sample rate for cinema, tv, soundtracks audio hd

[andy3]

Hello guys I'm going to recording my music.
I want to keep the option to fit my tracks on as many formats and situations as I can. (dvd, cinema, cd, tv, mp3, radio etc)

What is the simple rate I need to go for?

24 at 48 000?

New tech is coming up, and I don't want re-record my stuff!
Since I have to record I want to have tracks ready for every possible situations.

Thank you!(Really)

[Adam Dempsey]

The way I see it is if you go to 24/48k then you need convert to 44.1k only for compact disc, or low bit rate mp3's.
In my experience most clients doing video request 16 bit / 48k, and often a -10dBFS (peak) version.

[huejahfink]

Originally Posted by Adam Dempsey

16 bit / 48k, and often a -10dBFS (peak) version.

Yup, exactly this.

[Dave Rose]

Originally Posted by Adam Dempsey

The way I see it is if you go to 24/48k then you need convert to 44.1k only for compact disc, or low bit rate mp3's.
In my experience most clients doing video request 16 bit / 48k, and often a -10dBFS (peak) version.

Pretty much same here, although last time I did any TV stuff I was asked for 24/48 but peaking at -12dbfs

[andy3]

I would like to know a good for all sample rate.
I mean a sample rate where I can resampling without problem for anysituations from cd to hd tv etc, without loosing quality. without re-recording

thank you

[aleatoric]

Originally Posted by andy3

I would like to know a good for all sample rate.
I mean a sample rate where I can resampling without problem for anysituations from cd to hd tv etc, without loosing quality. without re-recording

thank you

I'd just record at 24/48khz. 48khz is the standard for film, tv, etc. If you then go on to do a CD release you can sample rate convert to 44.1khz using a high quality sample rate converter like Izotope 64-bit SRC or Voxengo's r8brain. Alternatively you could record at 24/96khz if you have the hard drive space and computer power and do the same as above and sample rate convert down to meet different standards (sample rate convert down to 48khz when required, 44.1khz, etc.).

[wado1942]

In a situation like that, I go for 96KHz. Most movies are recorded/mixed at that rate any way. It also converts to 44.1KHz better than 48KHz does.

[jayfrigo]

It has been my experience that post is still primarily a 48k world, 96k being used sometimes for scoring or at some intermediate stage, and that conversion from 48 to 44 is no worse than 96 to 44 when using a good SRC. Sill, keeping a 96k archive is not a bad idea, because then you can cover any rate asked for, now or for a future hi-res release.

[dcollins]

Originally Posted by wado1942

In a situation like that, I go for 96KHz. Most movies are recorded/mixed at that rate any way. It also converts to 44.1KHz better than 48KHz does.

I think everything is done at 48k, and there is no real difference in non-integer sampling rate conversion.


DC

[Silvertone]

I deliver everything at 16/48... 15 years and running now. Needless to say I don't delver at -10 dBfs... I'm mastering these music and SFX cues.

Just picked up a couple film companies over in England and they do 16/48K as well.

Some are now just starting to ask for 24/48K as well.

Many of these companies tell me the same thing... some editors are so lazy that they have to watch that the MP3 cue (used to audition the sound/music file) doesn't end up in the final film instead of the 16/48K file!

Here is one of the companies I do work for http://robertetoll.com/ Listen to the Visual Trail Montages (I worked on all this music and cues).

Hope that helps.

[nikkelk]

I'm working in a rather bigger post production company and we happen to deliver at 16/48 also.

Nevertheless some of our clients require different specs. for their material.
If you are not sure you should simply ask a technical specification sheet from your client.

Volume depends on the media it is for.

Do some of you have clients who ask for videos they can post on the internet? ( ads, informational,. . . )
If yes, how do you limit the file? (-10dbfs) or dou you go louder, lower,...?

[PaulMac]

As others have said, 48/24 or 48/16 is the standard in film. Although, I tend to work at 96 and convert down, just as I work at 88.2 for CD and then convert to 44.1.
I'd rcommend recording at 24bit even if they eventually are just looking for 16, as you can just dither the stems/master at your leisure.

[wado1942]

I think everything is done at 48k, and there is no real difference in non-integer sampling rate conversion.

Mostly 48K 16-bit or 24-bit, but more and more are going to 96K 24-bit.

My experience tells me different as far as downsampling goes. People argue that "with an ideal SRC, there's no difference". Unfortunately, ideal SRCs don't exist. There's always some kind of compromise.

some editors are so lazy that they have to watch that the MP3 cue (used to audition the sound/music file) doesn't end up in the final film instead of the 16/48K file!

I've had instances where the director would ask for a rough mix of what I had so far for timing on set and in editing. Then a few weeks later, I gave him the final mixes and he was like "What are these?"
"The mixes."
"What do you mean?"
"The final music mixes for the movie."
"I already have those."
"No, I gave you just some basic tracks of what I had like you asked."
"Oh, well, I already gave the final cut of the movie to the plant."


I'm almost glad he listed me as an electrician (which I was) but not audio engineer because the band wasn't even done RECORDING when he got those rough mixes. Upon seeing the final cut of the movie in a theater, I was somewhat relieved to find that my rough mixes were the least of their problems. What kind of idiot does a 4x digital zoom in editing?

For the record though, I usually submit 24-bit 48K audio when other engineers are doing the final mix. It's rare enough that they even have proper audio monitors, I don't take my chances on a proper post-production work flow.

[dcollins]

Originally Posted by wado1942

My experience tells me different as far as downsampling goes. People argue that "with an ideal SRC, there's no difference". Unfortunately, ideal SRCs don't exist. There's always some kind of compromise.

There is no ideal anything, engineering is the art of compromise................

Here is a representative FFT of 96 to 44k1 conversion. We could easily call it 'perfect.' The 96k to 48k is the same, or maybe even a little worse.

DC

[jayfrigo]

Originally Posted by wado1942

There's always some kind of compromise.

There are compromises with both the 96 and 48 source, and certainly there are some SRCs out there that look like one's pet cat could have coded them, but with the competent ones, the compromises should be and usually are quite similar, whether going from 96 to 44.1 or going from 48 to 44.1.

However, if you hear differently on the particular SRCs you've tried, there very well could be be something going on, so trust your ears and use what sounds better to you. I wouldn't necessarily generalize it to all SRCs however.

[wado1942]

I've tested a number of SRCs and I always trust my ears over graphs & charts. I really don't care what the FFT says because all it can do is help you SEE what you're hearing.

That said, of three different SRCs I critically tested, ranging from lousy to quite good, 2:1 and higher conversions always more closely resemble the unconverted version than less than 2:1 conversions and logically so. The ones that do more accurate filtering have more pre-ringing. You can counter that by leaning more towards minimum phase but that means more phase shift. I noticed there tend to be fewer overshoots with gentler, LP filters doing 2:1 conversions, which is useful for oversampled clipping/limiting. Higher quality converters tend to use a lot more processing power. Like I said, always a compromise.

[dcollins]

Originally Posted by wado1942

I've tested a number of SRCs and I always trust my ears over graphs & charts. I really don't care what the FFT says because all it can do is help you SEE what you're hearing.

That's fine. The FFT is a good way to get an idea of the quality of conversion, though.

I just didn't want people to get the idea that 96k is popular for TV/video use or that today's conversion is that different at non-integer ratios.


DC

[Bob Olhsson]

I believe 48x16 was an early limitation of final cut pro. When I was doing mostly post work before final cut showed up, everything I saw was 48x24.

[andy3]

Thank you!

[wado1942]

Originally Posted by Bob Olhsson

I believe 48x16 was an early limitation of final cut pro. When I was doing mostly post work before final cut showed up, everything I saw was 48x24.

By far the most over rated editor on the planet.

[mr.farkendorfer]

Originally Posted by andy3

Since I have to record I want to have tracks ready for every possible situations.

Thank you!(Really)

Hi,

If you want to be ready for "every possible situation", you should record / mix / master at the highest sample rate and bit depth.

You can always re-sample [re-record] your hi-rez version at any of the lower sample rates and / or bit depths that someone may ask for. That remains the most elegant solution. Despite the "techie" fascination with sample rate conversion, simple re-recording with a good set of converters will probably give you better results. There are some very good sample rate converters [Weiss], but they are expensive. Even the best will be hard pressed to do as well as simply re-recording through great converters.

Alternatively, when you do your final mix, just do a separate mix, in the first instance, at each of the available sample rates and bit depths. Then you are "ready for every possible situation" [but, eventually, someone may still ask for something weird that you didn't think of ].

Obviously, if you are doing a project where you know in advance what sample rate and bit depth will be required, then make sure you do a mix that complies with the requirements.

Also, beware of graphs and examples using only a test tone [or tones]. That's not necessarily the same as music. Not saying you can't learn anything that way, but its not the "be all and end all".

[dcollins]

Originally Posted by wado1942

By far the most over rated editor on the planet.

Walter Murch seemed to muddle through.


DC

[Herbeck]

Originally Posted by wado1942

The ones that do more accurate filtering have more pre-ringing. You can counter that by leaning more towards minimum phase but that means more phase shift.

Here is a test of the new filter that I used in the crossover earlier.
But now I used it in a SRC.
No pre-ringing and far less phase shifting than a min-phase filter.
The test also include a lin-phase and a min-phase conversion (Sox) for reference listening.
Any comments would be useful for further tweaking and development.

https://www.yousendit.com/download/Z...cG9PSHcwTVE9PQ


All the best

Herbeck

[andy3]

ty

Originally Posted by mr.farkendorfer

Hi,

If you want to be ready for "every possible situation", you should record / mix / master at the highest sample rate and bit depth.

You can always re-sample [re-record] your hi-rez version at any of the lower sample rates and / or bit depths that someone may ask for. That remains the most elegant solution. Despite the "techie" fascination with sample rate conversion, simple re-recording with a good set of converters will probably give you better results. There are some very good sample rate converters [Weiss], but they are expensive. Even the best will be hard pressed to do as well as simply re-recording through great converters.

Alternatively, when you do your final mix, just do a separate mix, in the first instance, at each of the available sample rates and bit depths. Then you are "ready for every possible situation" [but, eventually, someone may still ask for something weird that you didn't think of ].

Obviously, if you are doing a project where you know in advance what sample rate and bit depth will be required, then make sure you do a mix that complies with the requirements.

Also, beware of graphs and examples using only a test tone [or tones]. That's not necessarily the same as music. Not saying you can't learn anything that way, but its not the "be all and end all".

[wado1942]

Originally Posted by dcollins

Walter Murch seemed to muddle through.
DC

I'm not saying it can't be used, I'm just saying industry standards don't always deserve their status. There's better software out there for the money. It's easier to use and you can do more with it.

[dcollins]

Originally Posted by mr.farkendorfer

If you want to be ready for "every possible situation", you should record / mix / master at the highest sample rate and bit depth.

Probably true although there is certainly a legitimate question about whether more == more with regard to sampling rate.

Despite the "techie" fascination with sample rate conversion, simple re-recording with a good set of converters will probably give you better results. There are some very good sample rate converters [Weiss], but they are expensive. Even the best will be hard pressed to do as well as simply re-recording through great converters.

Not true. You can get free SRC that will be far more transparent than even the best A/D/A chain. If your claim was correct you would see people using that technique instead of dedicated SRC systems.

Also, beware of graphs and examples using only a test tone [or tones]. That's not necessarily the same as music. Not saying you can't learn anything that way, but its not the "be all and end all".

Actually, this is one case where two or three graphs can give you complete picture of the transparency / artifacts of the system as well as showing you which ones are not even worthy of a listening test.........................


DC

[mr.farkendorfer]

Originally Posted by dcollins

Probably true although there is certainly a legitimate question about whether more == more with regard to sampling rate.

I'm not sure what you "= =" is supposed to mean. But there's not really any legitimate question in my experience, at least not among people that have much experience using the highest sample rates.

There are rather obvious advantages to using e.g. 192kHz, if you are able.

Originally Posted by davecollins

Not true. You can get free SRC that will be far more transparent than even the best A/D/A chain.

Well, without any specificity, your statement is kind of vapid. Hmmm. I disagree with your statement, although I agree that sample rate converters have improved and that there are good ones. I don't know which one you are talking about, but I think Daniel Weiss makes a good one.

Originally Posted by davecollins

If your claim was correct you would see people using that technique instead of dedicated SRC systems.

That's a ridiculous statement, whichever position you take. I can't believe you posted it. FWIW, lots of people do simply re-record rather than use "dedicated SRC systems". So, if your criteria for whether it is a worthy technique is really whether or not people use it, they do.

Originally Posted by davecollins

Actually, this is one case where two or three graphs can give you complete picture of the transparency / artifacts of the system as well as showing you which ones are not even worthy of a listening test.........................


DC

Again, the statement is so nonspecific that it is basically meaningless, and somewhat argumentative. I know you really like graphs, and I really have nothing against graphs, I guess.

But graphs of what?

Yeah, if a piece of gear can't handle a simple test tone, its probably crap.

But maybe you think that a test tone tells you exactly how audio equipment will perform with actual music use? If so, I disagree.

[dcollins]

Originally Posted by mr.farkendorfer

I'm not sure what you "= =" is supposed to mean.

It's short for "equals" and is an equivalence operator. Certainly as a coder of the well-known Karyokiedokie d*ther you must have used it in your C++ programming?

But there's not really any legitimate question in my experience, at least not among people that have much experience using the highest sample rates.

There are rather obvious advantages to using e.g. 192kHz, if you are able.

What are the advantages?

Well, without any specificity, your statement is kind of vapid. Hmmm. I disagree with your statement, although I agree that sample rate converters have improved and that there are good ones. I don't know which one you are talking about, but I think Daniel Weiss makes a good one.

The R8brain free system is better than any A/D/A cycle. Perhaps you should try it.

That's a ridiculous statement, whichever position you take. I can't believe you posted it. FWIW, lots of people do simply re-record rather than use "dedicated SRC systems". So, if your criteria for whether it is a worthy technique is really whether or not people use it, they do.

There is nothing inherently wrong with going A/D/A, but digital can be designed to arbitrary transparency and analog cannot. Simple as that.

Again, the statement is so nonspecific that it is basically meaningless, and somewhat argumentative. I know you really like graphs, and I really have nothing against graphs, I guess.

But graphs of what?

The frequency sweep, the high and low level FFT, and squarewave.

Yeah, if a piece of gear can't handle a simple test tone, its probably crap.

What do you mean by "handle?" To pass it from input to output? To do so with one part per million error?

But maybe you think that a test tone tells you exactly how audio equipment will perform with actual music use? If so, I disagree.

Not exactly, but once you learn more about how to interpret the wide variety of measurements available today, you actually can get a good idea of the performance before listening.

Every so often there are surprises, but in digital all distortions sound bad, so the lower the better.


DC