold MP3s (and a few M4a/p lossless tracks). I'd like to move away from physical CDs and have high(er/est) quality digital audio that I can put into my USBs (or other memory cards like SDs). In doing so, I've been looking into separate DACs. That's where the whole ripping at *** kbps and DAC's sampling rates of *** kHz all got confusing.
Should I still want to play some CDs occasionally, I'm thinking to bypass the internal CD player's DAC (which I assume is 44.1kHz) by taking the pure digital signal from the digital out and using the external DAC's 24-bit/192kHz processing instead for arguably better music.

I'm probably mixing up the use of ripping rates and sampling frequencies not knowing if these apply to specific things only like compressed vs lossless content. And you can probably see my poorly directed questions as I don't know enough to ask specific questions. So summaries of this stuff and how they all relate is appreciated.

Thanks.

[ROTECH]

Re: Need a summarized lesson on digital audio and DACs

JSR,

Thank you for posting this. I also would like to know the answer to many of these questions. For me, it is more about when/where/how to perform conversions when ripping and downloading.
Perhaps and easy way for me to learn would be for people to explain the steps they perform from purchase of a CD, or download, through their system, naming the components along the way. What conversions are done, and why.

As of now, I dont use a PC for any sort of server, but I would like to get to that point. I do use a PC to transfer digital files to CD, or iThingy and what should I do while there, to get the best usable quality with any given equipment.

Thanks!

[greywarden]

Re: Need a summarized lesson on digital audio and DACs

Thanks guys. I'll start with some of those links when I get a chance to browse them. I don't want to get into any controversial debates about what's better. My goal here is to know what the various formats are, what the various ripping rates are from CDs, and what the sampling rates are and what they all mean both individually and together. Like the example I mentioned above, the reviewer said the DAC's USB channel supported 24-bit 96kHz sampling so his USB music wouldn't work. I'm confused with this specific scenario because if his USB tracks are ripped from CDs (into some form of compressed format), I don't understand what 44.1kpbs, 320kpbs, etc. have to do with the DAC's sampling rate of 96kHz. Or is he only referring to the DAC's USB port's capability of decoding the lossless content? Are those figures (kpbs bit rate vs kHz sampling rate) for different things (i.e. bit rate for compressed formats and sampling rates for lossless)?

I don't use my PC for music/HT. I use either CDs or a USB. The USBs hold MP3s (and a few M4a/p lossless tracks). I'd like to move away from physical CDs and have high(er/est) quality digital audio that I can put into my USBs (or other memory cards like SDs). In doing so, I've been looking into separate DACs. That's where the whole ripping at *** kbps and DAC's sampling rates of *** kHz all got confusing.
Should I still want to play some CDs occasionally, I'm thinking to bypass the internal CD player's DAC (which I assume is 44.1kHz) by taking the pure digital signal from the digital out and using the external DAC's 24-bit/192kHz processing instead for arguably better music.

I'm probably mixing up the use of ripping rates and sampling frequencies not knowing if these apply to specific things only like compressed vs lossless content. And you can probably see my poorly directed questions as I don't know enough to ask specific questions. So summaries of this stuff and how they all relate is appreciated.

Thanks.

CD is 44.1kHz/16-bit, DVD is 48kHz/24-bit, SACD is 88.2kHz/24-Bit. These are digital formats, whereas Tape, Vinyl, etc is analog.

The usual off-the-shelf DAC can handle information from a digital USB, Coax, Optical or AES-EBU signal to RCA or XLR, etc. from whatever the digital is - 44.1kHz/16-bit to 192kHz/24-bit.

If you are just ripping CDs, I recommend EAC either with or without the FLAC codec, I don't care for FLAC, I just rip in WAV format to 44.1kHz/16-bit, but I have alot of FLAC files from certain sources (Thanks Bryan!) that I enjoy as well.

I have a HRT Music Streamer II, the little red one, and going from laptop audio to an outboard DAC, even just a $150 one, was a BIG improvement. That being said, I really would love to have a nicer DAC one day, maybe an Audio Research or PS Audio.

Sorry if this is incoherent, I took a tramadol earlier and I'm a scatterbrain as it is haha.

So what you'll need is. Computer (With WAV or FLAC files) -> USB Cable -> DAC -> RCA Cable -> AVR/AMP/Etc.

And it takes a little setup with Windows, but it's not too challenging, and it will hardly ever give you problems.

[donradick]

Re: Need a summarized lesson on digital audio and DACs

I am far from an expert, but I am familiar with many of these aspects of digital audio.

A "standard" music CD always has 44.1Khz sample rate with 16 bits data depth.
http://en.wikipedia.org/wiki/Compact...Hz_sample_rate


If you multiple the sampling rate by the bit depth by 2 channels for stereo, you get 1411 kbits/sec
http://en.wikipedia.org/wiki/Audio_bit_depth

Lossy formats (MP3) compress or throw away much of the high frequency information to achieve a
bit rate of say, 128Kbps. Lossless formats don't throw away anything.

You can get Super Audio CDs and DVD-A (Audio) with higher sampling rates and bit depths.
Jitter is always produced by physical disks:
http://en.wikipedia.org/wiki/Jitter

There are certain quantize errors that are unavoidable when converting sample rates.
For example, audio on a DVD is sampled at 48Hhz, and to convert that into a 44.1Khz for
an MP3 file will result in a little inaccuracy. (Same thing happens when playing a 16/44.1 wave file
through a 24/96 DAC, although I would assume that the engineers did the best they could to minimize the
audio defects) I suspect this is what the OP had been told
about a 88.1 Khz USB bit stream, and a 96Khz DAC.
http://en.wikipedia.org/wiki/Quantization_error

I don't really understand the original question. A particular music file will have a particular sample rate and
bit stream, and the computer will just stream that out the USB port. The DAC on the other end of the cable
is responsible for turning that data stream into (analogue) audio waves. If all your music is 88.1, you've
sort of goofed up somewhere.

Hope this helps, there are many more layers of subtle things in the chain, and the subject for
many, many flame wars.....

[greywarden]

Re: Need a summarized lesson on digital audio and DACs

...
There are certain quantize errors that are unavoidable when converting sample rates.
For example, audio on a DVD is sampled at 48Hhz, and to convert that into a 44.1Khz for
an MP3 file will result in a little inaccuracy. (Same thing happens when playing a 16/44.1 wave file
through a 24/96 DAC, although I would assume that the engineers did the best they could to minimize the
audio defects) I suspect this is what the OP had been told
about a 88.1 Khz USB bit stream, and a 96Khz DAC.
http://en.wikipedia.org/wiki/Quantization_error
...

That's the reason why you should look for a DAC with dual clock crystals on the board.

[tyger23]

Re: Need a summarized lesson on digital audio and DACs

Working at an audio company (making codecs for the consumer electronics insudstry), I know more than I should about this. I'll try and keep it simple, as most everything posted above is quite true.

Digital audio is generally predicated on a balance of file size versus sound quality. The better the sound quality, the larger the file size. As Napster took off back in the mid-90's, mp3 sharing became ubiquitous. I was in college back in those days, and music sharing was rampant.

However, because internet speeds were mostly dictated by your 56K modem, downloads of anything over 2MB took an hour or so. Therefore, music for sharing over the internet was encoded at a rate that would be considered moderate to poor. However, also remember that most MP3's in that time were not portable. They were computer based and generally listened to on horrid PC speakers from CompUSA or even worse laptop speakers. Quality was not necessarily the goal - sharing as much as possible as quickly as possible was the norm. This essentially gave the MP3 a bad name. MP3 encoding can be excellent, and it's still the most ubiquitous format - easily portable across most all devices today.

Fast forward past the iPod and into today where users are demanding that their audio quality be just as good as CD's while still being portable. Let's ignore the fact that most of these users are still demanding that quality of music while listening on the beyond crappy headphones included with their iDevice. So how do we get a portable audio format with quality equivalent to CD?

To do that, let's look back at the CD and understand how it works:

First, let's understand the Nyquist Theroem. http://en.wikipedia.org/wiki/Nyquist...mpling_theorem Basically, this states that for a given "maxium" frequency, the sampling rate must be at least 2X that frequency to encompass the vast majority of the usable frequency range within the desired band.

Since the human hearing is generally considered to be 20Hz-20KHz, the Nyquist Theorem states that the sampling rate must be 2X larger than 20KHz. Therefore, you'd want a sampling rate to be AT LEAST 40KHz. CD's are recorded at 44.1KHz, so that checks off the Nyquist requirement. 44.1KHz > 40KHz. But, why 44.1KHz? That's a crazy number!

Turns out this has to do with the recording equipment available at the time the CD format became available. Basically, video-type recording equipment was used to master the audio format CD. Because the video equipment was based on the NTSC specification, it just worked on to use 44.1KHz. Here is the explaintion as to why.

So, we're faced with turing a digital CD recorded at 44.1KHz into a portable audio format such as MP3, FLAC, OGG VORBIS, WAV, etc.

Now is when you get into format wars. As I mentioned before, turning a CD into a digital format is a balance betwen file size and audio quality. Let's peel the first one of the list - the WAV format. WAV is a format based on PCM. Well, lo-and-behold, the CD is a PCM format as well. So, if you rip your CD into a WAV format, you're going to get an exact copy without any loss of audio quality whatsoever. You're also going to get the absolute largest file size. Ever notice how recordable CD's are listed as 700mB or 80minues? Well, there's your simple way to know how big your WAV file size is. If you have 80 minutes of music, you'll consume 700MB of storage.

Next up are FLAC and OGG VORBIS. These are "lossless" formats that use a proprietary compression technique to basically provide you WAV quality with reduced file size. Note, the issue here is that these format files are not overly well known, and may require a PC to decode them. You're certainly going to struggle to find car radios that can play these formats, and I'm pretty sure the iDevices can't.

Apple's lossless format (AAC) and M4A type formats are similar to FLAC and OGG VORBIS. They are mostly considered lossless, but really they have some compression in them. Also, note that Lossless AAC and standard AAC share the same file extension and both are encoded specifically by iTunes.

Finally, let's look at MP3. There are so many ways you can encode an MP3 - from crappy to excellent. There is no debate, however, that the Napster era of the mid-90's made this the most popular and most easily supported format. Just about any car radio or digital device on earth can decode an MP3.

So, MP3 is the most common and easiest to use, but it's the crappiest - right?

WRONG!

Again, this all depends on the encoding (ripping) settings that were available when the MP3 was created from the CD. MP3's can sound every bit as good as WAV while being more portable by consuming significantly less file space. But how?

Well, lets' go back to the Nyquist theorem and human hearing. It is generally understood that human hearing is 20Hz to 20KHz. However - very, very, very few people can hear 20KHz. These are the "golden ears" of the world and babies/children who've never been exposed to a loud environment. Basically, if you've ever taken a 2-hour airplane ride without hearing protection in place, I'll state that you can't hear 20KHz anymore. Most adults of high-school age and over are 18KHz or below.

So, let's relate this back to MP3 format. Since we don't really need anything over 18KHz (let's call it 19KHz to be safe), why do I need to store digital data of stuff that is completely inaudible??? That's the predication of the MP3 format. It saves storage space by discarding audio data that is completely inaudible anyways. So what's the best way to encode? Back in the 90's everything was encoded at a constant bit rate of 128KBps. This has some serious degradation of the high frequencies and gave MP3 the bad name.

However, when you encode at 256KBps or higher, I contest that you will NEVER hear the difference between the MP3 and the CD. If you do, I'll pay for that CD, or call BS because you were listening to the two on different sources which colored the music differently. In fact, read this article to see where MaximumPC only recommends 160KBps.

So, let's record at 256KBps or higher! But wait! There's more! There's another way to save file space! Why bother trying to encode silence (or very low level music) such as that at the beginning or end of music tracks at the full 256KBps? It's a waste of digital storage! Enter the variable bit-rate scheme, which looks at the music and adjusts the bit rate accoringly. A high bit rate is maintained when the music demands it, but the bit rate lowers as the music decreases to silence.

Now, how to do that? USE EAC and Lame! Exact Audio Copy is a CD ripper and LAME is the MP3 encoder. LAME and EAC make it easy to do what I've described above by including presets that do this for you. I've personally encoded my entire CD collection using the "-V0" preset. This used to be known as the "--alt preset EXTREME". This preset provides a low-pass of ~19.5KHz and a target bit rate of 245KBps. I recomend you read more about Lame in general HERE.

So, if you haven't picked up on it - I recommend MP3 encoded with a high but variable bit rate scheme. I'll post the exact settings I use in EAC later tonight when I get home. MP3 is the most popular and portable format - and I promise - you won't hear the difference when using the settings I recommend.

Oh and as to file size - a normal song runs about 6-9MB per song.

Enjoy!

[Mayhem13]

Re: Need a summarized lesson on digital audio and DACs

Excellent post Tyger.....and i share most of your opinions EXCEPT for the Mp3 thoughts. Although we...me included can't hear a lick past 17khz doesn't mean that the information isn't valuable as we loose the harmonic content as well. Not a silly audiophool statement but an observation doing ABX testing of MP3 vs other lossless formats..

But for me, it's all about the interface and ease of use that computer audio allows for. Setting up a Computer audio playback system can be highly functional if done properly from the start.....and can be really pretty cheap too. WAV files while offer the highest resolution (and file size) are nearly USELESS for cataloguing or tagging. I've found the iTunes mesh to offer an excellent user interface, easy compatability with iproducts and an excellent networking solution.

File resolutions.....lot's of opinions here but buyers beware. 'High res' tracks are often not what they appear...offen upsampled from 16/44.1 and resold. NOT GOOD. I'll offer that 16/44.1 CD rips to a hard disc remove transport problems and the like, providing as good a playback performance as possible. 'SOME' high res masters do sound better than their standard def files.....NOT because of the higher bitrate but the carefull mastering that took place so in my honest opinion, not much value in the high res download version.

DACs....again lot's of opinions, most audiophoolery IMO. There are a few factual things to consider on the computer side. Computer environments ARE noisey environments, and the USB BUS is about the noisiest of them all. While there's devices and DACS available to isolate the noise, they're often expensive options that complicate the playback system. USB DACs that offer Asynchronous clocking do exhibit extremely low jitter, but still don't isolate EMI or other power related issues. The only practical way to remove system noise is using a Toslink/Fiber interconnect from the playback source to the DAC. Fiber/Glass can't pass noise, only data. Network audio over wifi or CaT is alos an excellent solution to EMI/Noise. Apple sells an incredible little device called the Airport Express that's a Wifi network player which includes a mini TosLink/Fiber output for a DAC for $99....IMO the absolute best value in music streaming for the $$$. Paired with an iPhone, iPad or iTouch makes for an extemely flexible and powerfull computer audio playback system with no need to turn on a TV or monitor.

[tyger23]

Re: Need a summarized lesson on digital audio and DACs

Ok, so my post above covered "digital audio formats". Here's one to cover DAC's.

A DAC is, simply put, a Digital to Analog converter. It converts audio from a digital format (typically I2S for consumer audio or HDAudio for PC's) into an analog format that you can listen to. Please understand that the actual DAC itself does NOT do the conversion of an MP3, etc into analog.

Again, the DAC only converts audio in a digital format to audio in an analog format. Note that I didn't mention I2S or HDA in my post above. These are digital audio TRANSPORT mechanisms, not digital audio STORAGE mechanisms.

The conversion of digital audio in a STORAGE format (MP3, AAC, WMV, etc) to a digital audio TRANSPORT format (I2S, HDA, and SPDIF) occurs in a DSP. The DSP can exist in several places, including software. So, when you click on an MP3 to play it, it will first be converted from a digital audio STORAGE format to a digital audio TRANSPORT format, and then to an ANALOG format. Note however, that the I2S and SPDIF transport mechanisms essentially operate in a PCM format. So, when you play a WAV file and your DAC is using one of those transport mechanisms as an input, you're getting a digital one-for-one of the original source file. One more caveat, though, is 5.1 content over SPDIF. If you are playing 5.1 content over SPDIF, it IS COMPRESSED! Only setereo, PCM data will transport uncompressed over a SPDIF cable.

Now, because of the advent of WHQL (Windows Hardware Quality Labs, requirements placed on ALL PC manufacturers), audio in a modern PC has been significantly improved over the last 5 years. Today's PC Audio Codecs (codec = DAC + ADC + DSP in one chip) have significantly better fidelity that even most USB based cards or even those Creative add-in cards. However, you won't see Creative advertising that point. The PC Audio codecs from IDT (forerly Sigmatel), Realtek, and Conexant (formerly Analog Devices) easily approach >100dB SNR, >94dB Dynamic Range, and better than 0.01% THD+N across the spectrum 20Hz-20KHz. These numbers are phenominal, especially given the improvement over such a short timeframe.

However, DACs are always affected by the accuracy of the crystal supplying the clock to the DSP and DAC. The less accurate the crystal (more jitter), the less accurate the audio. PC Audio codecs can be affected by this, but all PC Audio codecs are forced to meet a number called Sample Rate Accuracy. The codec must be within +/- 0.02%. This speaks directly to the accuracy of the clock entering the codec, and that means that the clock is pretty darned accurate.

Now, I'm certainly not stating that on-board PC Audio codecs can't be outperformed by a dedicated "DAC" or "Audio Player". They can. I'm just stating that ~80% of the people in the world won't be able to tell the difference. Please understand that some people who drop the $1000 on a high end "DAC" claim a huge difference in audio quality because they have to justify the purchase in their mind. Now, there are other factors in a PC that affect the audio performance of an on-board codec, such as layout of the motherboard, component selection (such as caps), and even the quality of the audio connection (the jack). For example, did you know that you would get significantly better crosstalk results (left channel bleed-through to the right channel and vice-versa) if you use RCA plugs or a 1/4 TRS jack (instead of the common 1/8 or 3.5mm audio jack)?

However, what is interesting is that many audiophiles will claim the benefits of "cross-feed". Guess what - cross-feed is another term for cross-talk. They are one in the same. It's just that cross-talk is viewed as bad and cross-feed is viewed as good. Funny, hunh?

Back to the DAC thing. All I'm advocating is for you to use a modern PC (made within the last 5 years) as your PC based music source. With this, play your music and if you're happy with the audio quality, don't worry about dropping the dough on a dedicated DAC.

If you're not happy, then by all means - spend the money. But, this is where we go back to talking about clock accuracy and so on and so forth. There are some great DACs out there, and I won't talk bad about them (other than stating that some are not worth the exhorbitant amount of money the cost). I'm sure that there's some discussion about DACs on this forum, or you can check out the DIYaudio forum. You'll get plenty of opinions on it - just watch what you buy and look at the specs. For example, the popular "NuForce" USB based DAC has great reviews, but when you look into the specs you'll find that it actually UNDERPERFORMS most modern on-board PC-Audio codecs.

Why sped the money on something that doesn't actually give you better performance?

Converse this with the uber-high end NAD Masters Series M51, with THD+N better than 0.001%, SNR of 123dB, and crosstalk better than 115dB. If you want this, you have my permission. It's fantastic, and I'm sure it will blow the waters out of anything on a PC. It also costs $2000.

It's all cost versus performance. I simply want to hilight that buying a dedicated, external DAC may not actually get you better performance than what you already have on your current PC.

Enjoy!

[tyger23]

Re: Need a summarized lesson on digital audio and DACs

Not a silly audiophool statement but an observation doing ABX testing of MP3 vs other lossless formats.

I'd only counter with - how were the MP3's encoded? If using the method I described above (EAC+Lame using the -V0 preset), then I continue to believe you'd be extremely hard pressed to tell the difference. Any MP3 encoded in a lesser format, then yes - I agree with your sentiments.

[jsr]

Re: Need a summarized lesson on digital audio and DACs

Wow, thank you for the great info everyone!

Don mentioned the quantization error from a DAC due to re-sampling an audio file was already encoded at a certain sampling rate (i.e. a 16bit 88.1kHz file to a 24-bit 96kHz DAC). And Grey mentioned dual crystals for the DAC. If running a X sample-rate encoded file into a DAC that is Y sample-rate only causes some quantization errors, why did the reviewer say it wouldn't play on that DAC? I know the specifics of the article aren't available. It was a review of a NuForce 24-bit/96kHz DAC with USB input. The reviewer stated his files on his USB stick were encoded at 88.1kHz and thus couldn't be played on the DAC's USB input because the DAC is 96kHz. He didn't mention just quantization errors. He stated he would have to re-rip the files at 96kHz to be able to use the USB port on the NuForce DAC. So does the difference in the file's sample rate (sample rate used to generate the audio file) and the DAC's sample rate cause just quantization errors or complete inability to be read/played?

tyger -
Thank you very much for all that great info. That is some of what I was hoping someone to do...summarize much of the technology and concepts and how they relate together.
I'm still stuck on that statement from the NuForce review where the reviewer stated his USB audio files created at 88.1kHz sampling couldn't play through the NuForce's USB port as the DAC there is 96kHz.
And what specs should I look at on the Nuforce DACs that are underperforming PC on-board codecs (THD, SNR and xtalk?). Those NuForce DACs were some of the ones I was looking at since they're small and decently priced. I don't need NAD performance, nor can I afford it. I was just looking for something better. For the computer, I only have a laptop at home which may not have a very good on-board codec since most things are compromised on laptops for space. I was looking more for an external DAC to plug a USB stick or SD card into to play the music and maybe take the digital out from a CD player to bypass the CD player's DAC to use the supposedly better external DAC.

I read a lot of people recommending (and saying they hear the difference - let's not go into details on the hearing a difference as I'm sure that is one of the flame wars) using 24-bit 192kHz encoding. Per Don's formula to get kpbs, that's 9216kpbs for 2 channels. That seems extremely high compared to the original CD content. Is that only for lossless formats like FLAC or M4A? Did I calculate that correctly? Being so much higher than the original CD content, what can that much data offer if it is still based on the 1411kpbs original CD data? What is that extra data composed of if the original CD only contained 1411kpbs worth of data?

Thanks.

[tyger23]

Re: Need a summarized lesson on digital audio and DACs

I'll try and break this down part by part.

Wow, thank you for the great info everyone!

Don mentioned the quantization error from a DAC due to re-sampling an audio file was already encoded at a certain sampling rate (i.e. a 16bit 88.1kHz file to a 24-bit 96kHz DAC). And Grey mentioned dual crystals for the DAC. If running a X sample-rate encoded file into a DAC that is Y sample-rate only causes some quantization errors

Let's tackle this first. Here's the rub - if you start with a CD as a source, you need to be encoding at 44.1KHz and 16Bit. Anything above this will involve interpolation from some encoding scheme (interpolation = guessing). As you can quickly comprehend, you can't create data from a file where data did not exist. Any time you "upconvert" a 44.1/16 source to 16/88.1 or 96/24, the upconversion process will "guess" at the missing data. The same thing happens on upconversion DVD players.

Now, the guessing at missing data will cause a change in the sound of the file. Whether or not it sounds better is personal preference. But, the fact remains that the data has changed, and the file is no longer equivalent to the original.

Locating an audio file that was originally recorded (at the source) for anything above 44.1/16 will be extremely challenging. Essentially, this means your original source would be an SACD, DVD-Audio, or Blu-Ray. The content protection on these formats is strict, and I strongly doubt anything "recorded" from these files was actually a true 1-for-1 copy of the source. More than likely, any time an audio file is "encoded" at a format higher than 44.1/16, it was either up-converted, or it was converted from the original source into analog and then back to digital at a higher sample rate. Either way, I don't think this is "Music as the artist intended" (to borrow as quote from Beats).

why did the reviewer say it wouldn't play on that DAC? I know the specifics of the article aren't available. It was a review of a NuForce 24-bit/96kHz DAC with USB input. The reviewer stated his files on his USB stick were encoded at 88.1kHz and thus couldn't be played on the DAC's USB input because the DAC is 96kHz. He didn't mention just quantization errors. He stated he would have to re-rip the files at 96kHz to be able to use the USB port on the NuForce DAC. So does the difference in the file's sample rate (sample rate used to generate the audio file) and the DAC's sample rate cause just quantization errors or complete inability to be read/played?

This reviewer is full of it. Not only because of the "endocded" files format, but because he has no idea how a modern PC works. My answer here would vary, dependant on your PC's operating system, but I'm going to focus on Windows 7 as it's the most current and most common now. The NuForce DAC (or the on-board audio, or any other audio playback device connected to the Win7PC) will show up in the systray as a "speaker icon" on the lower right hand side of the system. You can right-click on that speaker icon and choose "Playback Devices". The window that pops up will show all of the available playback devices on the PC. The NuForce would show up here. Double click on one of those playback devices, and then click on the "Advanced" tab. Here, you'll be presented with the format for playback of the hardware device. Most devices are automatically set by windows to be 48KHz, 16Bit, but you can manually choose to set this to anything the device will acutally support. The NuForce should show 24-Bit, 96KHz as a choice, as will most on-board audio codecs.

Now, Windows will FORCE the hardware to operate in this mode. So, what happens when you play a 44.1KHz, 16-bit audio CD (let's frorget about MP3's for a minute). Well, Microsoft will AUTOMATICALLY do what's called a Sample Rate Conversion. When the sample rate conversion occurs (prior to the audio being given to the USB DAC), here's where your quantization error occurs.

Basically, unless your hardware settings in Win7 match your encoding format, you just introduced quantization error and didn't know it. You only thought you were setting your sound card to a better quality audio format.

My recommendation is to make sure your playback hardware matches the encoding format of your audio. Again, for anything that started from a CD, this will be 44.1/16. This will help you avoid quantization error.

tyger -
Thank you very much for all that great info. That is some of what I was hoping someone to do...summarize much of the technology and concepts and how they relate together.

You're welcome.

I'm still stuck on that statement from the NuForce review where the reviewer stated his USB audio files created at 88.1kHz sampling couldn't play through the NuForce's USB port as the DAC there is 96kHz.

So am I, but I'm ignoring this now. Rant done.

And what specs should I look at on the Nuforce DACs that are underperforming PC on-board codecs (THD, SNR and xtalk?). Those NuForce DACs were some of the ones I was looking at since they're small and decently priced. I don't need NAD performance, nor can I afford it. I was just looking for something better.

Wow, this is a tough one. Mainly because the computer vendors don't advertise the performance of their on-board codecs. In general, you're looking for SNR and THD as the main specifications. For SNR, higher numbers are better and for THD, lower numbers are better when talking in percentages. When the number is not in percentages (in dB), then higher is better.

For the computer, I only have a laptop at home which may not have a very good on-board codec since most things are compromised on laptops for space. I was looking more for an external DAC to plug a USB stick or SD card into to play the music and maybe take the digital out from a CD player to bypass the CD player's DAC to use the supposedly better external DAC.

Not necessarily. It depends on the make/model of the laptop, on-board codec used, and how well the manufacturer adhered to the codec maker's recommendations for schematics and layout. Does your laptop have a SPDIF output? If not, there's not going to be a digital format available for you to extract digital data from the CD. If you can let me know the make/model of your computer, I can try and figure out which codec is on-board and how well it's supposed to perform (assuming the manufacturer didn't really screw up the motherboard).

Can you hear audible noise from the computer's main analog outputs? To verify - find and audio file that's nothing but endoded "silence". If you use iTunes, then you should be able to search for one and play it back. You might also find one on the internet. Play back that silence file and listen for humming, hissing, beeps, clicks, and ticks. Anything that's not silence. If all you hear is silence, then you're likely good-to-go with the on board audio codec. A USB audio dac may improve the quality some, but is it worth the money? I can't answer that question for you.

I read a lot of people recommending (and saying they hear the difference - let's not go into details on the hearing a difference as I'm sure that is one of the flame wars) using 24-bit 192kHz encoding. Per Don's formula to get kpbs, that's 9216kpbs for 2 channels. That seems extremely high compared to the original CD content. Is that only for lossless formats like FLAC or M4A? Did I calculate that correctly? Being so much higher than the original CD content, what can that much data offer if it is still based on the 1411kpbs original CD data? What is that extra data composed of if the original CD only contained 1411kpbs worth of data?

I'll refer to my above statement on quantization error or interpolation. There's no way a person can take a CD as the source file and "upconvert" the file to a higher encoding level without introducing error (interpolation). Now, people may think this error sounds better than the original format, and perhaps that's valid. However, it is error. It is NOT a one-for-one copy of the original source.

Just experiment and let your ears be the judge.

Enjoy!