Taperssection.com

Gear / Technical Help => Ask The Tapers => Topic started by: caitjim on December 30, 2018, 02:28:03 PM

Title: 24 bit v 16 bit
Post by: caitjim on December 30, 2018, 02:28:03 PM
Apologies if this has been covered elsewhere but a lot of download sites are showing shows recorded in both 16 and 24 bit options. I've even seen some late 1960s audience recordings transferred to 24 bit and the sound is still woeful. You can't put in what isn't there in the first place quality wise. Anyway, back to today. Will I really miss much quality wise if I record in 16 bit as opposed to 24 bit as I'm going to compress it to MP3 anyhow.
Thank for your answers.
Title: Re: 24 bit v 16 bit
Post by: dyneq on December 30, 2018, 02:48:22 PM
The only way to know for sure is for you to do your own ABX test.

I can't tell the difference, and I believe that most people can't either. Here's a good blog post on the subject:

https://people.xiph.org/~xiphmont/demo/neil-young.html
Title: Re: 24 bit v 16 bit
Post by: Ronmac on December 30, 2018, 03:05:52 PM
Higher bit rate does not guarantee better recordings. It does allow for more dynamic range, although that will only be meaningful if the dynamic range is considered during the recording process.

Recording music that has a very limited dynamic range (most club and festival PAs) can not be made better by resampling at a higher bit rate.
Title: Re: 24 bit v 16 bit
Post by: ilduclo on December 30, 2018, 03:13:12 PM
recording wise, you can record at lower volume and avoid any chance of brickwalling when using 24 bit. After recording, the 24 bit can be amplified to the optimal level without adding hiss. My personal method is to record at 44.1x24, saving these to an archival file, then on a copied set of files, do all the adjustments and editing before converting them to 44.1x16. You can do the same from 44.1x24 to mp3 pretty easily. If you want the best sounding mp3's, I would recommend recording at 24 bit and stepping it down to mp3 after editing. As an additional method, flac encoding is a good way to save file size and not lose any sound quality, either at 24 bit or at 16.
Title: Re: 24 bit v 16 bit
Post by: EmRR on December 30, 2018, 03:15:03 PM
24 bit allows more effective changes after the fact: raising levels, equalizing, etc.  A 16 recording with very low level will suffer compared to the same in 24 bit, since the 'floor' is so much closer to the signal.  Most workstations are working at 32bit floating point for level changes, as example, it's 'penalty free' because dynamic range is so high.  16 bit, not so much, you are baking in a smaller range.   

It makes much less difference once you've made changes to a recording so it's ready for presentation, at that point it can make sense to reduce to 16 bit.    If you've got a really detailed listening system you may hear a clear difference between 16 and 24 at that point. 
Title: Re: 24 bit v 16 bit
Post by: daspyknows on December 30, 2018, 05:32:18 PM
Higher bit rate does not guarantee better recordings. It does allow for more dynamic range, although that will only be meaningful if the dynamic range is considered during the recording process.

Recording music that has a very limited dynamic range (most club and festival PAs) can not be made better by resampling at a higher bit rate.

So true, a sh#t recording will always be sh#t recording but a better recording there is more dynamic range to play with and running conservative levels will not impact quality.
Title: Re: 24 bit v 16 bit
Post by: ThePiedPiper on December 30, 2018, 06:39:28 PM
Higher bit rate does not guarantee better recordings. It does allow for more dynamic range, although that will only be meaningful if the dynamic range is considered during the recording process.

Recording music that has a very limited dynamic range (most club and festival PAs) can not be made better by resampling at a higher bit rate.

So true, a sh#t recording will always be sh#t recording but a better recording there is more dynamic range to play with and running conservative levels will not impact quality.

I'm in the same school of thought as Daspy. Personally, I am a "paranoid" taper so I run 24/96 at lower levels to avoid clipping and other related issues. My thought is this: as cheap as memory (SD & MicroSD cards) are these days, it only makes since to get as much info and dynamics as possible. It's only my way of doing things, not saying it's the best way, but I am pleased with the results.
Title: Re: 24 bit v 16 bit
Post by: caitjim on December 31, 2018, 05:56:36 AM
Thanks all. So 24 bit will give me more leeway for processing if I understand it correctly. Once processed I can compress to MP3, Flac etc.
Title: Re: 24 bit v 16 bit
Post by: willndmb on December 31, 2018, 10:44:26 AM
Thanks all. So 24 bit will give me more leeway for processing if I understand it correctly. Once processed I can compress to MP3, Flac etc.
yes
I would say do 24. The only extra step so to speak would be if you wanted to make/share 16
Title: Re: 24 bit v 16 bit
Post by: DSatz on December 31, 2018, 10:54:53 AM
caitjim, yes. But the "conservative" doesn't mean "anything goes" (its opposite!). You still want your maximum peak levels to be as high as possible without quite touching 0 dB.

You can say that a "conservative" level-setting strategy has worked well if your highest peak levels at the live event were -6 dB, maybe even -8 or -10 depending on your particular equipment. But if the peaks are down around (say) -12 or -15, the original recording will almost certainly contain unnecessary extra noise.

That's because each analog stage of your recorder's circuitry has its own noise floor. If your signal is low enough to expose the noise floor of ANY stage in your recording equipment, some unnecessary noise will be added to your recorded signal. Then when you normalize the levels later on, that noise will be amplified along with the desired signal.

--best regards

Title: Re: 24 bit v 16 bit
Post by: caitjim on December 31, 2018, 11:08:59 AM
Thank you all. I now understand and will stick with recording at 24 bit for the added flexibility in the editing stage. Ta.
Title: Re: 24 bit v 16 bit
Post by: Gutbucket on December 31, 2018, 11:41:32 AM
Helps to consider the recording, processing and delivery steps as completely separate aspects with different requirements, each following the other.  In other words, it's not about processing.

Recording- All else equal, if the live signal dynamics fit within 16 bits, the recording will not suffer from being made at 16 bits, nor will it be improved by recording at 24 bits.  However, it is often easier for the recordist to make the recording at 24 bits, since doing so is likely to increase the allowable dynamic range to somewhere between 17 to 20 bits or so (depends on the performance of the specific recording gear used), making the setting of recording level easier with increased safety-margin.  Personal conclusion- makes sense to default to recording 24 bits, retaining decent level setting for the reasons indicated above.

Processing- Best practice to do your processing in an editor which uses a sufficiently larger "calculation space" than the recording itself. 32 bit floating-point mathematics provides this for both 16 bit and 24 bit recordings.  If the recording fits in 16 bits, providing the same data to the editor in a 24 bit file rather than a 16 bit file file won't make it work better or do its calculations more precisely.  However, doing the calculations in a calculation space which provides increased precision over that of the original data can do so. Personal conclusion- makes sense to edit in 32 bit floating point regardless of input and output bit depth.

Delivery- All else equal, if it fits..  I think you get it.   Note that this is what folks other than the recordist actually get their hands on, and as such, justifiably or otherwise, what they are likely to be most opinionated about.
Title: Re: 24 bit v 16 bit
Post by: Gutbucket on December 31, 2018, 11:48:44 AM
That's because each analog stage of your recorder's circuitry has its own noise floor. If your signal is low enough to expose the noise floor of ANY stage in your recording equipment, some unnecessary noise will be added to your recorded signal. Then when you normalize the levels later on, that noise will be amplified along with the desired signal.

This also serves as an argument for providing the majority of the needed gain early in the recording chain, as gain stages later in the chain amplify the noise of those preceding them.
Title: Re: 24 bit v 16 bit
Post by: Gutbucket on December 31, 2018, 12:01:10 PM
^ Best to get levels optimized at the earliest stages of either realm.
Title: Re: 24 bit v 16 bit
Post by: ilduclo on December 31, 2018, 02:01:07 PM
caitjim, yes. But the "conservative" doesn't mean "anything goes" (its opposite!). You still want your maximum peak levels to be as high as possible without quite touching 0 dB.

You can say that a "conservative" level-setting strategy has worked well if your highest peak levels at the live event were -6 dB, maybe even -8 or -10 depending on your particular equipment. But if the peaks are down around (say) -12 or -15, the original recording will almost certainly contain unnecessary extra noise.

That's because each analog stage of your recorder's circuitry has its own noise floor. If your signal is low enough to expose the noise floor of ANY stage in your recording equipment, some unnecessary noise will be added to your recorded signal. Then when you normalize the levels later on, that noise will be amplified along with the desired signal.

--best regards

I tend to disagree. If you're up at the near -6 range, it is much easier to brickwall. One of my tricks to get optimum levels is ask the musicians, in the soundcheck, to play briefly at the loudest they will be playing that set. If you set to below that, you'll be good as gold. If you don't, and it's a band that starts quiet then builds to a loud crescendo, you're going to go over pretty easy. I amp up from -12 all the time, and it's not noticeable to me....(ymmv).
Title: Re: 24 bit v 16 bit
Post by: EmRR on December 31, 2018, 02:21:33 PM
Digital gain is free, sometimes a lower gain setting in a recording path is quieter than a higher one.  I'd be more worried about maxing signal levels if I were transferring out of a recorder in the analog domain. 

24 versus 16 is a real world difference of 25-30dB dynamic range, depending on the converter.  Recording at -20dBFS in 24 is still more dynamic range than 0dBFS in 16.....and ya can't successfully record at 0dBFS, hardly at -6dBFS in unknown situations.   Bring -20 up to -1 with digital gain in post, still more dynamic range than the 16 capture. 

16 versus 24 within a 32float scenario; it's about what the 32float does to the bottom bit of the 16, which is 8 higher than the bottom bit of the 24.  It's where the dither and quantization error lives.  That's further buried in the 24, and therefore is also after processing.  If you get into forensic fixes and noise reduction with something like RX7, there's a lot more in a 24 bit capture for the algorithms to work with.  It keeps the artifacts of the medium much further out of the way, regardless of the dynamic range of the intended capture. 
Title: Re: 24 bit v 16 bit
Post by: DSatz on December 31, 2018, 08:21:40 PM
[message text deleted; I had wrongly contradicted another member of the forum, who then politely pointed out my mistake]
Title: Re: 24 bit v 16 bit
Post by: EmRR on January 01, 2019, 05:04:00 AM
I think you're confusing 24 bit with 20 bit, 4 bits is 24.08dB theoretical. 

16 bit theoretical dynamic range 96.33dB
24 bit theoretical dynamic range 144.49dB
48.16dB theoretical difference

The measured input noise of my MOTU 16A AD gives a worst case result in the audible band of -135dBFS at 20kHz, around -143dBFS at 1kHz.  I've looked at it myself, that's not a number from a manual, that number is worse.   It's about 15dB worse in the previous generation MOTU 2408mkIII. 
Title: Re: 24 bit v 16 bit
Post by: DSatz on January 01, 2019, 06:42:29 AM
EmRR, you're right, I was thinking 4 extra bits rather than 8 when I wrote the above. Sorry! Major mistake on my part.

I'll go back and edit my earlier message now to indicate that it was wrong.
Title: Re: 24 bit v 16 bit
Post by: tim in jersey on January 01, 2019, 10:32:26 PM
caitjim, yes. But the "conservative" doesn't mean "anything goes" (its opposite!). You still want your maximum peak levels to be as high as possible without quite touching 0 dB.

You can say that a "conservative" level-setting strategy has worked well if your highest peak levels at the live event were -6 dB, maybe even -8 or -10 depending on your particular equipment. But if the peaks are down around (say) -12 or -15, the original recording will almost certainly contain unnecessary extra noise.

That's because each analog stage of your recorder's circuitry has its own noise floor. If your signal is low enough to expose the noise floor of ANY stage in your recording equipment, some unnecessary noise will be added to your recorded signal. Then when you normalize the levels later on, that noise will be amplified along with the desired signal.

--best regards

I tend to disagree. If you're up at the near -6 range, it is much easier to brickwall. One of my tricks to get optimum levels is ask the musicians, in the soundcheck, to play briefly at the loudest they will be playing that set. If you set to below that, you'll be good as gold. If you don't, and it's a band that starts quiet then builds to a loud crescendo, you're going to go over pretty easy. I amp up from -12 all the time, and it's not noticeable to me....(ymmv).

Agree.
Title: Re: 24 bit v 16 bit
Post by: EmRR on January 02, 2019, 10:52:17 AM
EmRR, you're right, I was thinking 4 extra bits rather than 8 when I wrote the above. Sorry! Major mistake on my part.

I'll go back and edit my earlier message now to indicate that it was wrong.

No worries, we all do it sometimes!
Title: Re: 24 bit v 16 bit
Post by: tim in jersey on January 02, 2019, 01:41:31 PM
[message text deleted; I had wrongly contradicted another member of the forum, who then politely pointed out my mistake]

Someone admitting a mistake? And then taking the blame for it? Unheard of.

Cheers to you, DSatz.  :cheers:
Title: Re: 24 bit v 16 bit
Post by: John Willett on January 03, 2019, 11:50:14 AM
Will I really miss much quality wise if I record in 16 bit as opposed to 24 bit as I'm going to compress it to MP3 anyhow.

Never record in 16-bit if you can record in 24-bit.

16-bit is fine as a delivery format, but 24-bit is far better for a master recording.

You have a much better headroom with 24-bits - ie: you set 0VU as -18dBFS (or -20dBFS) with 24-bit recording, but only -12dBFS with 16-bit recording. You need a good headroom to avoid any problems with inter-sample peaks that could cause problems, but may nit register on the meters.

You will then have the best available master and the end result, even if MP3 (yuk) will be the final result.
Title: Re: 24 bit v 16 bit
Post by: morst on January 03, 2019, 05:21:07 PM

Never record in 16-bit if you can record in 24-bit.

16-bit is fine as a delivery format, but 24-bit is far better for a master recording.
Words to live by!

I am finally delivering in 16 after years of resisting.
Title: Re: 24 bit v 16 bit
Post by: Chuck on January 29, 2019, 01:24:28 PM
Before I started doing mostly SBD/AUD mixed recordings, I almost always used just one source and I recorded strictly to 16 bits. I watched the levels the whole time, just to keep them as high as possible without hitting 0db. Now that I usually mix two sources together in post production I record in 24 bit trying to peak at -8 to -6 db. That way I can enjoy the show (not looking at the meters the whole time) and still make great easily edited recordings.

That also gives the added benefit of being able to mix the two sources together more easily in post production. In 16 bit I had issues when mixing two sources synced together in post that were peaking at -3db or 0db. With that scenario I'd typically have to lower the gain on one or both sources before mixing, because when added together with EQ etc... the two -3db sources combined would sometimes run over 0db.
Title: Re: 24 bit v 16 bit
Post by: DSatz on January 30, 2019, 05:10:47 PM
John, have you ever seen (or heard of) an actual intersample peak of 3 dB or more? Any studies that I've ever seen of this phenomenon, the worst-case examples were on the order of 2 dB, and typical examples were less. The authors of those studies had to look pretty hard for any examples of it at all.

Plus, any equipment or software that clips with such peaks is inadequate by definition. Some CD players in the 1980s did have such problems, but I've never seen any evidence of it in modern equipment or software (not that it couldn't exist; I'm just saying that I don't know of any reason to believe that it's prevalent).

--best regards
Title: Re: 24 bit v 16 bit
Post by: Gutbucket on January 30, 2019, 05:27:24 PM
Isn't the issue with intersample peaks confined to playback?  Specifically DACs with insufficient output-stage headroom to accommodate an intersample peak if and when it occurs?  If so most all folks will adjust level in post, which is where that would be addressed, either by monitoring with a meter or app which indicates intersample peaks or simply normalizing to a slightly lower maximum level, say -2dBfs?

It's not a problem in the digital representation, as the sample points to either side of the peak remain at or below full-scale by definition.
Title: Re: 24 bit v 16 bit
Post by: John Willett on February 01, 2019, 06:53:39 AM
John, have you ever seen (or heard of) an actual intersample peak of 3 dB or more? Any studies that I've ever seen of this phenomenon, the worst-case examples were on the order of 2 dB, and typical examples were less. The authors of those studies had to look pretty hard for any examples of it at all.

Plus, any equipment or software that clips with such peaks is inadequate by definition. Some CD players in the 1980s did have such problems, but I've never seen any evidence of it in modern equipment or software (not that it couldn't exist; I'm just saying that I don't know of any reason to believe that it's prevalent).

--best regards

No, not come across inter-sample peaks, but then I do not squash up to 0dBFS.

The -18dBFS is the EBU standard and the -20dBFS the SMPTE standard to set for 0VU as I understand it - which is why I quoted those figures.

I leave headroom as you do not know what will actually happen in a performance and performers nearly always play louder in performance than they do in practice.
Title: Re: 24 bit v 16 bit
Post by: Chuck on February 01, 2019, 08:03:49 AM
I leave headroom as you do not know what will actually happen in a performance and performers nearly always play louder in performance than they do in practice.

Agreed.
Title: Re: 24 bit v 16 bit
Post by: anodyne33 on February 01, 2019, 10:11:13 AM
I was always a 16 guy simply on the premise that "we're recording live performances, how much dynamic range do we need?".

I took a side gig working for Pittsburgh Digital (in fact, I think I'll bring that up elsewhere) and Michael and I have had a bunch a chats about it and I'm learning more about audio than I'd ever need to know. I'm now sold on a high bit depth knowing what it buys you when it comes to the dithering stage.
Title: Re: 24 bit v 16 bit
Post by: DSatz on February 01, 2019, 12:38:48 PM
> The -18dBFS is the EBU standard and the -20dBFS the SMPTE standard to set for 0VU as I understand it - which is why I quoted those figures.

John, I know that you know this, but for the benefit of those who don't:

VU meters are a very particular thing. Not every analog meter is a VU meter--not even every analog meter that has "VU" printed on its face. True VU meters are/were highly standardized "volume indicators" (VU = "volume units"). They were carefully designed in the early 1940s by Bell System engineers to give an idea of perceived loudness, mainly of speech transmission for telephone systems and AM broadcasting. The motion of a VU meter's needle has "syllabic response" with an integration time of 300 milliseconds, which is extremely long by today's standards.

True VU meters are dead-on accurate when fed continuous tones. But because of the long integration time, signal peaks occurring in live, uncompressed program material are typically about 8 dB higher than what you would see on the meter. (John, you will probably recall that PPMs were calibrated so that their zero point was effectively equal to +8 VU for this reason.) With some program material a VU meter would "under-read" signal peaks by a significantly greater amount--especially strong signal components that rise and fall quickly, such as percussion picked up at close range with condenser microphones.

Now, the scale on a VU meter doesn't stop at 0 dB; it continues to +3. The zone between 0 VU and +3 is marked in red, and with analog tape there was higher distortion above 0 VU than below it, sometimes audibly so. But even +3 VU wasn't a "brick wall" limit. The most conservative, purist classical approach to recording still involved "going into the red" sometimes--just not hanging out there for any length of time. Rock music, on the other hand, was often recorded with the needle well into the red a lot of the time--intentionally using tape saturation as a kind of compressor. I've seen some engineers "push" tape so hard that the VU meters were continuously "pinned", i.e. off the scale and all the way to the right, and they were proud of it.

In summary: For live recording, if you set -18 dBFS = 0 VU on continuous tone, your typical (often-recurring) peaks on the digital side will tend to be around -10 to -8 dBFS, with occasional peaks going maybe to around -5 or in truly extreme cases, a touch higher. Those would be very good levels for 24-bit recording in my opinion. So the standards you mentioned make excellent sense IF the fundamental difference between VU meters and peak-reading meters is well understood.

Fewer and fewer people nowadays have ever used real VU meters for live recording, though. So some people infer (wrongly) that their peak digital recording levels should be set to occur at -18 or -20 dBFS. That's not at all what those standards mean. Setting your levels that low is just asking for extra noise--if not from the recording channel itself, then from everything that comes before and after it.

--best regards
Title: Re: 24 bit v 16 bit
Post by: u2_fly_2 on March 18, 2019, 12:51:57 PM
If you have the option use 24 Bit.

As all good people has explained/said...more "headroom" to work with in post-prod. and also if you have space for 24-Bit, Why not?

One can always dither down to 44.1-16 Bit later on.

We already have recorders with 192/24-Bit...Do we really need it? Perhaps not, yet use it if there´s a chance/option is my opinion...then ofcourse it´s all about which mic´s and position you use.
Title: Re: 24 bit v 16 bit
Post by: Gutbucket on March 18, 2019, 06:07:18 PM
Why not?

Because some recorders have a high noise-floor and are effectively only recording around 16bits of actual dynamic range even when set to 24bit mode.  In that case you'd just produce a larger sized file with no benefit.  But to know if that's the case or not you'd have to measure, and since the difference in file size going from 16 to 24 bits isn't astronomic, it's a reasonable choice to record in 24 bit mode without worrying too much wasted file space. < Makes sense to worry more about setting levels (go a bit higher).

In contrast, going from 48kHz to 96kHz doubles the file size, and going to 192kHz quadruples it.  That's way more file size bloat than going from 16bits to 24bits and I'd be more hesitant to do so unless I'd convinced myself that the result was worth it.
Title: Re: 24 bit v 16 bit
Post by: MakersMarc on March 18, 2019, 09:52:10 PM
I ran super hot levels when recording to DAT, the only way thing I hate  about 😈 is having to guess at levels and having to add a bunch of volume, yeah I’m doing solid state 24 bit, but I’m losing quite a bit more than I gain by not being able to run hot. Imo.
Title: Re: 24 bit v 16 bit
Post by: MakersMarc on April 07, 2019, 11:57:46 AM
Setting levels correctly "cold" will come with increased opportunity and venue familiarity.

In short, you need to get out more.

<ducks, runs, covers ass>

Have that 620 down now. If I am hitting -12 levels I'm good with it, which generally results if I run 620 levels at -11 or so. So I could probably be fine at like -6 but still don't like to take that chance of clipping. I was worried I clipped Kansas, but I get it home and it's like -12. :facepalm:
Title: Re: 24 bit v 16 bit
Post by: MakersMarc on April 12, 2019, 07:07:44 PM
??? Sounded fine to me. Post work?

Normalize and amplified by 4db. That’s it.