upsampling then converting too mp3?

shortedaman

shortedaman

Avid Audio Analyst
OK. so recorded a song at 44.1 than transferred too my PC via usb. opened the track in my daw and up sampled, ( i think there is another term for this but i don't know what it is,) it too 192. then i will convert that file too .mp3, (downsampling too 48.)

is this a bad idea? i mean im just experimenting. please let me know your thoughts.

is this even important?
 
Last edited:
it's not a bad idea . . . but it's just not a very good one.

you won't achieve anything by doing it.

go straight from 44,1 wav to mp3 and you will be just fine
 
thats what i was thinking, then my daw gave me the option. so i said yes. but it doesnt seem to have affected anything. so im guessing it served no purpose, good or bad. ill skip that step next time. thanks man!
 
There are slight advantages of using a 24 bit file over a 16 bit file for converting to mp3, although I don't think there are any advantages to up-sampling from the original file if it's already 44.1.
 
Is this number:

123456

worse or less accurate than this number:

000000000000000000000000000123456

that's all you did when you upsampled, you just added zeros at the front of the number.

Upsampling doesn't make any difference to the quality of your music, it just makes the files larger and more cumbersome for other stuff to work with it.
 
TimOBrien said:
Is this number:

123456

worse or less accurate than this number:

000000000000000000000000000123456

that's all you did when you upsampled, you just added zeros at the front of the number.

Upsampling doesn't make any difference to the quality of your music, it just makes the files larger and more cumbersome for other stuff to work with it.
Click to expand...

It looks as if you are describing a bit rate conversion, not the effects of upsampling.

F.S.
 
You can't make somthing out of nothing. And upsampling attemps to do exactly that. Recording at a higher sample rate in the first place allows D/A conversion to move noise into higher frequencies to be filtered out, but people can't hear much above 16khz anyway so there is quite a bit of room for filtering already at 44.1khz which is giving you around 20khz of range.

I would not bother if I was you.


F.S.
 
TimOBrien said:
It's all numbers to the computer.

That's called 'teaching a concept', sort of like how they put 'not to scale' on a map....

;-))
Click to expand...
Yeah, but your illustration kind of missed the whole ball park.

Bit depth upsampling:
12345 becomes 000000012345

Sample rate umsampling:
12345, 12345 becomes 12345, 12345, 12345, 12345
 
Nah. I'm calling "12345" the single sample. Not sample 1 followed by sample 2 etc.

So in my "demo", I guess we are upsampling a square wave. :D
 
Freudian Slip said:
but people can't hear much above 16khz anyway so there is quite a bit of room for filtering already at 44.1khz which is giving you around 20khz of range.
Click to expand...
No its not, not that much. Sample rate has to be twice the highest frequency as far as I know. But it should be enough anyway.
 
Joza said:
No its not, not that much. Sample rate has to be twice the highest frequency as far as I know. But it should be enough anyway.
Click to expand...


Ya until everyones hearing evolves to be better than it now is and they put much better converters in consumer gear I don't see the point. I am half deaf though;)


F.S.
 
mattr said:
More like 12345 becomes 111222333444555 :p
Click to expand...

Or would it be from 12345 to 1....2....3....4....5

Does upsampling just duplicate the existing samples into the empty space to fill in the blanks us leave them blank and time compensate so that the track doesn't play too fast.
I'm not sure.
 
In a basic sense it will add samples in between the existing ones based on the same kind of interpolation used in a DAC. Its difficult to show with the 12345 example because (without rounding) the added samples would have values that aren't whole numbers... try and think of this when using 24-bit samples. Gets more difficult when you start talking about asynchronous SRC, etc, which I admit I don't understand too well yet :p
 
Joza said:
Sample rate has to be twice the highest frequency as far as I know.
Click to expand...
It has to be more than the highest frequency. Let's say the highest frequency is 20 hz.

If we sampled at 40 hz, any information belonging to a frequency higher than 20 hz could not be sampled accurately. This would cause large audible errors. So before the analog audio hits the digital converter, anything above 20 hz needs to be removed with EQ. The thing is, EQ is not perfect and cannot instantly and perfectly chop off everything after 20 hz. It needs to slope down. So maybe you end up with some stuff still hanging around really quietly at 22 hz. That will still cause encoding errors at 40 hz.

So we encode at 44.1 hz for a max frequency of about 20 hz.
 
You must log in or register to reply here.
Back
Top