Bonehead mistake, frustrated beyond belief

[Doublehelix]

We have been working on tracking my son's latest album over the last week or so, and things have been going relatively well. We have drum parts tracked for 9 out of 22 songs along with scratch vocals and guitar.

That is until tonight!

We had just got the 9th song in the bag, all of the drum parts analyzed and all punch-ins completed, and were in the middle of tracking song #10, when I happened to glance down at the computer screen and saw the words that made my skin crawl:

"16-bit, 88.2 KHz"

You could have pushed me over with a feather.

16-bit??? How the...????

This is Cubase 4.13, and for some reason, it defaults to 16-bit, even though all of my old templates are at 32-bit float. I must have created a new template using the default bit rate.

So now, we have 9 1/2 songs in the "16-bit bank" (more like the "16-bit dust bin") and I feel like a fool.

So now, we are going to have to start ALL OVER AGAIN, which is really disappointing. I am trying to look on the bright side, but it is difficult.

It is amazing that I could make it all the way through 9 songs and not even notice the difference in sound quality. Wow! I must be getting old and my ears turning to tin. We actually ran through the 10th song one more time at 32-bit float, and the difference was immediate and stunning.

Right now, I am pretty dejected as everything we have done up until now is all rubbish.

It is time for a good stiff drink... or two...

[wardo]

ah what a pain

have you listened back to the tracks though, some instruments may not sound at all bad at 16 bits, after all it was all that was available a few years ago, and i've done plenty of stuff at 16 bit that still sounds fine to me - no golden ears here mind you.

[Ted Kendall]

Well, excuse me, but it didn't sound too bad before you noticed, did it? And the floating point bit is more to do with processing than the original file. If the performances are good, go with it, and don't fret overmuch on a secondary issue like this. What goes into the mic is N times more important, where N is an indeterminate large number.

[Urthlupe]

Ooo DH...

Last week I was working in Cub4 with a client who remarked on the fact that the pitching on his material sounded off (it was VO). Looked down - 48kHz - where he wanted to be at 44.1. I had stupidly already recognised the slightly odd pitch of the voice, assumed poor transfer back down the line (originals from tape) and pitch shifted the files - stupid berk or what...??? Why had I not checked the clock?? Unlike you I simply lost some time and looked a bit of a nerk, but it really was a stupid baby mistake.

Music in our world is a peculiarly wonderful blend of the artistic and the technical, sometimes maybe those two sides of the old grey matter don't quite blend in the way they should.

Loss of clock settings for me is not a Cub4 issue incidentally - it's to do with the way in which the Apogee Symphony boots on the MacPro - haven't seen any other losses of clock on the Cub systems I regularly come across - poss a Lynx issue ? (assuming you still have that old bit of junk ).

Loopy

[Ian Stewart]

If you didn't notice any difference in sound then just go with 16bits, no one is going to say 'those songs are so cool but 16bit is so totally lame I just can't buy the CD that wasn't produced at 32bit floating point, I mean its so weird.'

[planetnine]

convert them to 24 bit before you do any further processing. there is a good chance that percusive tracks recorded with a reasonable level will not be noticeably different from 24-bit, unless you subject them to rigourous A-B testing in a high-quality environment.

to be honest, I've never really been bothered with 88.2 either providing good preamps and A-D are used, it seems superfluous to me. The issues when recording always boil down to the usual room sound, mic position, performance, tuning, consistency, etc, etc. I've never sat back and thought "I wish I'd recorded that at a higher sample rate" -I think most commercial recording is done at 44.1 24bit.

[wardo]

have you listened back to the tracks though

i mean individual tracks rather than the full mixes btw

[narcoman]

We have been working on tracking my son's latest album over the last week or so, and things have been going relatively well. We have drum parts tracked for 9 out of 22 songs along with scratch vocals and guitar.

That is until tonight!

We had just got the 9th song in the bag, all of the drum parts analyzed and all punch-ins completed, and were in the middle of tracking song #10, when I happened to glance down at the computer screen and saw the words that made my skin crawl:

"16-bit, 88.2 KHz"

You could have pushed me over with a feather.

16-bit??? How the...????

This is Cubase 4.13, and for some reason, it defaults to 16-bit, even though all of my old templates are at 32-bit float. I must have created a new template using the default bit rate.

So now, we have 9 1/2 songs in the "16-bit bank" (more like the "16-bit dust bin") and I feel like a fool.

So now, we are going to have to start ALL OVER AGAIN, which is really disappointing. I am trying to look on the bright side, but it is difficult.

It is amazing that I could make it all the way through 9 songs and not even notice the difference in sound quality. Wow! I must be getting old and my ears turning to tin. We actually ran through the 10th song one more time at 32-bit float, and the difference was immediate and stunning.

Right now, I am pretty dejected as everything we have done up until now is all rubbish.

It is time for a good stiff drink... or two...

I think you must be creatively hearing things!! ....for one thing - even if you use 32bit float as your storage format, without any processing all these files will be is 16 bit fixed point audio with padding for mantissa and exponent. Without processing there is NO difference between a 16bit AD file stored in 16bit and 32bit float. IF you were telling us that you had hoped to be using 24bit converters then you;'d have a point. As it is - with EXACTLY what you have said - there is no difference.

Just convert the 16bit fixed point files into 32bit float for mixing.... then you will be the same as the rest of the work you are doing.

So - clarify - were you hoping to use your converters at 24bit? There are no 32bit floating point converters. The concept is impossible. It would demand scalable hardware working at the sample rate ! ouch! There are only two bit depths that make any difference - 16bit fixed poitn and 24bit fixed point. 32bit floating point is indeed useful once you start processing stuff - but as a raw recording format has NO advantage whatsoever.

If your converters are 16 and you record at 32bit float, it will be no different to recording into 16bit audio files. If you do ANY processing then YES, things change - although I would be extremely doubtful of you hearing it - I think you ought to be mindful of "emperors new clothes" scenario. I use 24bit all the time - it's easier - you can leave BAGS of headroom. But for recording a loud drum kit - there'd be very little difference unless your 16bit peaks were at -15dB(FS) or worse.... and that's the point - there isn't an inherent difference in sound quality - it's usable dynamic range is the main issue.... yeah we can get into quantisation errors etc - but this isn't going to be discernible in a smacked out drum kit....

Convert the files - it'll be fine.

Oh - and just to make you feel better - three years ago I tracked at 16bit for a garage rock record. Got to number 1 in the indie charts - so I wouldn't worry !!

[Michael B]

If it's a good tune it'll survive whatever you throw at it

[thefruitfarmer]

Stay with the 16 bit tracks DH.

Like others have said you can convert it up for processing and the end result won't be that different....

If you do re-track everything you are heading for a can of worms and endless discussions about whether that re-recorded vocal sounds better than the original performance.

[Guest]

I've got a problem where everytime I open a project in Reaper, the MI4 Steinberg audio interface I use is always set to 16bit and I have to change the ASIO settings and then restart Reaper.

I've now got a post-it note on the wall coz I kicked off recording in 16 bit so many times

[SevenIndustries]

I keep forgetting that Logic defaults to 44.1, instead of higher rates - which kind of annoys me.

I've made a template, but in the heat of things I sometimes forget to open it instead of a new one.

[Jack Ruston]

DH it's just not a problem. Being drums, the levels are likely to be very healthy. Above the noise floor anyway. Just convert up to 32 float and crack on.

J

[narcoman]

Stay with the 16 bit tracks DH.

Like others have said you can convert it up for processing and the end result won't be that different....

As I said - if the interface is a 16 bit AD then it won't be ANY different - not one jot! The difference will be if the interface was 24bit and he recorded at 16bit. The 32bit float thing means nothing whilst recording.

[Doublehelix]

Thanks for all the suggestions folks, now that the sun has risen on a new day, I can certainly see things in a better light...

Just to clarify a couple of points that Narcoman brought up (thanks for your input, btw...):

I think you must be creatively hearing things!! ....for one thing - even if you use 32bit float as your storage format, without any processing all these files will be is 16 bit fixed point audio with padding for mantissa and exponent. Without processing there is NO difference between a 16bit AD file stored in 16bit and 32bit float. IF you were telling us that you had hoped to be using 24bit converters then you;'d have a point. As it is - with EXACTLY what you have said - there is no difference.

The recordings were done through Lynx converters, which of course are limited to 24-bit. The 32-bit float format I referred to, is as you mention, only a modified 24-bit format.

I would have certainly preferred recording at 24-bit, and that was my intent all along.

When I mentioned that I heard a difference, I was trying to say that we RE-RECORDED a single take of the new song through the converters at 24-bit (32-bit float in Cubase as the storage medium) and compared it to the 16-bit take. Now we may have only noticed a difference due to the Emperor's clothes phenomenon, but it sure SOUNDED better to our tired, frustrated ears!!!

The only "keeper" tracks here are the drum tracks, and I tent to record with about 10 dB of headroom on the highest peaks. Some of the quieter parts however drop WAY down, and I am a bit worried about the noise floor.

It seems as if the common recommendation is to convert and move on... I hate to compromise just to save time, and I would rather do this NOW then after we have recorded all of the overdubs.

Hmmm.... well you guys have given some things to think about, and we are taking today off, resuming tracking tomorrow night.

We will certainly proceed from here at 24-bit, and maybe leave those already-recorded songs for a decision to be made later.

The plan is to whittle those 22-songs down to 10-12 for the album anyway...

[James Perrett]

While sound quality is important, it should always be secondary to the performance. If the performances are great then there should be absolutely no thought about throwing the tracks away.

Probably the biggest problem is going to be the fact that the recordings are possibly truncated rather than properly dithered so cymbal tails may be a little grungy but I wouldn't throw the recordings away if the performance is good.

Cheers

James.

[narcoman]

Thanks for all the suggestions folks, now that the sun has risen on a new day, I can certainly see things in a better light...

Just to clarify a couple of points that Narcoman brought up (thanks for your input, btw...):

I think you must be creatively hearing things!! ....for one thing - even if you use 32bit float as your storage format, without any processing all these files will be is 16 bit fixed point audio with padding for mantissa and exponent. Without processing there is NO difference between a 16bit AD file stored in 16bit and 32bit float. IF you were telling us that you had hoped to be using 24bit converters then you;'d have a point. As it is - with EXACTLY what you have said - there is no difference.

The recordings were done through Lynx converters, which of course are limited to 24-bit. The 32-bit float format I referred to, is as you mention, only a modified 24-bit format.

I would have certainly preferred recording at 24-bit, and that was my intent all along.

When I mentioned that I heard a difference, I was trying to say that we RE-RECORDED a single take of the new song through the converters at 24-bit (32-bit float in Cubase as the storage medium) and compared it to the 16-bit take. Now we may have only noticed a difference due to the Emperor's clothes phenomenon, but it sure SOUNDED better to our tired, frustrated ears!!!

The only "keeper" tracks here are the drum tracks, and I tent to record with about 10 dB of headroom on the highest peaks. Some of the quieter parts however drop WAY down, and I am a bit worried about the noise floor.

It seems as if the common recommendation is to convert and move on... I hate to compromise just to save time, and I would rather do this NOW then after we have recorded all of the overdubs.

Hmmm.... well you guys have given some things to think about, and we are taking today off, resuming tracking tomorrow night.

We will certainly proceed from here at 24-bit, and maybe leave those already-recorded songs for a decision to be made later.

The plan is to whittle those 22-songs down to 10-12 for the album anyway...

RIGHT! So the issue is actually you were wanting the full 24bit thing you'd done and stored into your 32bit float for later processing BUT you'd started a 16bit session. Yeah - that would be the bummer situation. Indeed - if your recording at the levels I'd expect - ie 0dBVU = -18dBFS then yeah - not gonna sound quite as nice as the 24bit version... so I get your point now it's clarified!!

Well - at the end of the day - the performance is king - it certainly isn't going to be terrible - and there are plenty of big records STILL recorded at 16bit...So - if you have the performance you want - and there is at least some healthy level in the guitars, I'd stick with it. The Strokes STILL record at 16bit!! As does CLA (oh purleease...!!)

[Doublehelix]

As James and Narcoman point out, performance is key. So far all we have in the way of keeper tracks are the drums... Vocals and guitars are scratch, and the bass player is being replaced (although he has agreed to do the album if a suitable replacement is not found in time).

I would have to say that the performance is good, although I am not sure I would say that there is anything there that couldn't be reproduced if we gave it another go.

The one thing we have going for us is that nobody is paying for studio time, since it is my son's band. This is the first full-length project since I have remodeled the studio, so it gives me a chance to troubleshoot the cabling, etc. So in a way, they are my guinea pigs anyway!

My (current) plan is to continue tracking the rest of the songs at 24-bit, and then revisit the original nine that were tracked at 16-bit at a later time and decide if we need to re-track them. (I am guessing that we will.)

Thanks again to everyone for their advice. What a screwed-up situation to find yourself in! And here I am, teaching my son how to engineer, and then I show him how to also screw up!!! Haha! A good lesson to learn early, me thinks!

[Tomás Mulcahy]

While sound quality is important, it should always be secondary to the performance.

Agreed 100%. This should be engraved on the control room wall.

[Richard Graham]

With respect, you must be absolutely mental, considering chucking away good performances just because they are recorded at 'only' 16 bit, especially since you admit you'd only noticed the difference when you looked down at the sample rate display AFTER 9 SONGS!

According to the SOS interview with Trevor Horn, recorded 'Two Tribes' on an early digital recorder, and I daresay with worse converters than the ones in your Lynx. I'll bet it wasn't 24-bit either. Trevor Horn's records sound fantastic. I'll bet plenty of successful 80s records were recorded or mastered at 16-bit.

My jaw is on the floor! Re-record? All rubbish? How can you describe a musician's work as rubbish, just because you are hung up about bit-depth?

If it was 'rubbish', you'd have noticed without looking at the display.

I am gobsmacked!

[snipsnip]

While sound quality is important, it should always be secondary to the performance.

Agreed 100%. This should be engraved on the control room wall.

Agreed, but with a pinch of salt. I encountered someone on a forum who said he had recorded his lead vocal through headphones, and was not willing to redo it incase he lost the vibe.

He said the quality was high but, as you might expect, refused to post a link to the track in question.

If the quality is low enough it will detract from the value of the performance IMO.

[Doublehelix]

My jaw is on the floor! Re-record? All rubbish? How can you describe a musician's work as rubbish, just because you are hung up about bit-depth?

Before you put words in my mouth, I *never* said the musician's work was rubbish. It is the files that are rubbish, and *only* in the sense that I *might* be throwing them away. There has never been any commentary about anybody's work being rubbish other than mine.

The only performances that have been captured are the drummer's, and there is really nothing there that is awe-inspiring, and in fact, every song has many, many punch-ins to fix bad fills, out of time beats, etc.

TBH, it might be a good thing to re-do some of the performances as they are in fact, getting to the point where they are too clinical in my opinion. I actually had this conversation with my son *before* realizing my mistake.

So sure, they are all in time, and all the parts are there, but I would not call them once-in-a-lifetime performances. The drummer is a paid-for-hire musician who is getting a flat rate for the album, and has agreed to re-do his parts.

If it was 'rubbish', you'd have noticed without looking at the display.

I am gobsmacked!

A fair point, especially after 9 songs. As mentioned, the 16-bit files are not bad, but after re-recording the last song at 24-bit, everyone in the room instantly heard the difference in quality between the two. There was no doubt in anybody's mind which was recorded at 16-bit, and which was recorded at 24-bit.

If the goal here is to make the best album possible, you are right in your comments that the performance is the #1 criteria in making the best album possible. I am in complete agreement here. However as mentioned, the drum tracks are not what I would call so inspirational that they are irreplaceable.

Maybe if we had some dynamite vocal performances that were worth keeping, it would be a different story.

This is early enough in the process that I still have the liberty of re-doing the tracks. Any later, and I would have to think much harder before deciding.

So far, we have worked for 4 nights, about 3 hours each night. Nothing so drastic that I would lose any sleep over it.

And as I also mentioned, I have not decided which songs to re-record yet anyway.

One point I will not argue... An inspirational performance is worth fighting for and worth keeping at any cost.

I am sorry you are gobsmacked! I hope you get over it!

[Richard Graham]

Sorry for being a dick, DH. I was out of line. Thanks for being patient with me.

I'm not feeling at all right today.

I personally would never throw away a performance I was happy with purely to increase the signal to noise ratio, unless the noise made the take actually unusable - but your post suggests that you weren't happy with the performances in any case, so fair enough.

In any case, we all care about different things. I'm a drummer, and to me the drums are more important than the vocals (as weird as this may sound to others). The drums determine the energy and the groove of the whole piece.

You are a pro, and I am an amateur, I'm sure this accounts for some of the difference in our points of view.

I record my band with all kinds of spill (no gobos, in a small room, cranked up), in 16-bit, and my criterion for a good recording has very little to do with 'sound quality' in any normal sense.

Life, energy, vibe, of-the-moment-ness, those little bits of magic. They are what make a recording for me. I can put up with all kinds of technical faults if these things are captured and the recording sounds 'alive'. If you don't have them, I think you need to do it again anyway (as your last post suggests).

This isn't to say I think you're wrong to pursue technical excellence too... each to his own. I just don't have the time (or the money) to make sonic fidelity a priority!

Peace.

[James T Bigglesworth]

While sound quality is important, it should always be secondary to the performance.

Agreed 100%. This should be engraved on the control room wall.

Well said.

[Tomás Mulcahy]

TBH, it might be a good thing to re-do some of the performances as they are in fact, getting to the point where they are too clinical in my opinion. I actually had this conversation with my son *before* realizing my mistake.

Ah, so it sounds like a golden opportunity- the 16 bit stuff was a rehearsal, now he can play the parts properly... so get him in quick before he forgets...