Skip to main content
Topic: Sample formats: fltp (lossy) to s16p or s32p (ALAC) (Read 403 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Hello,

because of compatibility problems I need to convert some DTS, DTS-HD HR, AC-3 and E-AC-3 audio streams to ALAC.
The lossy formats are stored as fltp (depth: 32 bit).

So my question is:

To preserve all information, should I convert them to s16p-ALAC or better s32p-ALAC (although mediainfo sometimes tells me bit depth of lossy stream is 16 bit).

Best regards,
Felix

Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #1
I didn't think ALAC supported s32 samples? Assuming it does, that would be lossless, whereas 16 bit would lose some tiny bit of information.

Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #2
I've never used ALAC so I don't know what the options are...

But, just some reference points -

"CD quality" (16/44.1) is "better than human hearing".

The "pro studio standard" seems to be 24/96 (with the software using floating-point "internally"...  There a couple of good reasons for using floating-point for audio editing/processing/DSP).

I think one company makes a 32-bit DAC chip.  I'm not aware of any floating-point audio DACs.    And, from what I've read most 24-bit DACs (and ADCs) are only accurate to around 20-bits.   So, you can't directly-play a 32-bit or floating-point file.

Quote
although mediainfo sometimes tells me bit depth of lossy stream is 16 bit
Most lossy formats don't store integer samples so there is no bit depth.

Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #3
24 bit depth is more than enough.

Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #4
I think s32p for ALAC is 24 bit.
ffmpeg output:  Stream #0:0: Audio: alac (alac / 0x63616C61), 48000 Hz, 7.1(wide), s32p (24 bit), 128 kb/s

So again for stupid me.
fltp --> s16p: losing tiny bit of information; but I lose information.
fltp --> s32p: no lose of information.

Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #5
I only want to preserve the possibility to get back to the beginnig condition.
And I‘m not sure about the in-depth technics and algorithm the encoders use.

But when ffmpeg shows me a depth of 32 bit for the floating point algorithm and for the s32p algorithm also 32 bit. I see with my limited knowledge to stick with the s32p algorithm should be the best.
Am I right?

Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #6
ALAC should support 24 bit
"I hear it when I see it."


Re: Sample formats: fltp (lossy) to s16p or s32p (ALAC)

Reply #8
After intense googeling I found this: https://community.roonlabs.com/t/acc-mp3-lossy-output-24bit/45690/4

I'll use 24 instead of 16 bit.

The reasoning in that post is poor. The choice of output format is not logically connected to the fact that most decoders use floating point precision, nor is dither relevant.

 
SimplePortal 1.0.0 RC1 © 2008-2018