20-bit audio question

[echohead]

ive got a 24-bit audio source -- a .dtshd file from a bluray. the file is actually 20 bits of audio per sample with the remaining 4 bits filled with zeros, since next to nothing can actually play 20-bit audio.

anyway, the original file needed some editing, so after doing dtshd --> flac --> individual wav files and editing each one in audition, i attempted to save the file as a 20/24 .flac file. problem is, audition as well as every other program ive found wont save 20-in-24bit files. youre forced to either save them as "true" 24-bit which bloats file size (3.68gb result for me as opposed to 2.52 for the original .dtshd encoded to flac), or as a 20-bit file which almost nothing can play back let alone encode from.

so my goal is this -- 20-bit 5.1ch audio in a 24-bit file header (?), formatted as flac.

is there any way to do this, or am i screwed?

[ramicio]

I don't know the intricate workings in FLAC, but it should throw away those 4 empty bits because they are redundant.

[echohead]

flac wont read from 20-bit files; it only accepts bit depths in multiples of 8. and theres nothing in its documentation that mentions a switch to zero out the last 4 bits.

using eac3to, the bit depth IS handled correctly when converting to flac. it flagged the source as "20/24 bit". unfortunately it is only able to read FROM 20/24 bit audio. nothing ive come across can actually create files with that bit depth unless the source is already 20/24.

[ramicio]

I don't see the problem. Yes a 20 bit audio is going to have to go into a 24 bit file but a lossless compressor should see these redundant 4 bits.

[echohead]

original 20-in-24bit flac: 2.52gb
new 24-bit flac: 3.68gb

as i understand it, 20-bit and 24-bit audio are read in the same way. so writing a 20-bit file with a 24-bit file header is all thats needed here.

theres even a program called Twen that's supposed to do exactly what im attempting to do here. when it "finishes" re-encoding the audio, the result is only 90 minutes long (the movie im doing this for is about 2 1/2 hours) and the file can't be read by anything.

[ramicio]

I don't get it. What is the difference between those 2 files you just posted the size of?

[echohead]

2.52gb: my bluray disc --> eac3to --> converted from .dtsma to .flac


3.68gb: my bluray disc --> eac3to --> .dtsma --> .flac --> 6 mono .wav files --> audition --> tempo sped up by 0.1 percent (23.976 --> 24 fps conversion) --> trimmed a few seconds of silence from either end --> saved as 24-bit 5.1ch .wav --> .flac via eac3to

[ramicio]

When you process those files to speed up or slow down the file you are changing the audio data. Changing volume or anything about the actual audio data will use calculations. It is going to truly 24 bit now because of the new numbers that were calculated and rounded. Same thing goes for producing music, they edit things in 24-bit or higher and then the final production is 16-bit to not lose quality because of rounding errors. I hope it makes sense. If you want to keep the smaller file then the video needs to stay 23.976 FPS.

[echohead]

that'd be a great explanation if audition didnt have the option to save files as 20-bit.

maybe i should clarify a bit more. i have the 6 individual mono .wav files. i have two copies of each - one as 20-bit and one as 24-bit i then have two copies of the merged 5.1ch wav, one in 24-bit and the other in 20-bit. nothing will read from the 20-bit files except audition.

[ramicio]

That's because you are trying WAV, and that's not supported because any other software goes by the WAV standard and reads in 8-bit multipled word lengths. Is there an option for raw PCM? You're really best off just not going to 24 FPS exactly. A DTS-HD with 20 bits is going to decode as a 24 bits with 4 bits of empty zeros, and software like eac3to will show that the bit depth is only 20 bits.

[echohead]

i am working with two blurays of the same movie. one has 24 fps video and 16-bit audio. the other has 23.976 video and 20-bit (shown as 24-bit) audio. even if the video was the same framerate, there are slightly different lengths.

i am attempting to take the 20-bit audio from the second bluray and sync it up with the video from the first. i dont want to use the 23.976 video because it is lesser quality and there are issues with the on-screen translation. i am also aware that the 24fps video could easily be slowed to 23.976. however this is one of my favorite movies and i would strongly prefer to have everything play back as 24 fps.

i found a single hydrogenaudio thread (http://www.hydrogenaudio.org/forums/index.php?showtopic=36883) (last post) that explains what needs to be done in a situation like this. however they make no mention of what (or any) program that can be used to do this.

A DTS-HD with 20 bits is going to decode as a 24 bits with 4 bits of empty zerosim aware. this is my goal, only in .flac

software like eac3to will show that the bit depth is only 20 bits.eac3to shows the source .flac bit depth to be "20/24-bit". not 24-bit, and not 20-bit. it recognizes that theres 20 bits of information in a 24-bit container.


im not concerned with intermediate steps, only that the end result is a 20-bit 5.1ch flac.

[leeperry]

2.52gb: my bluray disc --> eac3to --> converted from .dtsma to .flac

3.68gb: my bluray disc --> eac3to --> .dtsma --> .flac --> 6 mono .wav files --> audition --> tempo sped up by 0.1 percent (23.976 --> 24 fps conversion) --> trimmed a few seconds of silence from either end --> saved as 24-bit 5.1ch .wav --> .flac via eac3to
as soon as you do some DSP, it's 24bit all the way again.

if you want to do 20bit and butcher the *useful* 4 LSB, you can do it in eac3to with "-down20"

If I were you, I would keep the untouched audio at 23.976, slow down the video to 23.976(with mp4box) and use Reclock in 24fps mode. This will have the nice added value that your video will playback butter smooth as well, and you haven't tempered with either A/V file.

[Ghitulescu]

I would keep it as 24b, with no processing other than fps change.
AFAIK, apart from some studio gear, no consumer one is able to cope with 20b. There's only a slight advantage of having 24b over 16b (I mean, of course there's one, but very few people could tell the difference using normal priced gear, actually 24b was created for studio editing purposes, for audiophile purposes was the other trend, eg 88.2, 96,192kHz), and reprocessing it again to add the 4b would waive all this advantage. So why? For gaining a few MB? Or, in other words, some 0.001€?

Should it be my favourite movie, I won't do this to it.

[pandy]

audition by default add some dither to file during processing - try export file with limited resolution to 20 bits ie convert file with explicit resolution as 20 bits inserted in 24 bit words (but remember that anyway if You do some processing then You change real resolution of samples).