Crash course in Bits

[boa12]

wanted to revisit here since i just got this dts 96/24 disk. below is a repost from the dvd section today.

48k sampling rate from my prepro was right, same as any other dvd. but the bit rate is a dif story & it sounds even better to me.

so if i have a dts encoded 96/24 & 96/24 dacs in the processor, why don't i get a higher than 48k sampling rate? thanks.

"ok it arrived today & sounds pretty darn good. i'm not a huge queen fan but think they were

pretty talented & pretty interesting to say the least. this has some of the old top 40 hits as

well as some fine non-top 40 on it. even ends w/ "god save the queen".

this has tracks for dvd-audio (mlp 5.1), dts 96/24 (5.1), & pcm 96/24. as i've mentioned i'm not

yet w/ a dvd-audio player, but this is the 1st i've found recorded at the much higher rates that

can be done on a

regular dvd-video player. & as I hoped, on my dvdp display it runs at much higher audio bit

rates. in fact it's showing TWICE the rate of a regular dts dvd and about 8 TIMES the normal

dts dvd rate for pcm 96/24 (stereo). the display on pcm 96/24 stereo also shows it as 24 bit). it

however is still showing the regular dvd sampling rate of 48k on my pre/pro.

as for sound WOW. the highs are just scintilating. so clear (as you you guys w/ dvd-audio know

& then some).

there is a definite dif from the other dts music dvd beyond the mix itself, which btw is also well

done imo.

brian may (the guitarist) put a lot of work into this, being postponed many times to get it right.

1st music dvd i've had that has video display of the lyrics

& other info that runs right along w/ the audio. it also has the infamous "bohemian rhapsody"

video & some other extras.

either the pcm stereo or dts 5.1 sound great to me, each in it's own right. they've done some

interesting mixing on the dts 5.1. again the b. rhapsody shines here.

my guess is the dvd-audio mlp 5.1 is 96/24 also. (don't know if they can do mlp 5.1 in 192/24

but maybe they have here). but i'm sure it's sound even better w/ dvd-audio & those 96k

sampling rates."

------------------

My Home Systems Page

[boa12]

oh by the way the disk is Queen: A Night at the Opera".

dvd just came out w/ these formats.

------------------

My Home Systems Page

[boa12]

bump

so if i'm getting these extremely higher bit rates I must be getting a higher sampling rate (per trooper info above. maybe the b&k just doesn't show it because

of the digital connection. (?)

------------------

My Home Systems Page

[skonopa]

Okay, this is bringing back the wonderful time I had in college when I wrote a 25 page paper on Digital Signal Processing!

Boa12 asks:

so if i have a dts encoded 96/24 & 96/24 dacs in the processor, why don't i get a higher than 48k sampling rate? thanks.

Please bear with me as this is a little fuzzy - been a good 10 plus years since I researched this stuff in any real detail (and the technology has certainly improved

in that time frame ).

Sounds like you are confusing RESOLUTION with Sampling Rate - two completely different things.

There is no need for a higher sampling rate. The audio sensitivity of the human ear is limited to about 20-22khz at the high end and as per Nyquists Thereom, to get the optimum frequency rate, you'll need to double the samply rate (thus typical audio at 22khz will yeild 44khz or so.) That is necessary to prevent aliasing in the sampling process.

The "96khz" mentioned in the (24/96) is an "oversampling" rate. The actual content of the disk will be sampled at 48khz when it is produced. Your DVD Player and reciever will apply the oversample to help the filters to remove the "garbage". What you see on the display is the actual "true" sampling rate of 48Khz. this site has a good explaination of how oversamping works and why our components do it.

Now the "24" part is the "resolution" of the audio. As Trooper correctly said, CDs are typically stored in 16 bits at 44.5khz. The format of a typical audio CD is known as "red book". 16 bits allows 65535 descrete values to be stored. 24 bits allows over 16 million descrete values - quite a big difference. To use video - would you think a picture that can be rendered in over 16 million colors would look more realistic than one that can only be rendered in 65 thousand? The same would apply to sound.

If you are interested in this stuff, just pull up Google and do a search on Nyquist Theorem, Red book Audio, and Oversampling. There is a TON of stuff out there, some of it gets down to the nitty gritty mathematics, complete with formulas and such.

PS. It seems that me and m00n has a lot in common here on these forums - the one big difference is that I DO get down to the low level stuff. I have written a BASIC compiler in assembler on a VAX-11/750 while in college. One of my recent projects involved me writting to SCSI devices at the sector level, passing the data directly through the SCSI bus itself (bypassing the OS drivers). It is not uncommon for me to look at hex dumps of object and executable files here at work to see if the compiler is screwing up (I have found bugs in our C compiler on occasion), as well as writting tools that reads in object and executable files and does some low-level processing with them (mostly dumping out the symbol tables).

------------------

Steven Konopa

Fredericksburg, VA

Denon AVR3802

RF-3II (Front)

RC-3II (Center)

RC-3II (Rear)

RS-3II (A surrounds)

Infinity RS2000.5 (B surrounds - recycled)

REL Storm III (Subwoofer 1)

Yamaha YST-SW40 (Subwoofer 2 - recycled)

32 inch Sharp (TV)

JVC XV-S65GD (DVD)

[boa12]

thanks skon. so maybe I can or should get better sound due to the higher resolution? - the pcm 96/24 does show 24 bit on my dvdp display. though from reviews of my pre/pro, though it has 96/24 dacs they say it really only goes at 20 bit. still better sound in theory anyway?

i need to study this when i get a chance. or take computer engineering classes. thanks for your help & the links.

------------------

My Home Systems Page

This message has been edited by boa12 on 05-08-2002 at 12:48 PM

[boa12]

maybe this can help - someone on another said that dts runs at higher bit rates because it's less efficient.

what are the variable in this "efficiency" formula?

------------------

My Home Systems Page

[imel96]

quote:

---

Originally posted by boa12:

maybe this can help - someone on another said that dts runs at higher bit rates because it's less efficient.

what are the variable in this "efficiency" formula?

---

algorithm.

but don't confuse efficiency with lossiness. dts is more lossless from higher bit rates and efficiency.

------------------

imel

[skonopa]

Boa12 asked:

thanks skon. so maybe I can or should get better sound due to the higher resolution? - the pcm 96/24 does show 24 bit on my dvdp display. though from reviews of my pre/pro, though it has 96/24 dacs they say it really only goes at 20 bit. still better sound in theory anyway?

The short anwswer is "yes"!

To better explain this, lets pretend the graphics on the computer display in front of you is the music your are listening to. Lets pretend each "pixel" on the display is a sample and the color of the pixel is the pitch (this is after all, how digital cameras and scanners work in converting the scanned/captured image into a digital file).

As I mentioned before, the typical human ear is only sensitive enough to handle sounds between around 20hz to about 22khz (I have been tested up to about 21khz in sensitivity myself). As I also mentioned before, according to Nyquist Theorem, to get a good approximation of a waveform, it must be sampled at least twice the maximum frequency sampled, otherwise to much "aliasing" will occur.

What is this "aliasing"? To use the graphics analogy again, if you look closely at diagonal lines and curves on your computer display, you'll notice a "stairstep" effect - that is aliasing. In audio, the music will sound harsh and "artificial". You no doubt have seen or heard those toys, such as the dancing cactuses or the singing santas. Those are usually sampled at only around 11khz mono - that is why they sound so tinny and harsh. Same with computer synthisized voices. There are websites out there that gets into the real mathematical nitty gritty proof of Nyquist Theorem. Even I have a tough time understanding it all (and I have a math minor along with my CS degree!). Again, using the graphics analogy, do you think a 1 inch diagonal line drawn with 1000 pixels would look much better than one drawn with only 500 pixels?

48Khz (or 48,000 samples per second - 1 second of music was sampled 48,000 times - 48,000 snapshots of the waveform is taken each second), which is slightly more than absolutly needed is a good compromise in maintaining audio fidelity and saving space (realize that all 48,000 of those samples must be stored on the medium for each second!).

It gets to the point where the ear would not be able to tell a difference, in other words, the typical person would not be able tell the difference if a piece of music was sampled at 48khz or sampled at say, 80khz. Same with graphics - it'll get to the point where the eye would not be able to tell if a one inch diagonal line was drawn with say, 10,000 pixels from one drawn with 20,000 pixels. Anything beyond that would just be a waste of space. Unfortunatly, current display technology is not anywhere near that kind of fidelity - even HDTV.

Now, where does the 96khz number play in? That is an "oversample" your component uses. What it'll do is try to "fill-in" the waveform with extra samples, thus trying to smooth out the waveform. Basically, it'll take two samples from the original signal, use some algorthim to calculate what the value of a sample would be if there was one between the two samples and assign a value to that and then shove it through the DAC, thus giving the DAC more samples to use to rebuild the original waveform to send to your amp (and ultimatly out the speakers). That is why there is also a processor in your component. In graphics processing - this is called "anti-aliasing". It is used to try to smooth out a diagonal line by filling in pixels between the "stairsteps" by making a guess as to what the color of the pixel would be if there actually was one there, thus yielding a smoother looking line.

Now, as to the 24 bits. Typical CD audio, as per the "Red Book" standard, is stored at 16 bits/sample. That is, exactly what the pitch of the sound at that exact moment is converted to a value and stored. 16 bits allows 65,535 descrete values, thus, say value 0 would be complete silence - value 65,535 would be 22khz at full volume. value 32,237 would be 11khz at medium volume - at the middle. Using the graphics analgy, standard VGA uses an 8 bit value to store the color of the pixel - 255 colors - 0 = black, 255 = white. The problem is that what if the actual wave yields a value between 4000 and 4001. A "best guest" would have to be made as to what value to assign the sample if the actual wave is between two sample values.

Now, DVD uses 24bits/sample! 24bits allows over 16 MILLION descrete values - quite a bit more than 65,535. That means, there is that many more values that can be used to assign to the samples - thus yielding a more accurate representation of the wave. Same with graphics. What if you want to assign a pixel that is not quite the color value assigned to 128, but greater than the color value assigned to 127. You'd have make a best guest and pick one of the two colors. But with, say 16 bit graphics, you now have 65 thousand colors to work with, much better chance of finding the right color to assign a given pixel. Why not say, 32 bits/sample - the problem is again, a compromise between space and quality. Again, most people probably would not notice a difference in quality beyond 24bits/sample. One second of audio sampled at 48,000 times and stored at 24 bits/sample would take up about 144,000 bytes of data. This would come out to over 17 megabytes/minute of audio in regular 2 channel stereo. At typical 5 minute song would require over 86 megabytes of storage! Just think how much storage would be required for a 6 channel DTS soundtrack for a movie, not to mention the video part of it also.

Bear in mind that this is a rather simplistic explianation - it does not even take into account pulse code modulation (PCM) encoding and even error detection/correction (why a CD still sounds good even if there is a scratch in it - of course if the scrath is to severe, to much data would be loss, thus the player would not be able to resolve the waveform and will "skip").

Thus, comes the answer to your next question:

maybe this can help - someone on another said that dts runs at higher bit rates because it's less efficient.

what are the variable in this "efficiency" formula?

In order to fit that much data on a typical DVD disk (which holds about 4.7 gigabytes for a regular single layer/single sided disk), it must be compressed. The video is compressed using MPEG-2. That is also how signals are transmitted from the satellite to my DirecTV reciever. That is also how it is stored on the 40 gig harddrive in my reciever. A video compressed with MPEG-2 takes up about 1 gigabyte/hour. MP3 audio compressed using 192kbps encoding takes about 1.2 megabytes per minute of audio (much better than 17 megs!).

The efficiency of a compression algorithm is the measure of how small it can compress a piece of audio (or video), but keep the quality as good as the original. Lets say algorithm A can compress a 1 minute song from 17 megs down to 1 meg - that is 17:1, but still sound as good when uncompressed and played back as the original. Let say algoritm B can compress that same one minute song down to only 2 megs. Algorithm A would be more effiencent. Now, lets say algoritm B can compress that song down to 1 meg also, but when uncompressed and played back, it sounds horrible. Again, algoritm A would be more effiecient because it would be able to retain the quality of the original.

MPEG-2 and MP3 are both "lossy" compression algorithms. That is, they throw away some data to achieve the 7:1 or so compression ratio. The "bit rate" encoding allows the encoder to determine just how much to "throw away", thus a lower bit rate would yield a better compression ration (say 12:1 for 96kbps compression), but a terrible loss in quality, whereas a larger bit rate (192kbps) yields only about 7:1, but nerely indistinquisable from the original, uncompressed audio. For MP3 - 192kbps encoding is a very good compromise for space versus quality.

You finaly say:

i need to study this when i get a chance. or take computer engineering classes. <http://216.37.9.58/ubb/smilies/Smile.gif> thanks for your help & the links.

Like I said, there is a ton of stuff on the web if you are interested in the nitty gritty of it all! Just pull up Google and do a search on some keywords such as "oversampling", "digital signal processing", or "Nyquist Theorem", just as a start. I sure as hell wish I had this resource avalaible when I was researching this stuff back in college all those 10 years ago .

I hope this explains what is meant by "sample rate" and "bits/sample" and so forth.

------------------

Steven Konopa

Fredericksburg, VA

Denon AVR3802

RF-3II (Front)

RC-3II (Center)

RC-3II (Rear)

RS-3II (A surrounds)

Infinity RS2000.5 (B surrounds - recycled)

REL Storm III (Subwoofer 1)

Yamaha YST-SW40 (Subwoofer 2 - recycled)

32 inch Sharp (TV)

JVC XV-S65GD (DVD)

This message has been edited by skonopa on 05-08-2002 at 10:51 PM

[boa12]

thanks imel. so the formula (algorithm) itself is less

efficient. so there is a tradeoff between higher lossless and higher compression. anyway ren. over in tech questions "dolby vs dts" says that dts is less efficient so it runs at higher bit rates. may want to check out that thread. w/ this digital tech i'm mostly nothing but confused.

------------------

My Home Systems Page

[boa12]

& thank you steven. trouble is i have studied a lot of the material about it on the net & still don't really get. perhaps this requires some real training &/or education.

so means to an end, all other things being equal (including recorded volume level), is there any reason dts should sound better than dolby. it sounded like as

imel said the superior lossless of dts should make for more audio dynamics (so i guess indirectly higher displayed bit rates mean better lossless but does that mean better dynamics?).

like i know in layman's terms that dvd-audio is supposed to sound (or maybe it's more a sensing) better because of the higher frequencies (higher sampling rate) and because of the higher resolution (bit quantity) which translates to a greater dynamic range.

does any of this apply to dts vs dolby on a regular dvd-video player?

------------------

My Home Systems Page

[skonopa]

boa12 asks:

so means to an end, all other things being equal (including recorded volume level), is there any reason dts should sound better than dolby. it sounded like as

imel said the superior lossless of dts should make for more audio dynamics (so i guess indirectly higher displayed bit rates mean better lossless but does that mean better dynamics?).

In my personal (somewhat expert) opinion, I would say yes with everything else being equal - speakers, pre/pro/amp, DVD player as well as the quality and volume of the material being represented. For the simple fact that in a less effieicnet/lossy compression algortithm, some of the dynamics of the sound cand and sometime does get lost.

like i know in layman's terms that dvd-audio is supposed to sound (or maybe it's more a sensing) better because of the higher frequencies (higher sampling rate) and because of the higher resolution (bit quantity) which translates to a greater dynamic range.

does any of this apply to dts vs dolby on a regular dvd-video player?

I guess this depends ultimatly on how good the DVD player can recreate the audio/video from the source disk. For that, I guess it comes down to the age old saw in audio/videophila - how does it sound and look to you. I would imagine one of the better quality high-end DVD players can process and recreate the images better than an $80 walmart special since they most likely will use more powerful processors/DACs, and better quality parts, such as a more sensitive pick-up head, and higher grade electronics. My JVC XV-S65GD DVD player, which is a very good mid-range model, does a superb job of rendering the images and playing the audio. The only downside is the MP3 playback capabilities is missing some much needed features (such as random shuffle and playlist setup) on that model.

ALL DVD players should at least be able to process a basic DTS signal per the DTS standard set forth by Dolby.

------------------

Steven Konopa

Fredericksburg, VA

Denon AVR3802

RF-3II (Front)

RC-3II (Center)

RC-3II (Rear)

RS-3II (A surrounds)

Infinity RS2000.5 (B surrounds - recycled)

REL Storm III (Subwoofer 1)

Yamaha YST-SW40 (Subwoofer 2 - recycled)

32 inch Sharp (TV)

JVC XV-S65GD (DVD)

[boa12]

steven, i know that dvd-audio does use mlp - meridian lossless compression - to be able to get those higher sampling rates (up to 96k for 5.1 and 192k stereo) and resolutions, and higher freq and dynamic range (like up to 144db i think) respectively. but of course u need an mlp decoder (dvd-audio player to do it).

anyway, u guys may want to visit that "dolby vs dts" thread under tech questions. it was said there that dts is less efficient which results in higher bit rates. efficiency is a broad term. maybe they meant dts is less efficient in compression.

------------------

My Home Systems Page

This message has been edited by boa12 on 05-08-2002 at 11:58 PM

[skonopa]

Boa12 said:

anyway, u guys may want to visit that "dolby vs dts" thread under tech questions. it was said there that dts is less efficient which results in higher bit rates. efficiency is a broad term. maybe they meant dts is less efficient in compression.

Efficiency in terms of required bit rates does translate directly to efficiency in compression. If a compression algoritm can retain the quality of the source material in a lower bit rate than another algorithm, than that algoritm is considered more efficient. A lower bit rate often translate into less space required to store the data. Using MP3 as an example, an MP3 file encoded to 96kbps takes up much less space than one encoded to 192kbps.

Some people say the Ogg Vorbis compression is more efficient than MP3. Those people say that an Ogg file encoded to 128kbps sound just as good, if not better, than an MP3 encoded to 192kbps. Not only that, but Ogg Vorbis is completely free and open source - no royalties required to use it in a commerical product, unlike MP3 compression. I have yet to try it myself to do a comparision - sounds like something to try this weekend.

Efficiency is completely a space issue - it would be better to use a compression scheme that is more efficient since the data would be smaller, thus taking up less space (thus allowing more material to be stored on a disk).

------------------

Steven Konopa

Fredericksburg, VA

Denon AVR3802

RF-3II (Front)

RC-3II (Center)

RC-3II (Rear)

RS-3II (A surrounds)

Infinity RS2000.5 (B surrounds - recycled)

REL Storm III (Subwoofer 1)

Yamaha YST-SW40 (Subwoofer 2 - recycled)

32 inch Sharp (TV)

JVC XV-S65GD (DVD)

[boa12]

thanks steven. yea take a look at that other thread. i found the dts rebuttal to dolby's review of dts & posted that link there. maybe u can make more sense of it, but dts does seem to have some good points for their methods. just seems to me that compression would compromise sound quality to some degree.

& my ears do seem to prefer dts overall. don't think i'll ever completely understand why though.

------------------

My Home Systems Page

This message has been edited by boa12 on 05-09-2002 at 02:15 PM

[skonopa]

boa12 said:

thanks steven. yea take a look at that other thread. i found the dts rebuttal to dolby's review of dts & posted that link there. maybe u can make more sense of it, but dts does seem to have some good points for their methods. just seems to me that compression would compromise sound quality to some degree.

A good lossless compression should NOT compromise the sound. A good example of a lossless compression scheme that is used commonly is the trusty old zip file format. To be able to run those programs that are zipped up, it better not lose any information in the compression otherwise the computer will have a fit when trying to execute one of those programs. However, a lossy compression scheme, ala MP3 will loose some sound quality. Ideally, it is information that would not be detectiable upon playback that is thrown out during the compression. That is why MP3 doesn't seem to sound as good, especially on the high-end stuff that many of us have on here, as regular CD audio. Usually it is the bass that suffers in MP3 files.

For effieciency - given equal quality, the more efficient compression would be more desirable. It seems that Dolby claims that DD is more efficient than DTS, but if people can tell a definite difference in sound quality versus DTS (DTS being better), than perhaps DD is not more efficient. If there was no difference in sound quality, than, yes, DD would be more efficient. I'll have to make a comparison on my own gear between DD and DTS.

------------------

Steven Konopa

Fredericksburg, VA

Denon AVR3802

RF-3II (Front)

RC-3II (Center)

RC-3II (Rear)

RS-3II (A surrounds)

Infinity RS2000.5 (B surrounds - recycled)

REL Storm III (Subwoofer 1)

Yamaha YST-SW40 (Subwoofer 2 - recycled)

32 inch Sharp (TV)

JVC XV-S65GD (DVD)

[cc1091]

Ah ha! Resolution. I think this gets closer to the reason why my 13+ year only Denon DCD1500II sounds better than my two year old Phillips dubbing CD deck.

The Phillips is certainly more detailed, but less musical. The Denon is more silky smooth. The Phillips more critically focused. The Phillips also seems to lack some of the bass depth.

[boa12]

ok should of remembered why, but as trooper figured even w/ a 96/24 dvd-video it won't do more than a 48k sampling rate thru the digital transport - because of those copyright restrictions (48k) which sony adamantly adheres to that's also why dvd-audio & sacd have to go around the digital connection.

if i connect the player w/ the analog outs to direct ins though i get a 96k sampling rate. sounds/feels a lil more dynamic this way.

also did somebody say dvd-video usually run at 48k/24 bit thru a digital connection? i thought regular dvd-video were also 16 bit (like a cd) though w/ SR of 48k instead of 44.1k. & hdcd is 20 bit & dvd-audio & like this new dts 96/24 are 24 bit.

anyway, this dts 96/24 seems to sound/feel more dynamic than a regular dvd-video (48k/16bit?) even w/ 48k/24bit using the digital connection.

w/ the 96/24 dts disk running in pcm from the analog outs, bit rates are up to 4608kbps & SR is 96K. use the processor to do 6 speaker stereo. sounds better to me.

------------------

My Home Systems Page

This message has been edited by boa12 on 05-20-2002 at 02:58 AM