Most customers want masters at 44khz, but occasionally people want them at 96khz. There are two issues with this:
If I capture the master at 44khz (which I think sounds better than sample-rate-converting 96khz down to 44khz), then to give them a master at 96khz, I have to upsample to do so - it’s kind of pointless.
If I capture the master at 96khz, in order to give them a master at 44khz, I have to sample-rate-convert.
AVAILABLE OPTIONS) I can do two masters - one at 96khz, and one at 44khz, but I have to recreate all of the fades, etc. - or I just sample-rate-convert…
So I thought of a potential solution… What if Wavelab could capture as DSD (just like the Korg MR-2000 or MR-1000)? And then DSD tracks would have selectable “decimation” settings - to output the track at any sample-rate and bit-rate desired. It would be a similar thing to the sample-rate-conversion we have currently built-in the master section, but it would be real-time decimation instead. The great things about this?
A) All fades and settings and any other processing happens post-decimation (at whatever sample-rate & bit-depth we set). If we want to work with our favorite ITB EQ at 192khz because we prefer the sound at that rate vs 44khz, no problem.
B) Fades could be left the same! There would be no need for multiple montages at different sample rates… no need for sample-rate conversion… One single DSD montage could do the entire process, and output any sample-rate or bit-depth requested… with all fades and plugins/processing at any sample-rate and bit-depth we want.
I imagine this might be a lot of work to implement, but the idea just popped into my head so I figured I’d share it. There would be no need to actually edit audio as DSD… all editing would remain as is, but being able to record/playback DSD and decimate to whatever rate would be cool.
There is no need to recreate all the fades to make masters of different sample rates. Check for the "Create Customized Montage Duplicate" feature that lets you recreate a montage at a new sample rate which is great if you like to use a 3rd party SRC on your files. All the fades, markers, metadata etc. are preserved. It’s a beautiful thing.
I always work from a 96k moorage and print everything to a new audio file at 32-bit floating point and 96k and create a new montage from the resulting file. All the markers, metadata, etc. are carried over to the resulting montage but having the audio printed to a new file in one render pass helps avoid some minor glitches with overlapping tracks etc.
From here I can i insert a dither plugin and render 24-bit/96k WAV files of each track.
Then I SRC that source 32-bit fp/96k WAV down to 32-bit fp/44.1k using Saracon and place it in a new sub-folder.
Then I can run “Create Customized Montage Duplicate” to recreate my 96k montage (the one with the fully printed audio) at 44.1k. Within a second I have a new montage that perfectly matches the source montage aside from the sample rate.
Now I change the dither from 24-bit to 16-bit (because in both cases, the source file is 32-bit fp) and I"m ready to render 16-bit/44.1k WAV files of each track, and/or a DDP. I can also remove the dither altogether and render reference mp3 files.
It’s a lot in writing but it’s really simple and effective. I would argue that while upsampling from 44.1k to 96k alone doesn’t add anything to the material, that doing all the digital (and analog) processing at 96k can yield better results than working at 44.1k if a good SRC is used and you’re careful about bit-depth management.
I basically work my way down to 44.1k but no matter how you want to work, “Create Customized Montage Duplicate” can be a big time and accuracy saver.
Interesting idea, If capture is the way you work. I guess you could sort of test it out if you have a DSD ADC record setup, DSD to a couple PCM sample rates using one of the options out there for that (or maybe you’ve done that already?), and use the montage copy function Justin suggested.
Not only that, HD Tracks won’t accept it if it doesn’t have frequency content indicative of the rate of the file, which can’t be higher than the mix file. The frequency content won’t be there.
I have experimented with capturing DSD files on my DSD recorder, and then working at 96khz. It works great and sounds amazing, but it’s a bit of a pain to capture there and then import to Wavelab, but I can probably figure out a way… Maybe this "Create Customized Montage Duplicate can help make this possible.
STILL, it would be even more convenient if Wavelab would handle the DSD directly as I mentioned above. Then, I could capture files directly into Wavelab as DSD, decimate appropriately to work at any sample-rate or bit-depth that I might prefer, and have only one montage handling EVERYTHING. This would be pretty cool.
Oh wow… So they won’t even take an up-sampled version…
I don’t foresee DSD happening any time soon but until then, I think the “Create Customized Montage Duplicate” can save you a lot of time and accuracy. The main thing is to keep the file names the same, just have the alternate sample rate files in their own sub-folder but with the same file names. There is a prompt to help work around some changes in the file names but i find it easiest to keep the file names the same and then keep the .mon in the same folder as those files.
You may want to consider upsampling to 96k before starting. The SoX SRC that was added to WaveLab 9 isn’t bad but I don’t think I’d want to run every project through it which is why I use Saracon (RX and Myriad are also very good) and why this feature is so useful to those who wish to use a 3rd party SRC.
Thanks I’ll try this. It’s a bit tedious, but for the occasional customer that wants a 96khz master, this may work well for me.
I’m confused - why upsample? In my case I’d be starting with a DSD capture… decimating to two files for each song (one at 44khz, one at 96khz), and then running a montage for 96khz - doing all fades, etc. Then use the “Create Customized Montage Duplicate” to do a montage at 44khz that has the same fades… I “think” this is how it would work - not sure I have the concept right though yet as I haven’t done it from beginning to end. Is this correct?
I think upsampling from 44.1k or 48k to 96k before doing any digital and/or analog processing can sound better than staying at those lower sample rates. This is especially true if you like to clip the input of the A/D converter on the way back in a little bit (or a lot).
As a bonus, you always have the ability to provide useful 24-bit/96k hi-res masters and also create the vinyl pre-master at 96k.
Upsampling alone doesn’t do any good (and can do harm), but upsampling with a quality SRC and then doing additional processing I think can have positive benefits. Even when using plugins that up-sample/oversample I think there is benefit to getting up to 96k with a quality SRC first.
It may be worth testing with your workflow to see if you agree but I don’t think I’d do it with the SoX built into WaveLab.
I see what you are saying as far as possible benefits to all processing being done at higher sample rates - I agree that it tends to sound better.
The way I work, is with two totally separate computers - one handles playback (at whatever sample rate), and then I capture/record into Wavelab on the 2nd computer at 44khz. I use the “input plugin” on the 2nd computer so I can do all of the work in real-time in both computers and in the analog domain. I try to handle as much processing as I can in the playback system (which would be 96khz in this case), and in the analog domain prior to 44k capture… Then finish in a 44khz Wavelab montage. It works very well and avoids entirely any sample-rate-conversion at all - unless the customer wants both a 96khz AND a 44khz master. That would require sample rate conversion - or integration of my DSD recorder (as below). That’s why I posted this thread with the DSD idea.
As far applying DSD the way Wavelab is currently implemented, I’d end up doing 96khz processing in computer #1 (I usually don’t upsample - although from what you’re saying, maybe with Saracon it could be worth it), playing back through analog equipment, capturing in my DSD recorder at 5.6mhz, using that to decimate to two separate files for each song (one at 96khz, one a 44khz), importing the 96k stuff into Wavelab for the final montage, and when done use the custom montage duplicate feature to recreate the same montage/fades to apply to the duplicate 44khz files… Sounds great and entirely avoids sample-rate-conversion, but it’s tedious… but hey, we are mastering people… trying our best to aim for perfection…
Yeah, there a lots of ways to work but I mostly wanted to point out that the “Create Customized Montage Duplicate” can save you a lot of time and headache but recreating the montage at another sample rate using another source file in one second.
I actually do all my digital pre-processing and analog processing in REAPER at 96k (or 88.2k if the files come in at 88.2k) so by the time the files reach the WaveLab montage, all they need is a bit of digital limiting and the occasional CLIP FX for an EQ touch or something minor.
I like to do all the final limiting and processing (other than dither) at 96k and render it to a new montage so all the plugin work is printed. From there it’s super easy to create any bit-depth/sample rate needed. I often make 48k masters for clients too so that if and when they make any video content, they can use 48k WAV masters from me rather then letting their video editor mangle the audio using who knows what.
With “Create Customized Montage Duplicate”, it’s a matter of a few clicks and a few seconds.
This is HD Tracks policy about upsampling and bit depth:
I’ve heard it said beyond what is on the page, that they can tell what is original ultrasonic content, and original low level signal below 16 bits, beyond just the visuals. Their policy is to not sell something that was recorded or mixed at 44.1, 48, or 88.2 as 96K Hi Res, no matter how it’s mastered (except according to the rare exception disclaimer on the page). The master might look like 24 bit or look like it has a bit of ultrasonic content from the analog or digital mastering process, but I’ve heard they can tell the difference. They’ve even released albums with tracks at different sampling rates because the track sources were at different sampling rates.
I would think other hi res and Blu-Ray audio sellers probably have similar policies. Things have probably slipped through the cracks over the years, but that’s what they attain to, and what they will reject.
To me it sounds like Mastering at a higher rate is acceptable (they notify the customer that the source material may be lower resolution, but the work was mastered at a higher resolution). Here is the exact quote from that page:
In the few exceptional cases where an album was mastered at a higher resolution than the multi-tracking or mixing. HDtracks clearly notes this in detail directly on the album page to the best of our ability to ensure transparency to our customers.
Toader you’re right but they specifically say “in the few exceptional cases”. I don’t think this is common for them at all, and they generally ask for a file at the native rate if they see something like this. Which would involve emails or the like, and it would still be marked with disclaimer on the HD Tracks page. Maybe they’ve had a few cases in the past where the client insisted, like if they mastered a 44.1 project straight to DSD, released it on DSD on a DSD site, and didn’t feel they should have to downsample all the way to 44.1 for the HD Tracks release (HD Tracks doesn’t release DSD). Or maybe the mastering engineer used a box or something that created ultrasonic harmonics and felt the master would need to stay in a range where that box effect wasn’t chopped off…
I think it’s worth looking at some upsamples in RX, after you’ve mastered the upsamples, exporting just a selection up in the 35k range from RX (view at linear scale) and then playing that 35k selection export at 1/4 speed in Wavelab to reduce the pitch to the audible range. There’s probably nothing there that correlates musically like there is in a normal 96k file (if you crank it up a little you should be able to hear it in the normal 96k). But in the upsample, there would probably be nothing at all up there.
HD Tracks has a free sampler worth getting to look at what they’re looking at, and listening to. The 2017 sampler has tracks at different sample rates, because those would be the native sample rates.
Or maybe just write HD Tracks and ask. I think it’s a legit thing to upsample if you think it’s beneficial for the plugins, but I think it would require a disclaimer that they wouldn’t usually use, so probably a lot more back and forth with the record label and with HD Tracks.
Does anyone know if Tidal sells 96k files? Haven’t gone to look at their policy if they have one.
EDIT: I guess HD Tracks does sell DSD files now. They didn’t used to.
btw, I tried to play spectrum selection around 35k with varispeed in Wavelab to bring the pitch down into the audible range, but it plays the whole frequency range, not just the selection, when varispeed is on. Also I didn’t see any way to export just the frequency and time selection rectangle like you can do in RX. That would be nice to have in Wavelab if it’s not already there and I just don’t see it. I tried render Selected Audio Range, but the render includes the full frequency spectrum, not just the frequency selection. RX exports just the selection.
The file I was using in the picture was 44.1 upsampled to 96k, showing basically nothing above 22k.
After looking at some more tracks (like on the sampler, a few of which might be questioned), I’m inclined to say you probably don’t have too much to worry about really. There are so many variables it will probably matter less rather than more in the future. But I still think it’s worth looking at your upsampled files after mastering, and maybe ask the question to HD Tracks or other hi-res sellers what their policy actually is, and is it changing.
