ShortScale.org

Is 24 really always the way to go? Does it actually make a difference?

You shouldn't skip right over 8 bit. 128k speccy and the NES were magical.

If you're recording a multi-track to mix later on, then the deeper the bit depth, the better as you can loose information when you sum them all up via the mixing process. Once it's mixed and mastered, you can then dither down to 16-bit for normal folkes to listen to.

moogmusic wrote:If you're recording a multi-track to mix later on, then the deeper the bit depth, the better as you can loose information when you sum them all up via the mixing process. Once it's mixed and mastered, you can then dither down to 16-bit for normal folkes to listen to.

gusman2x did not understand the question.

happened to me several times that I accidentally converted 16-bit fixed to 32-bit floating, and it sounded like shit every time, although it was not that bad before converting

moogmusic wrote:If you're recording a multi-track to mix later on, then the deeper the bit depth, the better

This is conventional wisdom. In terms of audio fidelity, I agree.

However, there's nothing inherently wrong with recording in 16 bit. I like to record tracks on my iPad using Garageband... it is more inspiring for me because it's so simple. However, iPad GB is 16 bit only. So I work predominately in 16 bit. My mixes may sound less detailed, but I'm not aiming for a slick studio sound. Remember that recording technology is continually evolving, but you can still make a great record with a Tascam 4 track.

I also don't plan to distribute my music in a 24 bit format. So I record 16 bit, distribute 16 bit (or mp3/flac). May not follow conventional wisdom, but it's simple and it works for me. Plus I never have to worry about dithering.

One important note: your recording techinque should vary depending on the bit depth. With 24 bit, you can record at much lower levels and tweak later to get a louder mix. For whatever reason, recording at 16 bit is more like traditional analogue recording. You want to record at louder levels without clipping to get a good mix. This is a major advantage of 24 bit recording: there seems to be a lot more flexibility down the line.

TL;DR: while 24 bit will sound more hi fi, 16 bit is perfectly fine.

Whatever you do, don't record in 24 and convert to 16 without dithering.

i'm new to recording digitally, just checked some tracks, says they're recorded in 32bit float, should i redo in 16? or is it fine? it sounds okay so far.

singlepickup24 wrote:Whatever you do, don't record in 24 and convert to 16 without dithering.

Avoiding the obvious gag, what does this mean? I've recorded in 16 bit up until recently (due to limitations of the hardware I was using) but have just started to use 24 bit and am about to (I hope, if I great a break from the endless decorating) embark on lots of recording. I'll need to convert to 16 bit for CDs (I'm old school) so interested to know what I need to watch out for.

BB, welcome to the world of 24 bit recording! I recorded exclusively in 24 bit for a long time before I became an iPad junkie.

Recording in 24 bit and then converting to 16 bit is very common. However, as one might expect, it is not a lossless conversion. When you convert to 16, lines of code are truncated. This results in some harsh digital distortion that may not be evident to the untrained ear but is certainly audible.

"Dithering" is the solution. Inserting a digital plug-in known as a dither adds a very low level of white noise to your mix. In essence, the white noise masks and balances any digital distortion created by the conversion process. It's sort of a blending technique that makes your mix far more pleasing to the ear. I think of it as similar to sleeping with the constant whir of a fan very quietly in the background to block out street noise.

I usually add the dither plug-in to the master channel of the mix panel before I mix down. I think a lot of people also dither again when they master. Moogmusic seems to mix and master in 24 bit and then dither the mastered track down to 16. There seems to be a few different methods!


Let me know if you have any more questions... I do love talking about recording. I just wish I did more of it!

^so in other words, it's better to start out at 16-bit?

Josh wrote:i'm new to recording digitally, just checked some tracks, says they're recorded in 32bit float, should i redo in 16? or is it fine? it sounds okay so far.

Nope, no need to redo. I'm not super familiar with 32 bit float, but it should give you more detail overall and more headroom.

You'll still be able to convert the files to the appropriate format for CD or mp3 later. Not sure exactly how that works with 32 bit float, but I'm sure google knows.

One note: the bit depth of your audio files is still based on the bit depth they were recorded at. So although you may be mixing in 32 bit float, I'm sure the tracks were recorded at 24 or 16 bit. Something to think about when you're recording in the future, especially because 32 bit float files take up a lot more space. May not be worth it to do all your recording in 32 bit float. Try recording some stuff in 16 or 24 and see if you can hear the difference.

Fakir Mustache wrote:^so in other words, it's better to start out at 16-bit?

Not necessarily.

Recording and mixing in 24 bit will produce richer, more detailed sounds. If high fidelity is your goal, start in 24 bit. You won't lose all of the detail when you convert to 16 bit. A 24 bit track converted to 16 bit will not sound the same as a track recorded and mixed in 16 bit.

Perhaps more importantly, 24 bit is more forgiving during the mixing process. As mentioned before, a deeper bit rate allows for more headroom. You can record in 24 bit at moderate levels (rather than pushing toward max volume) and still get a good sound. In essence, you don't have to worry about clipping when working in 24 bit.

Still, I think working in 16 bit is fine. It's mostly preference IMO.

Fakir Mustache wrote:happened to me several times that I accidentally converted 16-bit fixed to 32-bit floating, and it sounded like shit every time, although it was not that bad before converting

I would only convert down, not up... If you follow my meaning.

24 bit = a ton more headroom, to keep the "noisefloor" down.

a well recorded 16 bit song can sound just as "pro" especially if we are talking about loud rock music.

Dynamic recordings, maybe styles like Jazz, will see more benefit because those whispery quiet parts wont be masked in a noise.

"24-bit digital audio has a theoretical maximum SNR of 144 dB, compared to 96 dB for 16-bit"