Why is the given sample rate for TV 48 kHz and HD 96 kHz? - Page 3 - Gearslutz Pro Audio Community
15th January 2011
#61

Quote:
Originally Posted by Etch-A-Sketch
you know I never heard an official answer on this...but I always assumed it was because of the film speeds.

A little basic math tells a lot...

48,000Hz / 30fps = 1600 samples / frame

48,000Hz / 24fps = 2000 samples / frame

44,100Hz / 30fps = 1470 samples / frame

44,100Hz / 24fps = 1837.5 samples / frame

This may have nothing to do with it at all. Like I said, I've never heard any official word as to why film and TV uses 48KHz instead of 44.1KHz... but if you look at the timing, when shooting film at 24fps 44.1KHz audio won't sync perfectly.
No, films are shooted in 23,976 fps and gears like Sound Devices know how to sync 23,976 fps with 44,1 and you'll always have dropping, it's not such a big issue, can be resolved.

44,1 because SDDS can print digital signal encoded at 44.1khz, on 35mm film
96 because DTS standards are 24/96khz
48 because Dollby Digital standard delivery is 24/48 and because Broadcasting standard is 48khz and no, it's not about float because when you speak about broadcasting you speak 25fps and 30fps, not 24fps, not 29,970fps, not 23,976fps. If they would care about your concept => 44,1 at 25fps = 1746 and 44. at 30fps=1470: so no float here.

Why they set this standards ? I think for marketing. It really sound better ? Maybe you'll feel differences on a nice acoustic room on some nice speakers, PMC, B&W bla bla.

It's worth ? Probably if you watch the movie on 1080p on a nice anamorphic screen while listening 24/96 on 5.1 speakers from PMC, in a nice acoustic space.
15th January 2011
#62
Lives for gear
1) Films are actually shot at 24fps, only 23.98 if shot on an HD format

2) DTS is NOT 96K, only the 24/96 variant of DTS.

3) SDDS is hardly ever used for any cinematic releases anymore, and 44.1 is actually more to do with music CDs

4) Dolby Digital is 48K 16-bit

5) 29.97 is mentioned in broadcast, but often rounded up to 30fps for simplification
15th January 2011
#63
Lives for experience

Wow, here is the resurrection of a thread with a lot of misinformation in it.

The origin of 44.1 and 48 had nothing to do with marketing, or with an Executive wanting to be able to hear the entire Beethoven's 9th on a disc. These are the old wives tales of the audio industry.

I don't know if anyone noticed cinealta's posts, they pretty much give the answers. These things were actually decided by engineers. As he points out, the origin of 29.97 was to introduce Color into a Black and White signal infrastructure. But putting it at 30 fps caused a visible beating in the signal, so they pulled it down to 29.97 sub carrier so they could have Color without causing the B&W to beat.

The origin of 44.1kHz for the compact disc is due to early days in PCM digital where hard drives and such did not have the 1 Mbps per audio channel bandwidth or the capacity to store this much information. So they turned to a "pseudo-video" black and white binary waveform solution on video tape. This "binary" system is constrained by field rate and field structure. (As you know, in video, a frame is comprised of fields). To video was how the first Compact Discs were pre-mastered before they went to the CD manufacturing plant. And thus we have two standards to adhere to: 525 lines at 60 Hz and 625 lines at 50 Hz. Remember, this is B&W, so we are talking 30fps, NOT NTSC color.

In 60 Hz, there are 35 blank lines, leaving you 490 lines per frame or 245 lines per field, so the calculation for the sampling rate is simply : 60 X 245 X 3 = 44.1

With 50 Hz video, there are 37 blank lines, giving you 588 active lines per frame, or 294 per field. The math is thus : 50 X 294 X3 = 44.1 Khz.

Early Mastering Engineers in the U.S. had to make sure their machines were locked to a 60 Hz signal and not an NTSC Blackburst signal.

As to 48kHz, actually 60 kHz was considered to avoid any "leap frames", but remember we are talking very early PCM days (for the TV SR, this is the late '70's), so 60 was difficult to obtain as a sampling frequency and seen as wastefully high to the early digital engineers. Soundstream and 3M were using 50 kHz, and proposed taking 5005 samples over 6 fields of NTSC. But it did require "leap frames" for 525/60 monochrome B&W TV. The number of digital audio samples per video frame is an important number. If it is not an integer, then then you have put a different amount in some of the frames relative to the rest. If you create a "leap frame" (think like "leap years") you have to give it a digital flag. BBC/EBU was using 32kHz at the time, but this was not acceptable, so it favored 48 kHz because of the simple 3:2 relationship with 32 kHz and it caused only leap frames in NTSC, which was not the video standard used in Europe. Also, by this time, Decca was creating producing software that operated at 48k. Let's not forget that the film rate is 24 Hz, easy to work with this rate and 48k.

This happend all around 1981 for the 48k standard to be adopted.
16th January 2011
#64
Lives for gear

Quote:
Originally Posted by minister
This happend all around 1981 for the 48k standard to be adopted.
Tom, you sure sound like you've been there
16th January 2011
#65
Gear Guru

Quote:
Originally Posted by Big Andy
Higher sample rate GREATLY adds to the top end. Listen to an orchestral recording at 96k then listen to a down sampled 44.1k version of the same piece. There is a great loss of depth and clarity on the top end. You can actually hear the room in a 96k recording.
Only if your converters or SRC are very broken. Either that or it is expectation bias.

Alistair
16th January 2011
#66
Gear Guru

Quote:
Originally Posted by tom_lowe
If you were to record everything at 96KHz and work at that throughout the production, you'd be future-proofing your work,
Of course not unless you think we are going to evolve higher frequency hearing.

Quote:
The maximum audible frequency for 96KHz is 48KHz
The highest audible frequency will never be much above 20Khz.

Alistair
16th January 2011
#67
Gear Guru

Quote:
Originally Posted by tom_lowe
Except there are audible benefits. yes, are ears will not pick up sounds at 48KHz (audible, not sampling frequency), but they will pick up the harmonics from such frequencies that filter down in to our audible range.
Yet to be proven. Actually, the science predicts something different so there is no reason to believe this.

Alistair
16th January 2011
#68
Gear Guru

Quote:
Originally Posted by tom_lowe
a sound recorded at 96K and down sampled to 48K will sound clearer than a sound recorded at 48K.
Non-sense. If the converters you use has good filters it will sound exactly the same.

Alistair
16th January 2011
#69
Gear Guru

Quote:
Originally Posted by tom_lowe
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)
The Beethoven part is a myth created by marketing. Full story here:

Shannon, Beethoven, and the Compact Disc

Alistair
17th January 2011
#70
Lives for experience

Quote:
Originally Posted by danijel
Tom, you sure sound like you've been there
Ha ha, no, not THAT old. I was in high school and more interested in guitars and music and girls. Didn't know jack about digital until many years later. I just try to learn all I can. And I became curious some years back about these standards and did some research, asked questions of some people who were there.

And the comment that Dolby is merely a crutch of mediocre technology is just plain ignorant.
17th January 2011
#71
Lives for gear
I always thought that recording at 96k would yield benefits if you were going to mangle the sound, pitch shifting/time stretching et al. Would this be true? Not had the chance to compare myself.
17th January 2011
#72
Lives for gear

Quote:
Originally Posted by ThisIsSka
I always thought that recording at 96k would yield benefits if you were going to mangle the sound, pitch shifting/time stretching et al. Would this be true? Not had the chance to compare myself.
If there´s no audio beyond 20Khz there won´t be any benefits from using 96k. SOme people argue that 96 offers "higher resolution" but higher sampling rates give you more bandwidth in the frequency domain. If your mic doesn´t pick up anything above 20k you are simply wasting space and bandwidth using 96k since there won´t be anything to capture up there.
It can have benefits for SRC processing tuning down for effects but actually only if your recording has information above 20k.

It´s a bit like recording at 24bit with a zoom H2. Since the mics are noisier than the benefit of 24bit you are just filling your 8LSBs with noise. But since without a 24-bit-sticker on the box you can hardly sell audio equipment today. On almost all of the cheap audio gear the 24/96 is simply a marketing requirement.

But this topic has been beaten to death and beyond...
17th January 2011
#73

Quote:
If your mic doesn´t pick up anything above 20k you are simply wasting space and bandwidth using 96k since there won´t be anything to capture up there.
FWIW most mics record just fine well above 20khz despite the spec sheets that the manufacturers put out. Pull down my audio samples from this post and chuck them into izotope to see that even the built in mics in many sub-\$300 handheld recorders have frequency capacity well beyond 20k.

I've personally recorded frequencies up to and past 48k with the following mics (off the top of my head):

PCM D50
Zoom H4n
DR-100
AT 4050
Schoeps CMC-6
MHK60
VP-88
sennheiser 421
cold gold contact mics

ThisIsSka is correct in that high sample-rate acquisition has sound designer benefits, though the sounds are usually then downsampled after pitch shifts and before delivery to the final mix.

This post actually has a slowed-down example of a metal gate recorded with the internals of the PCM D50. You can see how much high frequency content was recorded above 20k and is retained when the source is slowed down to 20%.
17th January 2011
#74
Lives for gear

Quote:
Originally Posted by renec
FWIW most mics record just fine well above 20khz despite the spec sheets that the manufacturers put out. Pull down my audio samples from this post and chuck them into izotope to see that even the built in mics in many sub-\$300 handheld recorders have frequency capacity well beyond 20k.

I've personally recorded frequencies up to and past 48k with the following mics (off the top of my head):

PCM D50
Zoom H4n
DR-100
AT 4050
Schoeps CMC-6
MHK60
VP-88
sennheiser 421
cold gold contact mics

ThisIsSka is correct in that high sample-rate acquisition has sound designer benefits, though the sounds are usually then downsampled after pitch shifts and before delivery to the final mix.

This post actually has a slowed-down example of a metal gate recorded with the internals of the PCM D50. You can see how much high frequency content was recorded above 20k and is retained when the source is slowed down to 20%.
Yep. All I said was that if you don´t have more than 20k in your source you don´t need 96k. I didn´t say you can´t capture that with hand held recorders.

I just used the H2 as an example of wasted bandwidth. On an H2 24bit is definitely just a waste of space. The analogue front-end is more noisy than what you gain by recording 24bit instead of 16.
17th January 2011
#75

fair enough. I'd probably never bother recording production dialogue at 96k for that reason, though it is kind of amazing how much other stuff really does live up there.
17th January 2011
#76
Lives for gear

Quote:
Originally Posted by renec
fair enough. I'd probably never bother recording production dialogue at 96k for that reason, though it is kind of amazing how much other stuff really does live up there.
hehe. lots of production sound isn´t even usable at 48k. I´m not looking forward to rustling lavs at 96k heh
18th January 2011
#77

Quote:
Originally Posted by tom_lowe

1) Films are actually shot at 24fps, only 23.98 if shot on an HD format

2) DTS is NOT 96K, only the 24/96 variant of DTS.

3) SDDS is hardly ever used for any cinematic releases anymore, and 44.1 is actually more to do with music CDs

4) Dolby Digital is 48K 16-bit

5) 29.97 is mentioned in broadcast, but often rounded up to 30fps for simplification
1) Ask a DP that use one of this camera: ARRI, Phantom, RED, Panavision, etc. what FPS they shoot if director doesn't ask slow motion in post production. In pro film shooting we don't talk about HD we talk about 2k,4k and aspect ratios. In post production, yes you can conform 4k progressive to 1080,720 progressive or interlaced.
2) I spoked about 24/96 witch is use on Blue Ray
3) Yes, but still it's used in some cinema projectors and also Dolby ask you to have an optical magnetic film solution for accreditation.
4) My fault, I knew it's 16, just hurry
5) I didn't work too much for americans broadcasting NTSC (never twice same color) so you probably have right.

Quote:
Originally Posted by UnderTow
Non-sense. If the converters you use has good filters it will sound exactly the same.

Alistair
Could have sense. I don't know, maybe the AD PLL work better at high freq. (96k) and you'll hear only in blind test. I'm not very curios.
18th January 2011
#78
Gear Guru

Quote:
Originally Posted by soulviasound
Could have sense. I don't know, maybe the AD PLL work better at high freq. (96k) and you'll hear only in blind test. I'm not very curios.
The sampler is always working at the same rate regardless of the destination format. All modern ADC's work at 128 or 64 times the base rate with something like 4, 5 or 6 bits. The signal is then decimated down to the target rate. The ADC's sampling and clocking circuitry will be running at the same rate whether you set the output rate to 96Khz or 48 Khz.

The main difference between sampling at 96Khz and subsequently converting separately to 48 Khz is the quality of the filter used in the SRC. These days modern converters have very good filters. The differences are really not that big and might be better than some software converters (and worse than others).

You would have to do some serious blind ABX testing to determine if the difference is at all audible. (And no, anecdotal evidence fraught with expectation bias does not count).

Alistair
15th October 2011
#79

Do you filter your air before you listen? I always do that as well as making certain that I have a specific humidity and temperature in the room.

I hear audiophiles are very specific about these things...
15th October 2011
#80
Lives for gear

Quote:
Originally Posted by Sargaroth
Do you filter your air before you listen? I always do that as well as making certain that I have a specific humidity and temperature in the room.

I hear audiophiles are very specific about these things...
I haven't worked in a single facility tyat does that.
And we do nt deal in teh audiophile world of \$200 volume knobs, and \$1000 a foot audio cables
15th October 2011
#81
Lives for gear

Fwiw the main difference between sample rates from what I can tell (but have never seen anybody else proclaim) is this :

Take a signal generator route it into your audio workstation at 44.1 or 48k. Send the output to an ocilliscope. Set the signal generator to square wave and sweep the audible spectrum. At 44.1 and 48k, a square wave gets rounded to a sine wave above 7k I think ( its been a while since I figured this out)

So essentially 44.1 and 48 cannot accurately reproduce complex waves in the last octave and a half of the audible frequency range.

Try the same at 96k, and a square wave becomes a sine wave at 15k, giving you an additional octave of accurate reproduction above 44 and 48.

This is why higher sample rates sound better. Not because they reproduce audio beyond our hearing range or the harmonics that may enter the audible spectrum it is it ability ( or lack of) to accurately reproduce what we can hear

Nyquist theorem states you need at least 2 points to reproduce a signal. I don't remember anything in there about it being able to accurately reproduce high frequencies with only 2 sample points.

Sent from my DROIDX using Gearslutz.com App
15th October 2011
#82
Gear Guru

Quote:
Originally Posted by cananball
Fwiw the main difference between sample rates from what I can tell (but have never seen anybody else proclaim) is this :

Take a signal generator route it into your audio workstation at 44.1 or 48k. Send the output to an ocilliscope. Set the signal generator to square wave and sweep the audible spectrum. At 44.1 and 48k, a square wave gets rounded to a sine wave above 7k I think ( its been a while since I figured this out)

So essentially 44.1 and 48 cannot accurately reproduce complex waves in the last octave and a half of the audible frequency range.

Try the same at 96k, and a square wave becomes a sine wave at 15k, giving you an additional octave of accurate reproduction above 44 and 48.
This is really basic audio: If you remove harmonics from a square wave (or triangle or sawtooth etc) the waveform gets closer and closer to a sine wave. So yes, absolutely! If you filter out the harmonics, you end up with a sine wave. That is exactly what your ear would do anyway so nothing is actually lost that can be considered sound.

Quote:
This is why higher sample rates sound better. Not because they reproduce audio beyond our hearing range or the harmonics that may enter the audible spectrum it is it ability ( or lack of) to accurately reproduce what we can hear
No, we can't hear the harmonics. That is the whole point. The harmonics, that which make the waves other than sine shaped, are harmonics and if they are above about 20Khz they are inaudible harmonics.

So no, increasing the sample rate does not in anyway make the capture or reproduction of the audible frequency range any more accurate.

Quote:
Nyquist theorem states you need at least 2 points to reproduce a signal. I don't remember anything in there about it being able to accurately reproduce high frequencies with only 2 sample points.
ANY frequencies within the band-limited signal! You simply can't have a 15Khz square wave within a 20 Khz bandwidth. Neither in digital audio, neither in analogue audio (you could use an analogue filter and do the same experiment), neither in the signal that is ultimately captured by our hearing system.

Those harmonics are filtered out because they are outside the audible range. The fact that you can see them on your oscilloscope doesn't mean you can hear them or that they have any effect on the audible range!

I give some more details and provide animated pictures etc in this post in yet another thread about sampling rates: https://www.gearslutz.com/board/7048758-post480.html

Quote:
Better not. ;-)

Alistair
15th October 2011
#83
Lives for gear

Quote:
Originally Posted by UnderTow
This is really basic audio: If you remove harmonics from a square wave (or triangle or sawtooth etc) the waveform gets closer and closer to a sine wave. So yes, absolutely! If you filter out the harmonics, you end up with a sine wave. That is exactly what your ear would do anyway so nothing is actually lost that can be considered sound.

No, we can't hear the harmonics. That is the whole point. The harmonics, that which make the waves other than sine shaped, are harmonics and if they are above about 20Khz they are inaudible harmonics.

So no, increasing the sample rate does not in anyway make the capture or reproduction of the audible frequency range any more accurate.

ANY frequencies within the band-limited signal! You simply can't have a 15Khz square wave within a 20 Khz bandwidth. Neither in digital audio, neither in analogue audio (you could use an analogue filter and do the same experiment), neither in the signal that is ultimately captured by our hearing system.

Those harmonics are filtered out because they are outside the audible range. The fact that you can see them on your oscilloscope doesn't mean you can hear them or that they have any effect on the audible range!

I give some more details and provide animated pictures etc in this post in yet another thread about sampling rates: https://www.gearslutz.com/board/7048758-post480.html

Better not. ;-)

Alistair
I'll hold off on the white paper!

15th October 2011
#84
Lives for experience

Quote:
Originally Posted by cananball
Before you do this, first order of business would be to understand square waves and what happens to their harmonics as you go up in frequency -- remember, you don't have infinite bandwidth, you've limited it by the sample rate.

Also, square waves do not exist in nature.

EDIT: Everyone should read Alistair's post that he links to. It took a lot time to produce and it makes things very simple and understandable. Thanks for doing that, Alistair!
16th October 2011
#85
Moderator

"This message has been deleted by minister. Reason: Nevermind... This thread is getting silly."

...really guys....

I can absolutely promise you I can most definitely hear the "room" when I'm in a theater or home watching TV... no matter what the sample rate. and by the way, the dog, and the neighbors, and the popcorn, and the texting, and the blown speaker on the back rear, and the over pumped sub, and the....

cheers
geo
17th October 2011
#86
Lives for gear

Quote:
Originally Posted by tom_lowe
so that sound has something to travel through :-P
heh
17th October 2011
#87
Lives for gear
Quote:
Originally Posted by georgia
I can absolutely promise you I can most definitely hear the "room" when I'm in a theater or home watching TV.
cheers
geo
I'm surprised how many movie theatres I've been to that are really echoey, echoey, echoey.
18th October 2011
#88
Gear maniac

96khz and higher sample rates are a useful medium to have available but IMO it is most useful for reprocessing. It's already not clear as day to hear 96khz to 48khz on monitors, much less end user environments.

The biggest thing I feel benefits from the higher sample rates are certain digital filters. The most obvious one to understand would be a speed up or slow down or sample rate conversion. The kind of aliasing that happens as the source and destination sample rates rotate in and out of alignment with one another is a source of minor headaches for me.

Pitch shifting 44.1khz sounds slower starts to incur high frequency artifacts pretty fast, so from a sound manipulation front, it's nice to have a high end storage medium that allows for remanipulation. Just as 16 bit sounds fine for 99.9% of end users, if you need to gain up that audio for reuse, you will find your low-level audio to not have a lot of play room.

But, as an end storage medium, 48khz 16bit dithered is adequate for any consumer. Pros that require additional resolution for their work will find plenty of use for 96khz 24bit, but it's such a long stretch to say it's required. 4k images are not needed to make a Blu Ray look nice on your big screen. 96khz is not required to hear great sound in a film soundtrack. Some of those who chase "the best" will always attempt to make the argument that the ultimate standard is required. When the difference is hundreds to thousands of bucks for audio a consumer may not be able to notice, the reality of that technology being embraced is unclear.
19th October 2011
#89
Lives for gear

Quote:
Originally Posted by jampottt

That's my point, why does it sound better? What makes it better than using 44.1 kHz?
the higher the sample rate the more accurate the a/d/a will be (within limits of voltage resolution and clock jitter)

it has nothing to do with more bandwidth
that would be filtered off at some point anyway
it has all to do with a/d/a accuracy
9th November 2011
#90
Gear interested

Quote:
Originally Posted by soulviasound
for bats or for us ?
LMAO

Welcome to Gearslutz Pro Audio Forum!

Registration benefits include:
• Ability to make and reply to posts
• Interact with VIP industry experts in our guest Q&As
• Get instant lifetime access to the world's best private pro audio Classifieds for only USD \$20
• Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.

Replies
bigteets922 / High end
2
Kestral / High end
45
Azimuth / So much gear, so little time!
9
Carlos del Pino / Mastering forum
13

Forum Jump
Forum Jump

FAQ Search Today's Posts Mark Forums Read