On Thu, 11 Aug 2005, Robin Millette wrote:
but as i've said before, my ears are only 16bit anyway...so it's not big deal for me.
You should care about that extra byte if you're mixing a lot of tracks and need more headroom.
In PureData, all mixing is done using 32-bit floats, which are like 24-bit integers, plus adaptive resolution, so that the effective number of bits stays 24 no matter what's the dB. So that's a lot more headroom than anything you can get with 24-bit integers. That kind of resolution is mostly only useful during processing.
The 24-bit integers would only be for recording to .wav.
Honestly I don't really know what's the fuss about 24-bit recordings.
Maybe because some company got pushed into higher-end equipment than what
was ever desired, so they created the desire and thus the market. Each bit
of resolution requires 6 more dB of S/N ratio in order to not get masked
by the hiss. Multiply that by just 16 and try to find me a soundcard with
a S/N ratio of 2^16 = 65536 = 96.3 dB (!!!).
And then fetch me an ear with 96.3 dB of S/N.
BTW, one can cheat by recording at a very, very quiet level, and then playing that real loud. This allows your salesman and/or brother-in-law to effectively disable the top bits of your 16-bit range, making the 24-bit recording of the same thing sound so much better.
I must add that if you mix a lot of tracks together then the result still has to go through the same file, soundcard, loudspeakers, room and ears than before, and if there are 8 simultaneous sounds playing in the same frequency band with equal loudness then they compete for the available amplitude range in a way that each of them needs 3 _less_ bits in their pre-mixing recordings.
The float stuff of Pd is more useful than that because Pd's DSP is used for more than just sound: e.g. it is also used for indexing into recorded sounds, and that needs a lot more resolution (and, in fact, many would say float32 is not enough for that purpose).
(did i write anything inaccurate here?)
Mathieu Bouchard - tél:+1.514.383.3801 - http://artengine.ca/matju Freelance Digital Arts Engineer, Montréal QC Canada
On Fri, 12 Aug 2005, hard off wrote:
did i write anything inaccurate here?
i don't know...but is it still ok for me to have 16 bit ears?
Sure... I made all of my music compositions using 8-bit samples, and I still think they sound nice :-)
And then I remember spending a lot of time composing on a 1-bit (that's 1 as in _one_) sound output. The software I used was actually *dithering* the sound in pretty much the same way it's done for pictures to simulate the extra bits of colour. (!!!) It rendered in ultrasounds and then relied on the speaker's natural [lop~] to make it sound like smooth audible sounds.
But the thing I missed the most while composing using a PC Speaker, was that the damn thing didn't have a subwoofer... :-( I think I must then have switched to using my parents' computer because _they_ had a 8-bit soundcard that I could plug into normal speakers.
Mathieu Bouchard - tél:+1.514.383.3801 - http://artengine.ca/matju Freelance Digital Arts Engineer, Montréal QC Canada