Hi all,

I synced an AMZN WEB-DL of The Lady Eve (1941) to Criterion's BD.

Here's Blah-ray's coverage: https://blah-ray.blogspot.com/2020/12/la...-1941.html

Looking at the waveforms, I'm pretty convinced that the WEB-DL I found matches the one on Blah-ray:






Notes on the sync:

- The video streams differ in two places:
    1) Opening intros
    2) The WEB-DL has 2 additional frames that's missing from the Criterion BD. This is on a scene cut boundary. (WEB-DL frame #69703, #69704)

- Even after adjusting for both, the two audio tracks still wiggle in and out of sync within 20ms or so throughout the film. So I mostly synced by video, instead of trying to align the audios perfectly.
- I applied an additional delay so that the desync b/w the two audios is mostly within +/- 10ms. Spot-checking a few scenes, I didn't notice any A/V sync issues.
- This is my first amateurish sync involving more than a simple delay. Any feedback and suggestions are most welcome!

Input: E-AC3, 2.0, 224 kbps, 48 kHz, (mediainfo doesn't report a bit depth here)
Output: PCM, 2.0, 2304 kbps, 48 kHz, 24 bits (maybe 16 bits would've been enough?)

PM me for a link.
I also kept the original E-AC3 track, in case anyone is interested in aligning it more accurately to the Criterion track themselves.

Cheers!

(2023-05-16, 02:00 AM)axeyou Wrote: Input: E-AC3, 2.0, 224 kbps, 48 kHz, (mediainfo doesn't report a bit depth here)
Output: PCM, 2.0, 2304 kbps, 48 kHz, 24 bits (maybe 16 bits would've been enough?)

Someone correct me if I'm wrong, but I believe AC3 (and EAC3) are 16-bit.

(2023-05-16, 07:34 PM)stwd4nder2 Wrote:

(2023-05-16, 02:00 AM)axeyou Wrote: Input: E-AC3, 2.0, 224 kbps, 48 kHz, (mediainfo doesn't report a bit depth here)
Output: PCM, 2.0, 2304 kbps, 48 kHz, 24 bits (maybe 16 bits would've been enough?)

Someone correct me if I'm wrong, but I believe AC3 (and EAC3) are 16-bit.

Yes they are, as here you are switching .wav / w64 /FLAC etc. (lossless codecs) you would need to down to 16 bit using SOX or any preferred, probably add dither as I see the wave form having high peaks.

(2023-05-18, 07:39 PM)xwmario Wrote: Yes they are, as here you are switching .wav / w64 /FLAC etc. (lossless codecs) you would need to down to 16 bit using SOX or any preferred, probably add dither as I see the wave form having high peaks.

Thanks, xwmario! I must admit that audio is not my wheelhouse.

Practically in this case, will there be a difference in PCM 24-bit vs PCM 16-bit with dithering, other than the obvious file size difference? And what tool would you use to dither?

(2023-05-18, 08:19 PM)axeyou Wrote:

(2023-05-18, 07:39 PM)xwmario Wrote: Yes they are, as here you are switching .wav / w64 /FLAC etc. (lossless codecs) you would need to down to 16 bit using SOX or any preferred, probably add dither as I see the wave form having high peaks.

Thanks, xwmario! I must admit that audio is not my wheelhouse.

Practically in this case, will there be a difference in PCM 24-bit vs PCM 16-bit with dithering, other than the obvious file size difference? And what tool would you use to dither?

From What I know, WEB audio is usually Digital Dolby / which is a lossy codec, usually a 16bit audio. When you sync using an editor such as audacity and export the audio, you can export the audio to WAV / W64 it hits 24 bits. For which you would need to down it to 16 bit using SOX. If the audio is extremely loud it would be great by adding dither noise, aka noise shaping, and tbh there is "almost" no loss while doing it.

SOX usually gets the job done extremely well and dither.

(2023-05-18, 08:42 PM)xwmario Wrote: From What I know, WEB audio is usually Digital Dolby / which is a lossy codec, usually a 16bit audio. When you sync using an editor such as audacity and export the audio, you can export the audio to WAV / W64 it hits 24 bits. For which you would need to down it to 16 bit using SOX. If the audio is extremely loud it would be great by adding dither noise, aka noise shaping, and tbh there is "almost" no loss while doing it.

SOX usually gets the job done extremely well and dither.

Thank you! Just to clarify, is my understanding correct that editing a 16-bit lossy source and exporting at 24-bit lossless doesn't have any problems other than occupying more space? I assume 24-bit would be enough to properly encode the loud peaks. If so, I'll put this on hold for now and try 16-bit next time.

As you already have the 24bit PCM audio (wav) just resample it to 16bit, and you are good.

AC3/E-AC3/DTS/AAC/MP3 have no bit depth.
This is an error of software like mediainfo which tags a bit depth on lossy tracks.
A petition was made (on videohelp or doom9 I don't remember) for the authors of the software to remove this information, but in vain.
You can take the test:
Encode a 24 bit WAV in AC3, it will still display 16 bit.

DTS however marks the bit depth of the original WAV.
But the decoding must be done in 24 bit all the same.

DTS on Home Video (coherent acoustic) is closer to "classic" frequency compression formats (AC3, AAC, MP3) and has no bit depth, unlike DTS APT-X (Cinema DTS) which has a real bit depth.

If you want to preserve your audio tracks as well as possible, use 24 bit.
16 bit is fine at the mix output: the track will only be dithered once.
But your tracks have already been ditheed / compressed etc...

Comparison of a LD conversion exported in 24 / 20 / 16 bits :
https://slow.pics/c/ONYVFlaI