Sample Rate, Sample Format, and Storage Size in Recording

[harryk]

I have been doing some little experiments recording from Internet radio shows through Audacity (1.2.6). I have recorded the same 20-second sound (speech) sample using the six possible combinations of Sample Rate (44100 or 22050 or 11025) and Sample Format (16 bit and 24 bit). Of course, I'm doing this to see what settings I need to balance tolerable sound quality and reasonable storage size. The theoretical formula has generally held up. The formula says:
Storage Size (bytes) = (Sample Rate x Sample Format x Recording Length) / 8
Thus, for a 20 second sample at a 22050 Rate and 24-bit, I expect a storage size of
( 24 x 22050 x 20) / 8 …….. which comes to 1,323,000 or about 1.3 Mb.
Formula estimates were very close for the 44100 and 22050 sample rates. However, the actual storage sizes for the 11025 sample rate were greatly larger than what theory tells us is approximately correct. The 11025 – 24bit combination gave an actual storage of 0.744 Mb (theory 0.662 Mb), and the 11025 – 16 bit combination gave an actual storage of 0.916 Mb (theory 0.441 Mb). It doesn't make sense that actual storage for the 16-bit capture was bigger than that for 24-bit capture, and it doesn't make sense that the theory is so far off. I checked the obvious – did the Audacity screen REALLY say what sample rate and sample format I was doing the capture at ? Yes … and I repeated the experiment.

Can any smart "tech" type out there suggest what might be going wrong ??

Thanks !!

[kozikowski]

You're not the first to experience the "hey, wait!" moment. Somebody a while ago discovered that Audacity 24-bit was really 16-bit with extra zeros to make up the difference.

Koz