Topic: Scaling down 96 kHz/24 bit to 44.1 kHz/16 bit? (Read 18564 times)
0 Members and 1 Guest are viewing this topic.

## Scaling down 96 kHz/24 bit to 44.1 kHz/16 bit?

##### Reply #25 – 2007-02-07 16:11:09
Interesting. So it's a kind of lossy algorithm that meets the average ear? Can it be adjusted a bit lower within the source code so that it would meet those golden ears as well?

You can only change the "shape" by modifying the noise shaping filter (*). The average level (logarithmic) will always be the same unless you change the target bit depth as well.

(* you can also increase the level by using non-minimum-phase filters. But that'd be a stupid idea.)

Cheers!

## Scaling down 96 kHz/24 bit to 44.1 kHz/16 bit?

##### Reply #26 – 2007-02-18 21:09:49

I started to sketch a Mac OS X GUI frontend for the Shibatch sampling rate converter command line tool. One question. What does that p.d.f. in the commandline help text actually stand for?

Code: [Select]
`          --pdf <type> [<amp>]       select p.d.f. of noise                                       0 : rectangular                                       1 : triangular                                       2 : Gaussian`

## Scaling down 96 kHz/24 bit to 44.1 kHz/16 bit?

##### Reply #29 – 2007-02-20 18:07:34
It's not clear what actually happens if you set the slider under "attenuation" to -0.3 dB.
Is the signal attenuated by -0.3 dB (= +0.3 dB boost) or attenuated by 0.3 dB (= -0.3 dB gain change)?

AFAIK SSRC increases the gain when you attenuate "by negative values".
This is probably not what you had in mind.

Sebastian

## Scaling down 96 kHz/24 bit to 44.1 kHz/16 bit?

##### Reply #30 – 2007-02-20 20:54:40
Ok, I changed the label to »Likelihood dither signal» now.
AFAIK that's a very unusual term in the audio world and will likely cause confusion.
Why not simply use "Dither type", as described in wikipedia?

## Scaling down 96 kHz/24 bit to 44.1 kHz/16 bit?

##### Reply #31 – 2007-02-24 07:56:19
AFAIK SSRC increases the gain when you attenuate "by negative values".
This is probably not what you had in mind.

No, I didn't. In fact that behaviour was quite a surprise given that there is the normalize function as well.

AFAIK that's a very unusual term in the audio world and will likely cause confusion.
Why not simply use "Dither type", as described in wikipedia?

Changed that »Dither signal type» now.