Is 16-bit really that bad??!!!??

VU-1 wrote on 6/24/2003, 7:35 PM
OK, guys - do this simple test and see what you come up with:

In VV3, set up a song either for a mix or for mastering - with busses and plug-ins and all; the whole nine yards. BTW - the files I started out with were 44/24.

Now, render it as a 44/16 wav file for CD.

Leave that instance of VV3 open. Open up a new instance of VV3 and put up your newly rendered wav file on the timeline. Make sure there are no plugs on the track, volume set to 0, routed directly to the Master buss, etc, etc, etc. - nothing to change the file in any way.

Now - listen very carefully to the unrendered VV3 project vs. the rendered file on the timeline you just created.

Hear any difference? I did!! I won't tell you how yet so I don't sway your hearing.

Now open the rendered file in Sound Forge, CD Architect, WaveLab and whatever else you have and do the A-B listening test again.

What did you find?

I tried this comparing the same rendered file opened in VV3, SF6, CDA 4, CDA 5 demo, VV4 demo and WaveLab 4 demo against the VV3 project file. They were ALL off!!! The closest were VV4 and SF6. Even worse - the rendered file played in various apps DIDN'T EVEN MATCH EACH OTHER!!

I discovered this while mastering a live CD of 7 different bands - of course, each band had to have its own separate processing and 1 band required 2 different set-ups. I had it all done in separate VV3 projects, rendered and assembled into one Master VV3 project just for creating the CD (spacing & IDs). Nope. Had to go back and re-render all the projects - only this time I kept them at 24-bit. I made the Master VV3 project a 24-bit project too so when I do the DAO burn, it will just truncate the files. I don't know what else to do - the other way sounded horrible.

SO WHAT'S UP WITH THAT ANYWAY??? Is it just that 16-bit really does suck? But what about different apps making the same file sound different...? All through the same output, too.

Somebody do the drill and let me know what you get......

Comments

kilroy wrote on 6/24/2003, 10:30 PM


Try this...instead of doing an internal render, arm a new stereo track and record the file as a real time mix down back into Vegas. Save that as a 44/16 file with some good dithering, POW-er, UV22, maybe try Waves IDR or something if you have it. How's that compare?

Quality bit depth reduction makes a significant difference. As a rule we don't make a habit of using any internal rendering to sum anything we care about. The real time render approach is most likely to best reproduce what you have been hearing all along. If you are clocking from an external source you should definately take this route. If you have a nice pre you might also consider mixing down as above through that and see what you get. This *always* sounds better to our ears. It should be class A topology circuits thoughout.

BTW, you don't really say in what fashion the differences you describe are apparent. What do you mean by "way off" or "not matching", for example? I am not at work so I cannot repo the scenario you describe.

I can tell you this much, we have burned several masters to quality CDRs of the same make and batch run, burned in the same quality drive and played back on the exact same highly specified monitoring system...and they all sound different. So you see, there's plenty of room even at the end game for things to sound slightly "off". The only solution, to one degree or another, is to simply do alot of experimenting and listening.

"Is 16 bit really that bad??!!!??"

Well, look at it this way...Bob Dylan can't really sing...but that never hurt his career any, did it?
tmrpro wrote on 6/25/2003, 12:13 PM
******I made the Master VV3 project a 24-bit project too so when I do the DAO burn, it will just truncate the files. I don't know what else to do - the other way sounded horrible.******

"Give me death or give me truncation ....GIVE ME DEATH!"

Never truncate. Truncation is not ok. Truncating is very bad.

Here's what I do:

Render a file without converting sample rate or bit depth. Also, pass one back through your choice of converters to a new stereo channel. I have both Apogee and Lucid filters and I personally like the way the rendered file sounds better, more often than not.

Open both of these files in your two track editor; SF6 preferrably... & convert your files there using:

Process/Resample... at highest interpolation
Process/Bit-Depth Converter --I've found that using "Gaussian (2 bits RMS to RMS)" without noise shaping seems to reassemble the result with almost no noticable change.

One more thing, you should sonically process your files using a two track editor anyway and you should do this prior to Bit-Depth or Sample-Rate conversion. That way, you are processing your files at a full 24bits.

Listen and choose your favorite result.

Save these new files to a different folder. When you are finished converting your files, you should have 24bit mixes and 16bit masters in your audio file library.

Finally, create a new template at 16bit/44k for your master in Vegas for sequencing/PQ editing and maybe a touch of overall processing.
adowrx wrote on 7/3/2003, 8:04 PM
As tmrpro says, Death is preferable to truncation. Gearboxing always sux. Dither's the key or dare I say bounce your mixes to a nicely setup 1/2" tape machine, 1/4" machine, or any tape machine. Human ears love tape! I understand we're in the 21st century, but there are definitley more positives than negatives using "Real Old Gear".