96 Khz 24 bit conversion

[nalooti]

Hi,

For some reason, my Olympus LS-P2 recorder does'nt allow recording at 24 bit/44.1 Khz. It does allow 96 Khz/24 bit and 48 Khz/24 bit among others.

I've heard that 96 kbps is useless and it's better to use less. Could you please explain why (except huge file sizes) and should I go with 96, 48 (DVD) or 44.1 (CD)kbps frequency (still with 24 bit depth) in order to be able to play it anywhere with good (music) quality?

How can I reduce the sampling frequency from 96 to 44.1 Khz for example using Reaper ?

Many thanks

[DVDdoug]

The only downside to higher resolution is bigger files (and slower processing).

and should I go with 96, 48 (DVD) or 44.1 (CD)kbps frequency (still with 24 bit depth) in order to be able to play it anywhere with good (music) quality?

It depends on what final-format you want. FLAC? CD? DVD? MP3? AAC?

[nalooti]

Originally Posted by DVDdoug

It depends on what final-format you want. FLAC? CD? DVD? MP3? AAC?

I want WAV instead of FLAC, not all computers have FLAC players.

I want to have them on my hard disk (itunes library) uncompressed, and to be able to convert them to mp3/aac for portability and easier sharing when needed.


After some though (and tests) I may stay with 96 khz. In this case can itunes play them ?

How would I then convert them to mp3 for example ? Can Reaper do it directly (i.e. convert 96khz/24bit PCM to 256 kbps MP3) or should I first convert to 44.1khz ? how Reaper can do that ? just by drag&drop to a project set to 44.1 khz ?

Finally, how much CPU is needed for higher sampling rate (96khz) for importing to reaper and playing ? Please give me CPU power indications compared to a 44.1 khz file.

thank you

[drumphil]

In general doubling the sample rate roughly doubles the CPU usage.

I'd just use 48K, as your recorder and reaper are both happy working at that sample rate, and render to 44.1K at the end of the process if that is your target sample rate.

You could use a resampling utility to convert your recorded material to 44.1K before you import the files into reaper, but you don't really gain anything by working that way, and it's one more thing to do that isn't necessary.

There are a small list of real reasons to use higher sample rates, as follows:

You want to pitch shift down something from the ultrasonic frequency range in to the audible frequency range. (making sound effects for a video or something like that)

You want to achieve lower round trip latency from your audio interface when it doesn't allow you to set a smaller buffer size at normal sample rates.

(eg, the interface has a fixed minimum buffer size of 64. If you run at a higher sample rate the round trip latency at 64 samples will be lower. It would be better however if the interface offered a 32 sample buffer at the lower sample rates)

You are using a plugin processor or synth that doesn't properly filter its output, and sounds better when run at a higher sample rate.



Only the first of those is a situation where using a higher sample rate is the only and proper answer. The other two are workarounds for other things that don't work as optimally at lower sample rates as they could. The plugin could properly filter its output, and the audio interface could offer a smaller buffer setting at low sample rates, but if you want to pitch shift down stuff from the ultrasonic frequency range, you have to record it in the first place, which you can only do by running at a high enough sample rate to record those frequencies.

[grinder]

Nothing wrong with recording in 24BIT 96kHz or 1696
Reaper can handle the downsizing later on.
All depends on how powerful your computer is how much RAM it has,
how much memory in your hard drives you have.
Do not let this scare you your computer will soon tell you if it has "had enough".
Let everybody know what computer you have etc.
If you wish to record at 48 then record at 48 that can be made smaller at the end of your work as well.

Grinder

[Philbo King]

It's common for source video to be at 48k or 96k. Those rates correctly sync to video frame rates without needing to drop frames.

The merits of high sample rates has been discussed to death on the web... Basicly, no one has been able to demonstrate in blind listening tests that they can tell one from the other. But there are people who use it anyway.

Given the wording in your post, I'd suggest using the 48k rate, then dragging and dropping the file into Reaper, where you can easily render the sound into whatever format and rate you want. I use 44.1k/24 bit for all my audio mixes.

[mschnell]

Originally Posted by drumphil

In general doubling the sample rate roughly doubles the CPU usage.

With FFT based tools (such as Convolution Reverb or FIR Filters) it by far more than doubles the CPU usage.

-Michael

[Cableaddict]

Originally Posted by Philbo King

.

The merits of high sample rates has been discussed to death on the web... Basicly, no one has been able to demonstrate in blind listening tests that they can tell one from the other. But there are people who use it anyway.
.

That's true for raw audio, such as the OP is talking about.

Careful, though, as it's definitely NOT true for full sessions. Many plugins, and MOST VSTi's and guitar amp simulators, all sound better as a high SR. that HAS been proven with double blind tests, and there's good technical reasons for it.

Also, propagation delay goes down, so for live performance, (with VSTi's and no delay compensation) an 88.2K session with a 256 buffer will actually be faster (overall) than a 44.1K session with a 128 buffer. (If your CPU can handle it)


Apples 'aint always apples.

[grinder]

Nalooti
Reaper can convert your 48 or 96Khz wave files to mp3 and other bit and kHz values.
Time to read the Reaper manual.
I have video on Utube and audio on Souncloud along with many other people, the only way they got there was mainly Reaper.

Grinder

[drumphil]

Originally Posted by grinder

Nothing wrong with recording in 24BIT 96kHz or 1696
Reaper can handle the downsizing later on.
All depends on how powerful your computer is how much RAM it has,
how much memory in your hard drives you have.
Do not let this scare you your computer will soon tell you if it has "had enough".
Let everybody know what computer you have etc.
If you wish to record at 48 then record at 48 that can be made smaller at the end of your work as well.

Grinder

Well, there kinda is something wrong with recording at 96K. Unless you need to for one of the reasons I outlined above, all you are doing is wasting storage space and processing power for no good reason.

[nalooti]

Thank you very much all for your suggestions.

In case it makes any difference I'm going to record a jazz big band and this is the first time I'm using the recorder.

[grinder]

Hate to burst your bubble drumphil I can actually hear the difference
between the different rate Khz files.
At no time have I suggested to natlooti that he has to record at 96Khz
Other people who have years of intense engineering and mastering experience also have the opinion 96Khz is good/better who am I to argue and my testing has shown that I can in fact hear the difference.

Grinder

[serr]

Originally Posted by nalooti

Hi,

For some reason, my Olympus LS-P2 recorder does'nt allow recording at 24 bit/44.1 Khz. It does allow 96 Khz/24 bit and 48 Khz/24 bit among others.

I've heard that 96 kbps is useless and it's better to use less. Could you please explain why (except huge file sizes) and should I go with 96, 48 (DVD) or 44.1 (CD)kbps frequency (still with 24 bit depth) in order to be able to play it anywhere with good (music) quality?

How can I reduce the sampling frequency from 96 to 44.1 Khz for example using Reaper ?

Many thanks

The telephone game has been played here!

96k sample rate (HD) keeps the 'edges' of the format far away from the audio. It has absolutely nothing to do with capturing frequencies above the range of hearing! (That's an artifact and a moot point. But often used as a strawman argument against using HD.) Think of HD as the last level of perfection for audio with a wide margin.

Consumer level DACs will usually benefit from 96k sample rates and eliminate some distortion of the high frequencies at the top of the audio range that can get altered from filtering out the sampling frequency that is right on the edge of the audio band. (Fun fact, a consumer with such a system can actually benefit from upsampling their SD audio. The end result sounding closer to the original master. Which backs up the theory that HD is only about the converter performance and SD can still contain the full music program.) I hear better performance at HD vs SD on my MOTU converters. I'm very hard pressed to hear any difference between SD and HD with the Apogee units.

You may not be able to tell the difference between SD and HD for every source you hear. Nor should you! However, you will never run into any situation where you get ANY loss in the audio using HD no matter what has happened to that audio before (eg. audio that already has some generation loss that might start to stack up or snowball if you used SD). So you can stop with the shootouts and not feel any need to always re-qualify if 44.1k is still transparent enough for the next job.

If you hear something obvious between SD and HD, look for a different root cause! The difference between reducing 24 bit program to 16 bit is far more apparent for example. And you may know that you need a reference quality system to even hear that.

So none of this should affect a mix. If you recorded your tracks reduced to mp3, now you will start to affect how well your mix might be able to turn out if you were going for full fidelity and ultra exposed mix elements. But a lot of people might still be none the wiser. Record tracks on some cassette deck or recording 8-track deck and NOW you've limited what you'll be able to pull off in the mix!

So anyway, here we are in the 21st century with HD sample rates and the ability to deliver that in lossless FLAC files (or bluray or DVDA disc) directly to the consumer with stereo or full discreet surround sound mixes and everything eliminating all possibility of loss from the "container" format. I say set everything to 96k and forget about it and stop evaluating if 44.1k is "good enough for this recording".

But at the end of the day your mic technique and mixing chops are what matter.
That signature freeze-dried chirpy sound associated with many CD's is the sound of a brickwall limiter crushing the music at the hands of some studio intern BTW. Not an actual limit of the format.

My 2c. Have fun with that or dismiss the whole thing. Up to you!

[Lunar Ladder]

Originally Posted by serr

It has absolutely nothing to do with capturing frequencies above the range of hearing! (That's an artifact and a moot point. But often used as a strawman argument against using HD.)

Against? To me, it's absolutely essential. And using 96k has very much to do with capturing frequencies above the range of hearing. It only depends on what you are doing whether this is relevant or not. (Greetings from someone who does both music and sound design )

[serr]

Originally Posted by Lunar Ladder

Against? To me, it's absolutely essential. And using 96k has very much to do with capturing frequencies above the range of hearing. It only depends on what you are doing whether this is relevant or not. (Greetings from someone who does both music and sound design )

No argument there of course. Sound design is pretty far away and removed from simply capturing and preserving audio which is what I was focused on. Sorry for the omission!

For audio preservation, there's no data above the audio range that is relevant. For sound design, you're taking non-audio data and turning it into new audio. (Yes, based off the harmonic content captured up there!)

The frustrating part to me is the crude slammed and brittle mastering done for most CD releases combined with some of the bs you see in audiophile magazines (and throw in a sour grapes argument from the guy who converted his music collection to mp3 by accident and now claims to "hear no difference" on his ear buds) was the perfect storm of stupid!

[Lunar Ladder]

Originally Posted by serr

For audio preservation, there's no data above the audio range that is relevant. For sound design, you're taking non-audio data and turning it into new audio.

The term you mean is the "audible range", as in the range of normal human hearing, not the "audio range" -- the audio above human hearing is of course still audio, not non-audio. Especially for audio preservation, if there is actual captured audio taking place above the hearing range, in the original recording, it's arguably very relevant to preserve that as well. Again, when preserving audio, you can't be certain who is going to use it years down the line, and what the use is going to be, what there might be in store in the upper range, and so on. Someone might be making something completely different out of it at some point, using also the higher frequencies present in the signal.

I know this is a digression, and I basically agree with what you're saying, but the literal way you are saying it is just something I had to comment on, hah

Sooo yep, when I do sound design using the higher range, I'm definitely taking audio (not non-audio) and turning it into new audio. For example, I record impacts and resonances from architectural structures, and slow that audio down in order to bring the higher elements into the range of human hearing.

[Lunar Ladder]

Before you respond, I have to say: I googled the definition of audio, and it is listed in many reputable places as being by definition AUDIBLE sound. So I'm wrong in this way. When working, files containing ultrasound are interchangeably referred to as "sound" and "audio", and I have come so accustomed to that, it's a difference in terms I clearly needed to bring to my attention. However, the point is, there is actual captured sound up there, and in many cases it's relevant

[serr]

My intention was to avoid the argument for HD that suggests that any content above the audible range adds to the audible fidelity in some way. Talking about this purely as a recording device for audible sound as used in systems where the speakers don't reproduce above audible anyway (or even sometimes the amps).

That you can capture harmonic or whatever content way up there and it's obviously part of or related to the original audible sound (at least if you pitch it down anyway) does beg the question of it's role in the original sound.

Here's what I think happened:
We never captured content above the audible range in the past with analog technology. We simply had very high quality analog devices for the good stuff. The examples in the digital world cited as "missing something" are not missing some mystery inaudible content. They simply suffer from cheapness. Digital gear (and especially the analog/digital converters) ranges from the good stuff to cheapness too. And it has nothing to do with capturing inaudible content or not.

Then in a quest for a analog/digital converter free of having to deal with filtering out the sample rate at the top of the audible range, we got this system that captures data up into bat cave territory. And it turns out you sound design folks can do some pretty cool things with that.

[Cableaddict]

Originally Posted by grinder

Nalooti
Reaper can convert your 48 or 96Khz wave files to mp3 and other bit and kHz values.
Time to read the Reaper manual.
I have video on Utube and audio on Souncloud along with many other people, the only way they got there was mainly Reaper.

Grinder

Indeed. & Just to add:

Reaper's built-in SR converter, if set to highest quality in the project main preferences, is actually VERY good. Good enough that I have no need for third-party or hardware solutions.

(And I'm a very VERY picky guy when it comes to sound, being a life-long pro engineer & heavily versed in techniques like Blumlien, where subtle stuff really matters.)

[drumphil]

Consumer level DACs will usually benefit from 96k sample rates and eliminate some distortion of the high frequencies at the top of the audio range that can get altered from filtering out the sampling frequency that is right on the edge of the audio band. (Fun fact, a consumer with such a system can actually benefit from upsampling their SD audio. The end result sounding closer to the original master. Which backs up the theory that HD is only about the converter performance and SD can still contain the full music program.)

I'm not sure that that is generally true any more. Nearly all modern converters are sigma delta based at very high sample rates with very precise filters. If there was an issue it would show up in the usual audio fidelity measurements.

[emarsk]

I think this is relevant: https://xiph.org/~xiphmont/demo/neil-young.html, and there you'll also find some files you can use to test the intermodulation distortion of you interface.


Also, be sure to watch his videos:

https://xiph.org/video/vid1.shtml
https://xiph.org/video/vid2.shtml

They're excellent and should be interesting to anybody here.

[drumphil]

Yep, it is quite possible to listen to a higher sample rate version of something, hear a difference, and declare yourself able to hear the superior fidelity of high sample rates, when you might just be hearing intermodulation distortion.

[Cableaddict]

Originally Posted by drumphil

Yep, it is quite possible to listen to a higher sample rate version of something, hear a difference, and declare yourself able to hear the superior fidelity of high sample rates, when you might just be hearing intermodulation distortion.

Very interesting. Worth some research for sure.

[drumphil]

Also worth noting is that the specs for the device he has only claim 20Hz to 20kHz frequency response using the built in mics. And that's without specifying whether the measurement is +/- 3db, or whatever.

If you're using an external mic the specs claim 20Hz to 23kHz at 48K. But of course that will depend on the mic you connect.

No figures are stated for dynamic range or noise floor.

There is absolutely no point going above 48K with the internal mic setup, and still no point going above 48K with the external mic input unless you have a very special microphone. (well still no point then unless you want to record stuff you can't hear)

Heck, I'd be surprised if there is any point going past 44.1kHz 16 bit.

All of the other technical arguments are basically irrelevant in this case.