Any benefits of converting 16bit to 24 bit wav files?

[titser_marco]

Okay, so I have an old Fostex D108 recorder that records at 16bit 44.1khz, but I always make it a point to transfer it to my PC-based station via ADAT. Now, I have several questions:

1. Are there any advantages to converting files originally recorded in 16bit/44.1khz to 24bit/44.1khz? If so, does it outweigh the extra space that the new files would be taking?

2. If my Fostex spits out 16-bit/44.1 khz on its ADAT output, do I need to set my PC's ADAT ins to 16-bit/44.1khz during the transfer [note that this is more like a re-recording, as I'll be playing back the files from the machine and record from the PC as it plays] or can I set the ADAT ins of my PC to 24-bit/44.1khz so I can skip the trouble of batch converting the files once they're in the PC?

Thanks in advance guys!

[skyturn]

if you're converting A/D, playback and all, a good 24/48 or 24/96 gives you a good processing headroom... pero kung D/D, then just stick to what it is nalang.  well that's how i would do it. 

but i think some more experienced guys could give you a better advice... so don't take my word for it.  hehe

[BAMF]

If you already started out recording at 16 bit, there is absolutely no advantage porting it to 24 bits. In fact, you'll even introducing more dithering noise.

If you're recording direct and can still select your bit width, 24 bits all the way would be a good idea. But not f you're conserving disk space, using FX Teleport or if your mix involves loud passages with little or no dynamic variation (as is the case with most loud rock music).

[rakrakan]

if you're converting A/D, playback and all, a good 24/48 or 24/96 gives you a good processing headroom... pero kung D/D, then just stick to what it is nalang.  well that's how i would do it. 

This is a good strategy especially if you're going to add new material - working at 24bit just gives you better S/N ratio, higher headroom. But remember, play the 16/44 files out to analog - then capture the analog to your DAW at 24/48.

[titser_marco]

I see. So what I'm getting here is that 1) I should stick to the original bit depth since I'm doing a digital to digital transfer anyway, and 2) if I really need it to be in 24 bit, I need to play the tracks through the analog outs then re-record it at 24-bit. Is that correct?

On an unrelated note: BAMF - how's FX Teleport? Is it useful during recording [i.e. when I need to use Amplitube when recording guitars, or would need a VST compressor when tracking vocals] or can I just use it during the mixing stage?

[KitC]

Converting from 16 to 24 only adds zeroes to the file; no additional information will be added since it was not there to begin with. This is unlike when a file was originally recorded in 24 bit where there is the possibility of data in the last 8 bits.

No need to convert since most daws process natively in 24-, 32- or even 64-bit, applying their own dither during processing, if you prefer. BAMF, the conversion process usually doesn't not add dithering noise, although some cheap conversion programs do add their own artifacts to the converted file; normally, it should only be 8 zeroes butted on to a 16-bit word.

[BAMF]

Actually what I meant Boskit is the entire process of going up to 24 then back to 16 when you're ready to burn into audio CD. There won't be any bit-padding then bit truncation if the original bit depth was kept (coz AFAIK dithering occurs when transitioning between higher to lower bit depths).

@marco: I was able to pull off using Amplitube via FX Teleport with minimal latency aka the delay is almost unnoticeable it won't affect your playing. But this was at 44.1kHz/16 bits only. Anything higher than that and it becomes quite untenable. Better nga yung "Reamp" solution where you split the clean signal, keep a copy and let the tracking guitarist hear it via his own setup or through a decent fx-amp signal path. I'm having a set of transformers wound so that the line-level from the outputs can drive the amp. Kasalanan ni MarvinQ to e...binigyan ako ng idea with his Radial Reamp box, which costs $200 ouch I'm going DIY on this one.

Converting from 16 to 24 only adds zeroes to the file; no additional information will be added since it was not there to begin with. This is unlike when a file was originally recorded in 24 bit where there is the possibility of data in the last 8 bits.

No need to convert since most daws process natively in 24-, 32- or even 64-bit, applying their own dither during processing, if you prefer. BAMF, the conversion process usually doesn't not add dithering noise, although some cheap conversion programs do add their own artifacts to the converted file; normally, it should only be 8 zeroes butted on to a 16-bit word.

[titser_marco]

Actually what I meant Boskit is the entire process of going up to 24 then back to 16 when you're ready to burn into audio CD. There won't be any bit-padding then bit truncation if the original bit depth was kept (coz AFAIK dithering occurs when transitioning between higher to lower bit depths).

@marco: I was able to pull off using Amplitube via FX Teleport with minimal latency aka the delay is almost unnoticeable it won't affect your playing. But this was at 44.1kHz/16 bits only. Anything higher than that and it becomes quite untenable. Better nga yung "Reamp" solution where you split the clean signal, keep a copy and let the tracking guitarist hear it via his own setup or through a decent fx-amp signal path. I'm having a set of transformers wound so that the line-level from the outputs can drive the amp. Kasalanan ni MarvinQ to e...binigyan ako ng idea with his Radial Reamp box, which costs $200 ouch I'm going DIY on this one.

uy uy uy, reamp box. balitaan mo ako if you're going to sell some units eventually.

[KitC]

Actually what I meant Boskit is the entire process of going up to 24 then back to 16 when you're ready to burn into audio CD. There won't be any bit-padding then bit truncation if the original bit depth was kept (coz AFAIK dithering occurs when transitioning between higher to lower bit depths).

Good one, Jobet. Dither usually adds noise in those last 8 bits, but bad conversion programs only truncate those last 8 when converting from 24 to 16-bit. Not only is this lazy encoding, but it sounds real bad, too. Since most DAWs usually process internally at 32 bits, the dither often occurs there when processing audio. The real test for a good DAW is when they export files to lower bit rates - I've had this happen to me with a couple of my early programs, where exported files sounded different compared to the same song as it was being played back by the DAW. It was one reason why I composed/recorded in one program, then mixed the stems in another DAW, but that was way back in 2004.

[BALDO]

so Boss kitC.. what i have done in the past..recording at 176.4 mixing to stereo at 88.2 then mastering in 44.1 ..was it the right or wrong way of doing things?

[KitC]

so Boss kitC.. what i have done in the past..recording at 176.4 mixing to stereo at 88.2 then mastering in 44.1 ..was it the right or wrong way of doing things?

Depends. I think you should have mixed down to 176.4 instead of downsampling to 88.2 - that way, you avoid downsampling errors during mixdown, then another downsampling error during mastering. One other recourse would have been to use analog summing, but that would mean adding additional gear.

If by mixing down to 88.2, you mean exporting a mix using the DAW's export algorithm, try to compare using a 'null test'. You're using Nuendo, di ba? When you export, you have the option of sending the exported mixdown to the audio pool. Drag that mixdown back into your project, and see if you can invert the waveform using the DAW's audio editor. Play back both the mixdown and recorded tracks making sure nothing is muted. If using stems, make sure they are unmuted along with the inverted mixdown track. If the downsampled mix and playback algorithms are excellent, you should hear nothing because of complete phase cancellation.

[BALDO]

yes kitC what i mean is i mix in 176.4 and export the files to 88.2 as stereo file. So you mean i have to bring back the mixed 88.2 file back to the 176.4? it may not work because the whole project is in 176.4..putting an 88.2 file will sound SLOW and will not sync with the remaining 176.4 files di ba?

[KitC]

When importing a file, a DAW will usually convert it to the project samplerate, regardless of the original file's sample.

What  meant was, export the mix as 176 instead, then do the mastering also at the same rate so that you avoid conversion errors. A good mastering DAW should be able to export the final product as 44.1.

[BALDO]

Boss Kit i haven't heard any audible audio degradation or what ever you may call it - down sampling from 176.4 to 88.2 .( BTW i am using Nuendo 4).does this mean the program is handling it well or the ad/da is doing a great job?  i have been eyeing the summing amp by dangerous audio but i can't justify the cost since i don't hear anything wrong with the downsampling from 176.4 to 88.2 then 44.1... i even monitor the mix on a genelec 1032a and on a  Beyerdynamic DT770.. do you think there is really a difference in doing the mix in 176.4 then 44.1 than 88.4 then 44.1??? i will try to do it differently next time maybe i will hear it hehehe.. ear wax maybe?    those critters - big track counts can eat a lot of bytes i tell you.. one song i did before (33 plus tracks) totalled a whopping 13 gigs 

[KitC]

big track counts can eat a lot of bytes i tell you.. one song i did before (33 plus tracks) totalled a whopping 13 gigs 

All the more reason to track according to the final output, i.e., if the final product is CD, just track at 44.1 24-bit and completely forget about conversion errors. The 24-bit files are more important, IMO since this is where you get your added processing headroom.

Better yet, sum to tape which you then convert to a 44.1 CD master. Analog summing completely takes conversion and downsampling out of the picture.

[titser_marco]

Hahaha another reason to justify getting a two-track tape machine!

[BAMF]

Hahaha another reason to justify getting a two-track tape machine!

But you'll need another mixer (summer). Sabagay, mura lang yun sa Pier, line level mixers, halos nabubulok na nga. I think the people don't know what it's for. I'm thinking of getting one myself, baratin ko siguro ng 2 thousand. I'll use it as a returns mixer for my outboards.

[titser_marco]

Yeah, I'll probably need that; although I think I can use my old Yamaha MT8x multitrack as a summing mixer na lang

[BALDO]

Hahaha another reason to justify getting a two-track tape machine!

i think that is the wrong way of doing things because analog tape can't capture the full dynamic range of digital recordings..

[KitC]

i think that is the wrong way of doing things because analog tape can't capture the full dynamic range of digital recordings..

In theory, analog had no limits when it came to frequency range. I guess a limiting factor was the range of the instruments used to measure frequency sweeps, coupled with the fact that it was decided that human hearing was limited to about 22 kHz so most measurements never exceeded that threshold of measurement.

Digital, however, has a limit, and this is decided by the Nyquist frequency. Anything above Nyquist gets hit by a brickwall filter in order to avoid aliasing. I guess that's why you want to record at 192 kHz most of the time.