This goes along with the topic about the size of files. I get some of it, but what I'm wondering about now, is this:

1. On the Edirol (probably other recorders too) you can go the the 'Recorder Setup' and choose what number bps to use (128, 160, 192, 224 etc.) So if you have it set to say, 192, then when you save it to your computer with an editing program, you are asked again what number to use. Do you think that using different numbers in these two applications screws up the sound quality? I never noticed this before, but I use different numbers.

2. Sample rate - KHz: Not sure what that is, but you get two choices: 44.1 and 48.0 Does it matter?

I can't recall what was in the manual when I had my edirol.

You should do an experiment!

I think the manual is in a junk drawer. But I don't think it will tell me about having a certain number on the recorder, and then using a different number to save the file. I've probably be doing it wrong all this time.

If you always postprocess your recordings on the PC, *and* have a big enough flash card in the Edirol, I recommend setting the Edirol to WAV recording. There is no reason why the Edirol should firat compress to MP3, then your Wavepad should uncompress this, and compress yet again (with possibly different parameters) when saving. If you record to WAV format at least you do not have any losses on that end. The files are bigger but actually load faster in your audio editor as they don't need to be decoded.

My 2GB flash card allows for 2 hours recording in WAV format. That is a bit of a pain but most people will have bigger cards by now.

If you still decide to record in Mp3 format, set it to 44.1 Khz sample rate (higher is overkill) and 128 to 192Kbps compression, so at least you already have the required quality. Then when saving the mp3, make sure you do not use higher compression (a lower Kbps number) so as not to lose quality.

I always record in WAV on my Korg MS-1000 DSD in 24-bit, 128kHz. Engineering studies have been done to see if the human ear (a pathetic sensory organ really as compared to other mammals) can differentiate higher levels of fidelity from 16-bit, to 24, and upward. The answer is no, the ear cannot distinguish improvements beyond 16-bit recordings. Yet 24-bit is fairly standard on today's moderately priced equipment, but anything above that, while increasing the profit margin for the manufacturer and vendor, is actually lost on the ear.

Similarly, when it comes to sampling, the human ear stops hearing enhancements beyond 20kHz. My Korg lets me set sampling all the way up to 192kHz, which, of course, is way beyond the ear's feeble capabilities. I once encountered a "test" on the Internet where a recording was made at 44, 88, 176kHz, etc. One had to match each recording with the sampling rate and submit it to the website for scoring. The "test correcter" said that the results were always comical, as nobody could guess the right answers. The ear simply cannot differentiate those increased levels of fidelity.

I mentioned that my Korg does WAV, but also does WSD, DSF, and DFF formats. Most importantly it does the DFF, which is Direct Stream Digital (DSD) at 1 bit 5.6mHz, which is cutting edge technology. I still use the WAV option though, because the conversion programs can do WAV to MP3, but have never even heard of DSD, never mind have the ability to convert it. So when I do the conversions to MP3, I always select the fairly standard 128kHz that I also used for the original recording.

Someone once suggested to me that if you record and convert at, say, 196, or 224kHz, that it will greatly enrich the recording and counteract the terrible loss of fidelity resulting from MP3 compression. Not so! Again, the human ear can't tell the difference. I actually did an experiment with a recording and subsequent conversion using 224. To me I could hear no enhanced fidelity, but the sound overall seemed more nebulous and weird somehow. So I've stuck with the 128k bits/s setting for both the original and the conversion settings.

This is strange because 128 kHz is not a multiple of 44 kHz or 48 kHz like the standard sampling frequencies. For MP3 conversion, I guess that your setting is 128 kbit/s rather than 128 kHz ?

Well, I still don't know what that 44 khz and 48 khz thing means. But I have figured out how to change some of my recordings that were 48 into 44. Not that that really matters for this site, but it does elsewhere.

44 kHz and 48 kHz are sampling frequencies for digital audio recording. In a recording at 44 kHz, the sound signal on each channel is represented by 44 000 numbers, the samples, per second, each number being coded usually on 16 bits. So the bit rate of an uncompressed stereo recording is
2 channel x 44 000 samples per channel and per second x 16 bit = 1.4 million of bits per second, that is 1.4 Mb/s.

The size of a recording file is the product of the bit rate and the duration. For instance, the size of a 5 minutes uncompressed stereo recording as above is 1,400,000 bits per second x 300 seconds = 420,000,000 bits = 52,500,000 bytes = 52.5 MB because 1 byte = 8 bits.
Hence after MP3 compression at a bit rate of 128 000 bits/s, that is 128 kb/s, this recording will be 1,400,000 / 128,000 times, that is about 11 times, smaller. Indeed, its size will be 128,000 bits per second x 300 seconds = 38,400,000 bits = 4,800,000 bytes = 4.8 MB.

Didier - can you say that in English, please?

Seriously, thank you so much for the explanation. For some reason, this is still confusing to me. But I'm going to save your words in one of my files so that I can read it a couple dozen more times. Maybe one day it'll sink in.

Monica, just keep in mind that the sampling frequency is the number of samples per second, for instance 44 kHz means 44,000 samples per second, and the bit rate is the number of bits needed for the binary coding of these samples within 1 second, for instance 128 kb/s means 128,000 bits per second.
The relation between both these quantities is

bit rate = number of channels x sampling frequency x number of bits per sample / compression factor.

Most often the number of channels equals 2 (stereo recording) and the number of bits per sample equals 16. The compression factor, which is the ratio of the recording file sizes before and after compression, equals 1 for a .wav file, which means that there is no compression, and about 11 for a 44 kHz and 128 kbit/s .mp3 file.

Well, I am not sure that I could do my explanation clearer.

Thanks, Didier. You're a sweetheart!

Hi,

Yes, in my post above, last sentence, I meant bits not kHz. I just went back to edit that. Sorry for the error.

David