Alright. Will do that.

Look, in the preservation world, audio for most analog audio formats is 24-bit. If you seek preservation bodies' requirements for audio format digitalization, it's 24-bit. I don't decide that. Generally, for analog video, it's opposite. Audio is supposed to be 16-bit but they want specifically 24-bit...

I don't know if you are joking or not but... we're not building something specifically for that. There must be something that can be done with what we have.

We DO have some AIW systems on XP. I can and use those but they are super janky volume wise.

Alright... deactivated the "Always suggest RGB format for output" in the Huffyuv config in compression.

I was using "H264" as some kind of catch-all. I know x264 is a thing.

Yes, that's what I'm doing. But I also do raise the volume as it tends to be quite quiet most of the time and I love seeing the waveforms to do that kind of work (and also "zero" what's supposed to be silent, etc.). In other words, some kind of "mastering" process.

Not months... We *just* got the tapes to digitize them. It took quite a lot of time to get here though. My previous thread was to see my options when it came to that issue.

Before posting this, I've messaged my boss and asked to come around when he has the time.

The org had a second option in their docs: ffv1. It's noted to be "Acceptable" and "v210" to be preferred. They explained as such because "we'd need to transcode it to make it work with our systems." (paraphrased)

At least ffv1 is an export option in vdub2. But I'm not banking on it, tbh. It will depend on what my boss will say about the subject when I'll be able to talk to him.

I'm open to your longer way to do things - especially if it can help me make better quality outputs - even in lossy H264.

FFV1 will make this process much easier for you.

Before I proceed with any workflow suggestions, let's see where the FFV1 option lands. If it's a go, we can make this far more painless for you, possibly even skipping Premiere entirely.

BTW, this is normally far more detail than I'll do in a thread, but I'll do it for you (and your org). :)

What I had in mind, was to open the audio recorded in the AVI file in Audition alongside the one recorded in 24-bit in parralel. Make sure they have roughly the same amount of silence at the beginning (based on the AVI file ofc). The end does not really matter because it's just noise that's meant to be silent. Then, in VDub, I saw the option "Audio from other file" in the "Audio" menu on top. Use that to load the edited 24-bit PCM file. Crop the beginning and the end. Save as FFV1 or huffyuv.

Done for their "Preservation" file. They can deal with the v210 transcode and see if I care.

After, use VDub to add a Deinterlaced filter. Once done, export as x264 16mbitCBR 320kbps AAC MP4 file. (also, for some reason, x264 in vdub2 is broken so that will need to be fixed).

My boss will come by tomorrow afternoon. I will confirm with him what to do and/or if it's OK to fo the ffv1 route.

Let's swap to Hybrid for QTGMC deinterlace + x264 encoding. Far easier, and much better quality. So win-win.

What is the version of your vdub2?
Also, why Deinterlaced filter? Haven't your client asked for interlaced?

Soooooo... about the parralel audio recording... the previous user that said that it wouldn't work was right. The 24-bit audio file has a slight .25sec delay on the audio versus the audio in the video file. I would imagine that this kinda varies throughout.

But yeah. That's a robot effect when overlayed. That'a a no-go.

Will check that out!

v1.9.11 for SD capture
v2 build 44331 for HD cpature and other things.

They asked for two versions. One for "preservation" that preserves interlacing and one for "access" that is a deinterlaced MP4.

I see.
Still, what version of vdub2 do you use?

I edited my post above but in case you didn't see it...

v1.9.11 for SD recording
v2 build 44331 for HD recordings and other AVI conversions. (Had issues with recording where the preview would get slower and slower until it only displays less than 1 frame a second. Resulting file is correct though).

Current work PC specs, if you missed them earlier:
Ryzen 9 5900X
32GB DDR4 RAM 3600mhz CL18
Dell Quadro RTX A4000 8GB
1TB OS SSD, 2TB other SSD + 18TB HDD

Your vdub2 is outdated. That's why you have issue with x264.
You need v2.3.0.845 from 2025.
You'll get there x264 for 8 and 10 bits, lossless, SAR, interlaced video explicit flag and compression optimization.

Thanks. Updated.

Will still need some instructions with QTGMC that LS mentionned. What version of Avi/Vapoursynth should I use (if possible, I'd like a portable install)? What command-line is the best for some catch-all scenario for tapes? That kinda thing. When I said I'm allergic to CLI, it's not because I don't know how to pass a command, it's just that I just cannot figure out on my own what would the command to use and, mixed with bad memories, it makes me dread anything without a GUI.

As for the 24-bit audio, I think I have an idea that I'll need to talk with my boss this afternoon. I don't know how well it's going to work out but... yeah.

VirtualDub2's VFW implementation of x264 is outdated, no matter how "updated" the software is. It's like trying to cram an EV engine into a 1995 Mazda (or whatever). Yes, people have done these silly conversions, but it's not at all the same as a ground-up software for modern encoding.

VirtualDub(2) has it's place. This isn't it.

Hybrid has built-in Avisynth (and Vapoursynth), and learning raw Avisynth is not needed at this time.

That 24-bit audio mess will probably never work. Software to ingest SD video was max'd out at 16-bit for a reason. HD allowed more (or even expected more?), but then the SD was given the step-child treatment, not built for that SD task. I would just "upscale" it from 16 to 24, and go on with my day.

Odds of anybody being smart enough to spot it are zero, given how they're not smart enough to give a legitimate ingest request. You're dealing with low-knowledge people, who "think" they know more than they do (because they read some BS from some unknown somewhere). Zero knowledge unto themselves, zero experience at any of this. If you or your boss tries to explain this to them, it will fall of deaf ears. Honesty will not be rewarded.

100% agreed.

My idea was to add transparent (VERY quiet noise) 24-bit white noise that they won't ever know it's there. So that if they ever decide to check for legitimacy, they'll see a real 24-bit audio but that's just because some very low sound white noise I added. I tried it out and it works wonders.

How?
- Extracted audio with VDub
- Open in 16-bit WAV Adobe Audition
- Create a first new empty 32-bit float file
- Create white noise of needed length that 32-bit float file
- Create another new empty 32-bit float
- Copy & pasted entirety of audio from AVI (which upscales)
- Finally, copy & mix-paste very quiet noise within the original audio
- Save file as 24-bit PCM

You can check by opening any DAW that supports VSTs. Install "Bitter" VST by Schwa. By experience, 99.9999999999999999% will never know how to check for that and, if in the small chance they actually do, it'll be a disguised 16-bit file from the white noise. haha Then just use that file in VDub and make final ffv1 encode. That's the plan I'm going to present to my boss, whenever he happens to be able to pass by.

I'm going to give Hybrid another chance but I'm not having high hopes as, last time, it wouldn't even open.

What you are talking about? What 24bit noise? Are you kidding?
Simplest way is to use Adobe Premiere together with Adobe Audition. Premiere has option "edit in Adobe audition" so you can edit Audio file in Audition, convert sample type, normalize etc, save, close Audition and go back to Premiere. Then export.

Read that Studer article I attached above or at least Wikipedia:
https://en.wikipedia.org/wiki/Audio_bit_depth

That's a brilliant idea. Do that. :congrats:

Those "extra steps" realistically should not take much extra time. Just some clicks.

Make sure whatever extra noise is added is below the VHS audio noise floor, which I'm sure it is. But it'll be 24-bit range for saving. (Realistically, that's all 24-bit would have been here anyway: richer noise.)

I'll walk you through Hybrid, if needed. Just let me know where you run into a roadblock. You want to return QTGMC deinterlaced to the client, not lousy ancient VirtualDub deinterlacers. Deinterlacing is one are where I'm pretty unforgiving, because non-QTGMC has such a steep quality drop-off in almost all cases.

No.
That workflow would be a nightmare here, and would not accomplish what's needed anyway.

However there is simpler solution. At export Adobe Premiere offer to set any sample frequency and bitrate. So if original audio track is 16bit, set to 24 and export, keeping all other (video) setting the same.

Anyway there is nothing related to noise. No one will be able to tell particular audio track was 16bit before :)

There is simple test to be sure - after export, in Premiere open "get properties from file" and look what you can see.

P.S. And there is possible to set even video to 10bit at export with appropriate codec. How that stupid mess with format affect video quality I have no idea (with audio all will be OK) but it is simplest way to get required format :)

If 16-bit is simply converted to 24-bit, it won't pass "is this real 24-bit audio" tests. But adding near-silent noise that is beyond 16-bit will fool the tests.

This isn't about quality anymore, it's entirely about stupid tests. The max possible quality was extract from the VHS tape, which is arguably not even 16-bit @ 48kHz (which is what the ingest/capture is). He already has all the quality.

The entire situation is because he's dealing with low-knowledge goobers. They read random specs online, and want those used for no qualified or quantified reason. Those are not specs for consumer analog SD formats (VHS, etc).

This is his out, and I think it's quite clever. Assuming it works as planned, and I think it will. :)

He does the best job that is possible, and they get their dumb files. Win-win.

In my opinion, in this case he do not want to worry about it.
What the heck is "is this real 24-bit audio" test? :)
There is one trick with sample frequency - if you record at 44.1hz and then upsample to 48khz original noise cuts at about 20khz, not about 24khz so you can tell something is wrong - probably something is filter between or it is upsampled. But with 24 bit - it is not related to noise so you can not tell it is fake. :)

So background noise difference is true with sample frequency not bit depth.

-- merged --

P.S. However, if he bother someone, by using special test programs, will recognize those added 8 bits... Then he will recognize them even if there will be mix with true 24bit :)

Cheating should be simple. Complex cheating almost always fails, as Remarque said :)

I do not know how Premiere converts 16bit to 24bit, but Audition not simply add 8 zeros, it dithers file. So it would be nearly impossible to tell it was 16bit before. But there is only one way - to try and check result.

First, check the VST I'm talking about. "Bitter" by Schwa: https://www.kvraudio.com/product/bitter_by_schwa

Yes, Audition simply adds 8 zeros. Dithering only happens when you are changing the samplerate (aka the Hz). I'm only asked to dither (or not), in multitracks, when sample rates mismatch. When they do, software asks me dithering questions before upsampling up or down (and makes a copy).

That being said, doing what I describe will show 24 bars pop out and all near all equal in length if you are using the VST I linked above. If you pop in a fake 24-bit file with 8 padded bits, the VST will show 16 bars even if the file is 24-bit. You can try it with any DAW that supports VST.

It's all to fool those that knows how to check that.

Going to be honnest, it felt good to see my idea approved by you! My boss also agreed with the strategy. So, we all good on that.

Going over Hybrid would be great. I noticed that there is a "Defaults" option and I suppose that I could technically use that to plug in your settings as default as I only intend to de-interlace using Hybrid when the client is extremely picky. Which only happened once and it's this very client.

I imagine that you are starting from a huffyuv AVI file. I'll have that ready tomorrow.
I'd also know if there's a good way to force an "interlaced" flag on that ffv1 AVI file just for, you know, pass "the test".

Thanks a lot in advance.

24 bit denotes the size of the container. But the container is not the contents. Putting contents the size of a matchbox into a shoebox won't make the contents worse but won't make it any better either. A little knowledge can be dangerous. We can add very quiet "24 bit noise" but it will be overwhelmed (masked) by louder system noise upstream.

The bit length is only a potentiality. The signal chain is only as good as its weakest link, in this case the least noisy link. As with good picture, good audio starts at the very beginning of the production chain and at every step along that chain. The delivery format is only one link at the end of that chain. That is what is so often not understood by people who don't know any better. Usually they're just consumers of other people's work.

Just for interest, VHS can have two soundtracks, one about 9 bit, the other about 14 bit, and that's best case.

Hmm.... Something new to me because I never did such foolish thing as to convert lover bitrate to higher, only vice versa (24 bit is better for editing but final can be 16bit without problems, however today, because CD mastering is in past, this conversion is not actual any more).

But it should not be true Audition simply add 8 zeros. It is brutal conversion. Did you tried to convert bitrate in edit view under "convert sample type"?

-- merged --

Try following (at least under Audition 3.0 what is still the best for me):

In edit view under Convert sample type convert 16bit audio file to 32bit (float).
Then under "save as" click options and choose 24bit packed int (type 1 24bit).
Enable dithering.

Save. Test with your plugin what it say :)