How does Scope deal with 64 bit?

[Music Manic]

Ok so DAWs are now 64 bit float and Scope is 32bit integer.

My question is, when the ASIO signal from the DAW enters the Scope domain, what happens?

I ask because I have quite o hit vocal coming in which sounds slightly distorted but is not clipping anything. I can only think it's ISP when it's converted.

Could someone explain what's happening to the bit depth conversion?

Thanks

[dante]

A DAW application being 64 bit refers to its execution / executable. That's entirely different to and independant of the bit depth of Audio it's processing.

[Music Manic]

So what happens after the mathematical process in the DAW and the stream to Scope?

Thanks

[garyb]

one is talking about the flow of data in and out of the processor, and the other is about the actual audio data, which are completely independent and different things.

there really isn't any such thing as 64bit audio, although it's possible.

certain processes within Scope may be different bit depths. if i remember correctly, the STM mixers use 48 bit math.

[Music Manic]

certain processes within Scope may be different bit depths. if i remember correctly, the STM mixers use 48 bit math.

Ah yes! Thanks for the reminder. I think I know where my distortion occurs now. There's a sweet spot.

Thanks

[astroman]

call me dumb, but I never could tell a good 16bit recording from a 24bit one ...
so I record 16bit and just chop the sh*t off
Bayreuth would be a different story - but my humble guitar and basses @home...

the float thing in Scope is 32bit integer plus 8 bit exponent
while 32bit float in PC world means 24bit integer plus 8 bit exponent

cheers, Tom

[hubird]

welcome back Tom

[Music Manic]

call me dumb, but I never could tell a good 16bit recording from a 24bit one ...
so I record 16bit and just chop the sh*t off
Bayreuth would be a different story - but my humble guitar and basses @home...

the float thing in Scope is 32bit integer plus 8 bit exponent
while 32bit float in PC world means 24bit integer plus 8 bit exponent

cheers, Tom

Ah yes! I thought that the DAW was the problem. So Scope is an even safer environment to be in?

[astroman]

I'm not the brightest one in math theory, but 1 thing I remember well was the statement:
the numeric range of any floating point system can NEVER continous
regardless of precision there WILL always come a point where a value must switch 'scale' (so to say)
as signal processing IS in particular about continous data, I'd rather stick with integer whenever possible
(and use float stuff for carefully considered 'special' cases)

but that's just about about the most basic use of numbers...
dozens of ways exist to implement a model of a device - the 48bit Gary mentions for the mixers my well apply
Scope's preferred 32bit integer (at least my impression) offers significant headroom while retaining continuity
(or at least makes the latter easier to achieve)

cheers, Tom

[garyb]

here's my confusing, but still oversimplified take-

compared to visual data, bit depth is the number of dots in the space that makes the picture on a printed page.
the sample rate is the number of pictures of the waveform per second. bit depth is the size of the grid. it depends on the point of view(how close to, or far from the picture you are), how small the dots need to be to give a clear picture. with audio, if the dots are too small, the file size gets stupid(and quickly!). if the dots are too large, then high frequencies and quiet passages will be gritty and lack detail. the preferred bit depth is the one that sounds the best, with the most reasonable use of space. there's a point where "better" becomes inaudible, or stops making enjoyment of the sound any deeper.

for pop sounds with limited dynamics, 16 bit is more than plenty. there are cds that sound phenomenal, assuming that the playback system and room can display it. for live sounds, which tend to be very dynamic, i usually use 24bit, since it has enough resolution(dot size) to make a clean and clear recording of a signal half way up the meters, leaving plenty of head room so that i don't need to compress while recording. the lower the signal is up the meters, the lower the bit depth. when the bit depth gets too low, the sound becomes grainy, so that's why you need to keep the signal fairly high with 16bit. when the signal gets below halfway up the meter in 16 bit, you're looking at 6 bits or less. a 24bit scale would up the bit depth 8 bits from 6 bits to 14 bits, a much sharper recording. however, my finished product will end up compressed and limited and will be 16bit.

higher bit depths offer better resolution for fx and combining signals, which leads to clearer highs and less of the distortion that you aren't purposely adding. once those processes have occured, it really doesn't matter that much if the final product is 16bit(coarser resolution, bigger dots) because the lower resolution of 16bit is the playback standard for most consumer listening devices and the limiting of the dynamic range will keep good resolution for the music.

there's no reason that a 16 bit file shouldn't sound AMAZING, assuming it was well prepared and recorded. that's the reason that 16bits was chosen as the consumer standard for digital gear, because file sizes are reasonable and it sounds good. if 16, 24 or 32 bit is needed to make the music incredible, then the music itself may need work.

[Music Manic]

Yes Gary. I agree that more bits are good for processing and that 16 bit can produce great quality. It's also good to know what's going on.

[garyb]

yes, of course. since there seemed to be some confusion, i wrote a book. my apologies if it's not helpful.

[krizrox]

Interesting conversation. For the past many years I had standardized on 24/48 for all my recording projects. I just decided to use the highest rate that made sense for me and my business. It didn't make sense to me to use 16/44. Why would you if you have the option for something better (even if it's just shades of gray)?

Anyway, a few months ago I accidentally recorded a session at 16/48. When I realized what I had done, I spent some time listening and comparing and came to the conclusion that 16 bit wasn't so bad after all. The music sounded great. The client couldn't tell the difference. Since then I've done more 16 bit recordings. They all turned out fine. This isn't Abby Road and I'm not recording Elton John or Led Zeppelin. It's a lot of cover band demos and various other things. The file sizes were smaller which I like. All those backups and archives take up space lol. I've started to pick and choose depending on the nature of the session. Rowdy, thrashy, noisy punk band? 16 bit. Acoustic folk/roots with a lot of air? 24 bit.

[astroman]

here's the stuff I'd track with 24 bits anytime... (no 1176 hardware available)...
https://www.youtube.com/watch?v=wBJP1Ljc_Ng
the dude is cool, takes 2 minutes until he get's down to action - but he explains nicely

regarding 16 bits (and even 14) check out the comments below this video
https://www.youtube.com/watch?v=JS1QK2Wavig

cheers, Tom