Why is the given sample rate for TV 48 kHz and HD 96 kHz?

[loujudson]

Oh, cool! I've been looking for the answer ever since Bill Cosby asked it. Here it was, right here on Gearslutz on my Mac. That is so cool.

So, why are there animals? So we can eat?

Lou

[gsilbers]

interesting posts.

i just ASSumed 48k over 44.1k was decided because someone skimmed through the theory and ASSumed it sounded way better back when digital audio started.
the whole frame rate math theory is pretty interesting and makes more sense.


for higher freq range the term would be foldown or aliasing. where the frequencies above our listening range will be folded back down to our freq range thus creating an unclearer sound. minimal of course but noticeable when u compare to 96k.

and as for the marketing comments.. i agree. just look at dolby! c'mon man do we really really (think hard) really need them

[tom_lowe]

Originally Posted by gsilbers

interesting posts.

i just ASSumed 48k over 44.1k was decided because someone skimmed through the theory and ASSumed it sounded way better back when digital audio started.
the whole frame rate math theory is pretty interesting and makes more sense.


for higher freq range the term would be foldown or aliasing. where the frequencies above our listening range will be folded back down to our freq range thus creating an unclearer sound. minimal of course but noticeable when u compare to 96k.

and as for the marketing comments.. i agree. just look at dolby! c'mon man do we really really (think hard) really need them

a sound recorded at 96K and down sampled to 48K will sound clearer than a sound recorded at 48K.

as for do we need Dolby, well, without them cinema sound would be a lot worse if you think about their developments that made analogue sound tracks far better.

[dr.sound]

Originally Posted by gsilbers

interesting posts.
and as for the marketing comments.. i agree. just look at dolby! c'mon man do we really really (think hard) really need them

Gsilbers,
Yes, We DO need them much more than you realize!
They do quite a lot in keeping and setting standards.
When everyone delivers on HiDef without Film it's going to
be the "Wild West" without a Sheriff.
Standards are important. Dolby helps in the setting of standards.

44.1 for CD was originally done to fit a specific Concerto onto a CD.

48k was implemented because of storage.While the cost of storage has changed in recent years 48/24 is fine for where we are right now.
Studio's who make movies care more about 3D and selling more tickets,
not audio.

[tom_lowe]

Originally Posted by dr.sound

Gsilbers,
Yes, We DO need them much more than you realize!
They do quite a lot in keeping and setting standards.
When everyone delivers on HiDef without Film it's going to
be the "Wild West" without a Sheriff.
Standards are important. Dolby helps in the setting of standards.

44.1 for CD was originally done to fit a specific Concerto onto a CD.

48k was implemented because of storage.While the cost of storage has changed in recent years 48/24 is fine for where we are right now.
Studio's who make movies care more about 3D and selling more tickets,
not audio.

44.1 was use for CD as a conductor who was a friend of the Sony CEO insisted the whole of Beethovens (I think 5th) would fit on the CD. 44KHz would have enabled this, but 44.1KHz came from the fact it had to be stored on U-matic tape (I'm sure Google will give you the exact figures)

Why would 48KHz be used for storage? if they wanted to save on storage space, surely a lower sampling frequency would have been chosen?

as for CD, this will explain far better than I can

http://www.cs.columbia.edu/~hgs/audio/44.1.html

[gsilbers]

Originally Posted by tom_lowe

a sound recorded at 96K and down sampled to 48K will sound clearer than a sound recorded at 48K.

.

dont understand your post or u didnt understand mine. i agree 96k sound better than 48k i never said anything to the contrary.
the opposite, its exactly my point.

[gsilbers]

Originally Posted by tom_lowe

as for do we need Dolby, well, without them cinema sound would be a lot worse if you think about their developments that made analogue sound tracks far better.

that is true. good ol' dolby just needs to keep re inventing its wheel to sell new products.
hey not bad if im them, selling snake oil.
hey.. that was harsh. its a good company. gets a lot of money.

[gsilbers]

Originally Posted by dr.sound

Gsilbers,
Yes, We DO need them much more than you realize!
They do quite a lot in keeping and setting standards.
When everyone delivers on HiDef without Film it's going to
be the "Wild West" without a Sheriff.
Standards are important. Dolby helps in the setting of standards.

44.1 for CD was originally done to fit a specific Concerto onto a CD.

48k was implemented because of storage.While the cost of storage has changed in recent years 48/24 is fine for where we are right now.
Studio's who make movies care more about 3D and selling more tickets,
not audio.

dont get your point. most hidef tapes dont have dolby at all. id say 80% of hidef tapes are discrete audio. i know, ive seen plenty from fox , univ disney etc.


i thought 44.1k was because its double the sampling rate of the audible human hearing frequency range. nyquist theory and all. but it does have some extra frequencies there so if u would care to elaborate, sounds interesting.

[tom_lowe]

Originally Posted by gsilbers

that is true. good ol' dolby just needs to keep re inventing its wheel to sell new products.
hey not bad if im them, selling snake oil.
hey.. that was harsh. its a good company. gets a lot of money.

You REALLY need to go away and learn a bit about the history of sound ;-)

[gsilbers]

Originally Posted by tom_lowe

You REALLY need to go away and learn a bit about the history of sound ;-)

i might do that, and re read all the reasons why dolby has been only a crutch to mediocre technology.

i think they might have hired a lobbyist to pass the loudness law so they can sell the DP600.

hey its a hell of a company. is it public? i might invest in them.

[tom_lowe]

Originally Posted by gsilbers

dont get your point. most hidef tapes dont have dolby at all. id say 80% of hidef tapes are discrete audio. i know, ive seen plenty from fox , univ disney etc.


i thought 44.1k was because its double the sampling rate of the audible human hearing frequency range. nyquist theory and all. but it does have some extra frequencies there so if u would care to elaborate, sounds interesting.

I posted a link above explaining this ;-) despite the extra frequencies, most CD players circuitry is limited to 20KHz.

This explains all

Compact Disc - Wikipedia, the free encyclopedia

and sorry, it was Beethoven's 9th, not 5th symphony, but that article explains that may have just been an excuse during an argument between Sony and Phillips.

[gsilbers]

Originally Posted by tom_lowe

I posted a link above explaining this ;-) despite the extra frequencies, most CD players circuitry is limited to 20KHz.

This explains all

Compact Disc - Wikipedia, the free encyclopedia

and sorry, it was Beethoven's 9th, not 5th symphony, but that article explains that may have just been an excuse during an argument between Sony and Phillips.

oh, its a little more in depth tha that but nice. interesting info.

Fergus Cassidy

[kk@jamsync.com]

Originally Posted by tom_lowe

HDTV doesn't HAVE to use AC-3, they could have gone with AAC, MPEG, Dolby TrueHD, DTS, the list goes on. They stuck with AC-3 for backwards compatibility with home decoders/theatre systems. If they truly wanted 96KHz they could have done so.

TrueHD wasn't around when the HDTV transmission standard was defined. Doesn't matter "if they truly wanted 96kHz", they truly standardized on AC3. Dolby has been more instrumental in developing broadcast standards and the equipment to maintain them than any other company. They had the most influence, so they won the ballgame.

[kk@jamsync.com]

Originally Posted by tom_lowe

48KHz was chosed for TV WAAAAAAY before AC-3 was developed. AC-3 maxes out at 48KHz due to the reason I mentioned earlier, not the other way round, the broadcast specs dictated what AC-3 could cope with, the broadcast specs weren't written around AC-3's capability.

When I speak of "TV", I mean "HDTV", the modern standard.

HDTV is required to carry AC3 and has been for over a decade and a half. Suggest you read the doc where the following quote originates (AC-3 of course started with AC-1):

"The genesis of the AC-3 technology came from a desire to provide superior multi-channel sound localization for High Definition Television sound. In the United States, the High Definition Television (HDTV) standardization process formally began in 1987 with the formation, by the Federal Communications Commission (FCC), of the Advisory Committee on Advanced Television Service (ACATS). The initial HDTV system proposals called for analog based picture transmission and digital sound transmission. The initially proposed sound systems matrix-encoded a multi-channel audio source into a stereo pair, which was then digitally encoded with Dolby AC-1, a low cost delta-modulation based coding technology. At the receiver, the two audio channels would optionally be decoded into four channels using a matrix decoder. This proposal basically used the 4-2-4 multi-channel matrix system as a 2-1 bit-rate reduction system, since the matrix technology reduced the bit-rate required to convey four channel audio by a factor of 1/2. By 1989, advances in audio coding technology and DSP hardware made it possible to move from the AC-1 audio coding technology to the transform based AC-2 coding technology, simultaneously raising audio quality while reducing bit-rate. The use of the multi-channel matrix technology remained." etc., ect., etc....refer to the doc for the eventual ATSC determination.

http://www.mp3-tech.org/programmer/docs/ac3-flex.pdf

I'm not going to slip into an argument about where 48kHz originated for AC-3. The history is self-evident and it is the current standard for HDTV. Until the ATSC changes its mind (not likely in the near future), that's where it will remain.

Anyone who owns a Dolby Digital encoder knows that it does not take inputs higher than 48kHz. HDTV is required to use Dolby Digital. Hence my statements: "HDTV uses Dolby Digital which is pegged at 48kHz for delivery standards. Unless they develop a much more efficient codec than AC3 it will probably stay there for awhile." It doesn't matter what they *could* have used.

[cinealta]

Originally Posted by jamiedonnelly

...why for TV/Film the sample rate is 48 kHz....

Reverse chronological basis for 48kHz digital audio standard:

*1981: AES adopts 48kHz as professional digital audio sampling rate standard. 48kHz was proposed by Alastair Heaslett of Ampex. Heaslett's rationale was based on the following considerations:

1) Data "block" (integral samples) boundaries should coincide with frame boundaries of NTSC/PAL/SECAM/Film sound track (for ease of editing).

2) System clock frequency should be multiple of NTSC/PAL/SECAM/Film horizontal frequency (for lock).

3) System clock frequency should be multiple of sampling frequency.

4) Sampling frequency should be twice that (Nyquist theorem) of 22.5kHz (20kHz human hearing range + 2.5kHz, so anti-aliasing filter has required attenuation to prevent aliasing components from appearing in audio passband, ie > 45kHz).


*1977 (Snowbird, Utah): AES Standards Committee Meeting
Ampex want 48kHz
SMPTE want 60kHz
3M want 50kHz
Soundstream want 50kHz
BBC/EBU want 32kHz (broadcast Sat uplink; Euro telecom at 8kHz)
Matsushita want 44.1kHz
Philips/Sony want 44.1kHz
Technics want 49kHz
Denon want 47.25kHz
Mitsubishi/Teac want 47.5kHz
Toshiba want 50kHz
Hitachi want 35kHz


*1954:
RCA convinces FCC (NTSC) to adopt frame-drop system so that sound sub-carrier frequency did not produce visible beat in color sub-carrier frequency. FCC refused to change sound sub-carrier, but agreed to change color sub-carrier.


*1941:
FCC (NTSC) adopts field refresh rate exactly matching 60Hz AC power in U.S., to avoid beating intermodulation (rolling bars on screen).

[gmarinov]

Originally Posted by tom_lowe

The problem with your theory is that NTSC actually runs at 29.97 (my theory is that people the other side of the pond can't actually count properly) which would be 48 000 / 29.97 = 1601.6016 samples per frame, so sorry, but your theory just doesn't add up.

http://archive.digidesign.com/support/docs/FilmSync.pdf

[ErikG]

Here's the important part when it comes to what difference it might actually make to audio:

4) Sampling frequency should be twice that (Nyquist theorem) of 22.5kHz (20kHz human hearing range + 2.5kHz, so anti-aliasing filter has required attenuation to prevent aliasing components from appearing in audio passband, ie > 45kHz).

My understanding is:
The higher the sampling rate the smoother the digital (and also any analog) filtering will be. Look at a properly designed parametric EQ. If you put a Lowpass filter in, the steeper you make it the worse gets the resonanse and phase issues in the top end (near the filter frequency). So the main benefits of higher samplingrates as I see it is not to actually "hear" the stuff above human hearing, but the fact that digital audio REQUIRES filtering to recreate proper audio signals. This filtering is better of the less intrusive (steep) it has to be.

The other possible use for high sampling rates is in sound design, when you capture audio that is at a higher frequency than what we can normally hear and slow it down/resample you can hear stuff you never heard before, not always very interesting, but sometimes. Another benefit of using high sample rates is that resample, and pitch stretch etc often sound better if performed at a higher sample rate.

[musikwerks]

Originally Posted by terminal3

All the HD video I post here is 48khz. Just as with DVD, HD video formats may support 96khz, but old habits die hard.

I was going to add the same thing. HDCam SR still only records audio at 48khz.

48khz is here to stay.

[musikwerks]

Originally Posted by gsilbers

i thought 44.1k was because its double the sampling rate of the audible human hearing frequency range. nyquist theory and all. but it does have some extra frequencies there so if u would care to elaborate, sounds interesting.

Correct me if I'm wrong but the sample rate has nothing to do with the human hearing range. It's how many times the sound wave gets "looked at" by the digital conversion system, how many cycles per second. Nothing to do with sonic frequencies.

[Whitecat]

Originally Posted by musikwerks

Correct me if I'm wrong but the sample rate has nothing to do with the human hearing range. It's how many times the sound wave gets "looked at" by the digital conversion system, how many cycles per second. Nothing to do with sonic frequencies.

Yes, but the reason that the "optimal" sampling rates are above 40+khz is very much related to that - human hearing never goes beyond 20khz, and Nyquist's theory states that you need to double the highest audible rate in order to reproduce it properly. So in order for things to sound "real" they have to be sampled at double the frequency (or frequencies) that they exist at.

Read here: http://www.lavryengineering.com/docu...ing_Theory.pdf

[musikwerks]

Originally Posted by terminal3

Yes, but the reason that the "optimal" sampling rates are above 40+khz is very much related to that - human hearing never goes beyond 20khz, and Nyquist's theory states that you need to double the highest audible rate in order to reproduce it properly. So in order for things to sound "real" they have to be sampled at double the frequency (or frequencies) that they exist at.

Read here: http://www.lavryengineering.com/docu...ing_Theory.pdf

Cool. Thanks.

[cinealta]

Originally Posted by ErikG

So the main benefits of higher sampling rates as I see it is not to actually "hear" the stuff above human hearing, but the fact that digital audio REQUIRES filtering to recreate proper audio signals.

Precisely. The main benefit of higher sample rates (> approx 45k) is not to enhance the fidelity of what's being recorded, but to stop the artefactual degradation from digital filtering (to send aliasing etc higher out of audible range).

[mikevarela]

Recording at 96k almost seems like a futile effort when you consider the audio chain and the final delivery.

Imagine the playback scenarios... home theater in a box, cinema with 10 year old drivers, ipad with headphones and movies on an airplane.

If you can hear the clarity in any of those environments, then you should be a highly paid acoustical engineer.

Also, think about bit reduction.. as the medium of delivery becomes more aligned with internet streaming, doubling the sample rate just doesn't make economic sense.

Where I CAN see a use is in preservation transfers. Anything else seems academic at best.

[Etch-A-Sketch]

Originally Posted by tom_lowe

The problem with your theory is that NTSC actually runs at 29.97 (my theory is that people the other side of the pond can't actually count properly) which would be 48 000 / 29.97 = 1601.6016 samples per frame, so sorry, but your theory just doesn't add up.

NTSC 29.97 is still 30 frames per second with a 0.1% pulldown. So the theory still adds up.

[bizzle]

96k for broadcast? Forget it, I'm gonna low pass most everything anyway. Plus, you have to look at the whole delivery chain, most importantly the final audio playback system capabilities.

The name of the game is intelligibility, and 96k only makes bigger files, not better sounding productions. This is truly where the audiophile ideals don't fit the practical goals. (For specialized music or high end demonstration purposes it is perhaps something much different. I'm in the middle of a music project at 96k that sounds amazing.) I will however, argue that the difference between 16 and 24 bits, is way more important than 48k vs. 96k.

[cinealta]

Originally Posted by bizzle

I will however, argue that the difference between 16 and 24 bits, is way more important than 48k vs. 96k.

No need to argue, it's a given from audio theory. 16 bit is 96 dB of dynamic range (65k discrete levels of sound), while 24 bit is 110 dB of dynamic range (16.8 million discrete levels of sound). No comparison. In addition, 24 bit pushes the noise floor down further than 16 bit.

Whereas, the difference in audibility of 48k vs 96k, is arguable, with no concrete basis from audio theory, nor that of the human hearing range.

[apple-q]

Originally Posted by cinealta

No need to argue, it's a given from audio theory. 16 bit is 96 dB of dynamic range (65k discrete levels of sound), while 24 bit is 110 dB of dynamic range (16.8 million discrete levels of sound). No comparison. In addition, 24 bit pushes the noise floor down further than 16 bit.

Just a side-note:

It´s not "in addition". Higher quantization leads to lower quantization noise meaning more (theoretical) dynamic range. But since there are very few true 24bit converters the difference is often theoretical. On machines like a Zoom H2 for example 24bit is simply a waste of space since the analogue section is so noisy you simply fill the additional 8bits with noise. Recording at 16bit will not make a practical difference.

[tom_lowe]

Originally Posted by cinealta

No need to argue, it's a given from audio theory. 16 bit is 96 dB of dynamic range (65k discrete levels of sound), while 24 bit is 110 dB of dynamic range (16.8 million discrete levels of sound). No comparison. In addition, 24 bit pushes the noise floor down further than 16 bit.

Whereas, the difference in audibility of 48k vs 96k, is arguable, with no concrete basis from audio theory, nor that of the human hearing range.

24bit in THEORY will actually give 144dB of SNR (dynamic range) - 6dB per bit. Of course, no converters actually give this, most being between 110dB and 120dB

[pechnatunk]

Informative thread - I thought another reason (the main one for me) for higher sample rates and bit depths was in preparation for digital processing.

In the box - doesn't any effect, even as simple as an EQ - or mixing two sounds together for that matter change the information, therefore needs to rewrite the samples. So a higher res will retain better quality after lots of processing.
I thought that is what was meant when DAWs say 32 bit float etc etc.

If you took a sound and repeatedly applied +6 db of a certain frequency then -6 db of the same frequency over and over again would it degrade the sound.

Even when reversing a sound (in music) - if I didn't like it i'd make sure I pressed Undo rather than reverse it back - thinking it would reduce the quality - but yes, thinking about it - that shouldn't have any effect at all.

[8BallProductions]

Originally Posted by dr.sound

Gsilbers,
Yes, We DO need them much more than you realize!
They do quite a lot in keeping and setting standards.
When everyone delivers on HiDef without Film it's going to
be the "Wild West" without a Sheriff.
Standards are important. Dolby helps in the setting of standards.

44.1 for CD was originally done to fit a specific Concerto onto a CD.

48k was implemented because of storage.While the cost of storage has changed in recent years 48/24 is fine for where we are right now.
Studio's who make movies care more about 3D and selling more tickets,
not audio.

Yes the doc is right here... how ever what dolby dont do is check that cinema's are putting out the film at the same high standards .... great example EVERYMAN CINEMA's in the UK a supposedly high end boutique cinema for film lovers, yet play they're films through what i can only describe as a marshall stack or some kind of 150w PA system ... 48k, 44.1khz 96 ... on a S**T system they all sound the same.