Blackmagic Forum

Still no way to output anything but 48khz audio

On the deliver page there is no way to choose sample rate.

96khz audio imports fine, but there is no way to export it.

This is a serious omission when working back/forth with protools or for
using Resolve for final conform.

DCP output support 96khz, but no way to output 96khz pcm through the
easydcp plugin so I have to render the the j2k picture elements through the
plugin and then use the standalone easydcp to combine them with the 96khz
sound track.

Does this mean that all internal audio processing is happening at 48khz?

So if I import 96khz sound it gets truncated to 48khz before being
processed by a plugin?

I know the decklink hardware does not support 96khz, monitoring is not
essential but being able to pass through 96khz from source file to output
file is.

A year later - is this still the case? Is there NO 96khz audio output in davinci resolve?

Even with new Fairlight integration, if they want to make this a robust audio finishing tool, sample rates above 48k need to be properly handled. Until then, I'll continue to use Nuendo with a "nice effort" pat on the back for Resolve.

Going on 2 years, 96khz audio imported is still truncated to 48khz, and still no way to output 96khz.

It makes handing off to protools or making a dcp deliverable a ton more work than it needs to be.

I’ll review again post NAB but a quick search of the forum finds less than 10 requests, thus it’s not been pushed up my list. Can you define who requires deliverables at this rate?

Peter Chamberlain wrote:I’ll review again post NAB but a quick search of the forum finds less than 10 requests, thus it’s not been pushed up my list. Can you define who requires deliverables at this rate?

Don't quote me on this, but I was under the impression that the Dolby ATMOS spec required audio at 96khz.

Personally I think it's overkill, and while an argument could be made for recording audio at 96khz, I don't think it makes any more sense as an end delivery format than doing a video output at 8K. Nonetheless, it would be a good thing for Resolve to support 96khz as enough people do get delivered final mix files at that rate and it would be nice to keep the quality all the way to delivery.

Kays Alatrakchi wrote:I was under the impression that the Dolby ATMOS spec required audio at 96khz.

Dolby's Atmos RMU only accepts 48kHz.

DTS:X supports 96kHz for object-based mixes, and 192kHz for stereo and multichannel mixes.
Auro-3D also supports 96kHz.

Neither DTS:X nor Auro-3D require 96kHz audio deliverables.

Peter Chamberlain wrote:Can you define who requires deliverables at this rate?

Netflix has moved in that direction by now preferring 24/96kHz deliverables, but they aren’t required quite yet.

Netflix Originals: Production and Post-Production Requirements v2.1
All final sound mixes (near-field 5.1 and 2.0), Stems, M&Es
48kHz/24-bit minimum, 96kHz/24-bit preferred

M&E Creation Guidelines
Please deliver a fully-filled Music & Effects mix in both 5.1 (6 + 2) and 2.0
(48kHz/24-bit minimum, 96kHz/24-bit preferred)

It may be questionable as end listening format, but I would say it's not that questionable as an exchange format when further steps may add own processing. 96KHz export is present in about any (if not all) NLEs out there. Surely it should be an option in Resolve regardless if only 10% people may need it. Sometimes you just need to pass it.

There are 2 problems: 
1. 96khz Deliverable  -  Netflix prefers 96k, all flavors of DCPs support 96k except atmos.  
In addition, as most projects have a 24fps deliverable for DCP and a 23.976 deliverable for Bluray/Streaming/etc - it is preferred to at least do the conversion from 96khz.

Currently we render dcp output in Resolve, then use EasyDCP to add the 96khz sound - adds a considerable amount of time when your on a deadline.


2. Truncating 96khz source on pass through.  This is the major problem.  Most film productions and all concert films record 96khz.  

If Resolve is used to create dailies/proxy for offline, sound is clipped to 48k.

If Resolve is used for the edit, 96khz source BWF is clipped to 48khz on import.  

There is no way to use internal resolve/fairlight for audio mixing and preserve the the 96khz source for processing

There is no easy way to export to Protools  and preserve the 96khz source - the resulting PCM files are clipped to 48k

Conformalizer and Ediload which are always suggested on here by people who have never used them,as they do not work for this application - they are for conforming edit changes, you need to have a protools session to work from, they conform, they do not convert.

Currently its a convoluted process exporting to Premier XML and using AAtranslator, but in the process automation, clip names, etc. are lost - far from ideal, the mix studio is not happy as the resulting protools session levels does not match the edit.

Decklink hardware is currently limited to 48khz, 96khz playback is not essential - just output.

A compromise would be a working Protools AAF reference export - not embedded media. This would allow cutting in Resolve, and exporting an AAF that points to the original 96khz media for working in protools. Doesn't solve the daily or deliver problem, but an AAF reference export should be much easier for you to implement. The current Resolve Avid AAF export does not work for this, nothing re-links in protools.

We just need to be able to hand off sound without compromising it.

Thanks for listening.

Peter Chamberlain wrote:I’ll review again post NAB but a quick search of the forum finds less than 10 requests, thus it’s not been pushed up my list. Can you define who requires deliverables at this rate?

Hi Pete,

Who requires Delivery in 8K?

Just because it isn't required now, doesn't mean that it won't be required at some stage in the future.

Netflix "prefers" 24/96 delivery. NHK will possibly demand 24/96 for their super high vision format when it launches. Any iPhone 7 or greater will play 24/96 audio, the iTunes music store has been accepting 24/96 Masters for years (still not selling in 24/96 yet though). In the media, 24/96 is being referred to as "HD Audio."
Plus, as soon as one distributor starts advertising HD Audio, or 4K Audio it will be full steam ahead for everyone:
"Did you know most TV audio is the same quality today as it was 30 years ago? You don't watch TV on an old square box anymore, so why are you still listening to SD Audio? Only with Nitflux 4k can you get the full 4k experience for both your eyes and your ears. Upgrade to 4k today, and leave SD audio in the 90's where it belongs."

Andrew Kolakowski wrote:It may be questionable as end listening format, but I would say it's not that questionable as an exchange format when further steps may add own processing. 96KHz export is present in about any (if not all) NLEs out there. Surely it should be an option in Resolve regardless if only 10% people may need it. Sometimes you just need to pass it.

I would not need it as a deliver format but I usually work in 96Khz when moving audio between programs so I would appreciate it for that reason alone.
Edit: the movie I'm working on now has 96Khz recorded for all audio.

/Andreas

I just thought I would post some thoughts on this.
I'm currently reviewing our media from the jungle of Sri Lanka.
First got 9TB of Red 5k media. For sound we recorded mono and/or stereo. All the recordings were 96kHz, 24bit.

Now the main reason for me to record in 96kHz is not to deliver with it, but for post processing. Mainly:
- Izotope Noise reduction, frequency repair etc. If I have 96kHz this will just make Izotope much more accurate (same in Audition of course).
The reason is that usually when we record sound in the jungle there is always some background noise. You can hear a loud car in the background, some idiot guy with the camera (me) pulls a zoom that you can audibly hear etc. Izotope or Audition can really clean this up working in frequency mode.

The way I would work with this in Final Cut Pro X is I would send over to Pro Tools and keep working with the audio in 96kHz format. Is that not possible currently in Resolve?
I guess a workaround would be that I process the material first in Izotope and renders them in 48kHz. Then later I could do the final master in for example Pro Tools or maybe even in Resolve.

Btw, I love the Fairlight implementation so far even though I miss several things, but what is there now is very nicely designed.

A few things I missed from an audio workflow currently:
- 96kHz as mentioned
- I miss that I cannot trim audio files like we can with video. They just end up with the same length. First step in my organization process is trimming video and audio. If I try trimming an audio file it just spits out an identical length file to what I started with.
- I really miss a way of renaming the source audio file. In Audition I can set Ranged Markers and then export all of them and USE the markers name to rename the audio file. That way you can get an audio file called SriLanka_Peacock etc. A feature like this in the Media Management Copy module with rename from DisplayName or metadata would be a godsend for me since then I could do more of our media work inside Resolve instead of being in Audition. I would like a feature like this for videos too. Just Copy and rename to Displayname.
- I love having the audio preview in Frequency Domain. It gives a visual indication of what the sound will sound like. I have seen this in any NLE, but it would make working with audio more pleasant in Resolve.

/Andreas

AndreasOberg wrote:I guess a workaround would be that I process the material first in Izotope and renders them in 48kHz. Then later I could do the final master in for example Pro Tools or maybe even in Resolve.

Just keep the source media as 96kHz, or whatever the native rate is.

Resolve supports all rates up to 384kHz, and downsamples on the fly to 48kHz.

AndreasOberg wrote:Now the main reason for me to record in 96kHz is not to deliver with it, but for post processing. Mainly:
- Izotope Noise reduction, frequency repair etc. If I have 96kHz this will just make Izotope much more accurate (same in Audition of course).
The reason is that usually when we record sound in the jungle there is always some background noise. You can hear a loud car in the background, some idiot guy with the camera (me) pulls a zoom that you can audibly hear etc. Izotope or Audition can really clean this up working in frequency mode.

I actually tested this once. I recorded some dialog at 96KHz, and also at 48KHz. There was some fairly noisy HVAC noise in the recording, so I used Reaper's noise reduction on it. With the 48KHz recording, the result was... nasty. Lots of digital distortion. With the 96KHz recording, it was surprisingly clean.

ChristopherSeguine wrote:Most film productions and all concert films record 96khz.  

Hold on: name one. All the feature films and TV shows I've ever worked on (about 700, last time I checked) have been 48kHz / 24-bit. Not one of them has been 96kHz or above, including some major concert projects like the Rolling Stones Shine a Light film. I glanced at your IMDB resume and those projects are not familiar to me. I just delivered projects to Disney and Sony within the last few weeks, and believe me, they were all 48kHz. [I don't dispute that Netflix is among the few companies beginning to allow 96kHz / 24-bit deliverables, but I would bet you that all the location dialogue is still being recorded at 48kHz.]

I think you can make a good case for 96kHz for classical concert work going to SACD (at least the few thousand people who have an SACD player or a high-res streaming device), but I think this is a niche market at best. I buy some of these releases myself, but I don't consider it a mass-market format... especially for video.

Conformalizer and Ediload which are always suggested on here by people who have never used them,as they do not work for this application - they are for conforming edit changes, you need to have a protools session to work from, they conform, they do not convert.

They're only editorial tools -- they just rebuild an XML edit list and were never intended to convert. If you need to take a picture edit session and rebuild it in Pro Tools, both will work, as will Virtual Katy. They also work well as change tools, keeping up with picture edits over time. 48kHz / 96kHz / 192kHz are not part of what they do in any way.

But these accessory programs do work and they've saved my sound crews tons of work on the indie films we've done. There are some kludges involved, particularly from FCPX, but the smart sound editors know the workarounds and still get the job done.

Fairlight is an interesting question mark right now, but while it's clearly a work in progress, it's a step in the right direction. I think we'll see big changes at NAB in a month, and I think it'll be a good option for people looking for this kind of flexibility between picture editing, color, and sound.

Marc Wielage wrote:Hold on: name one. All the feature films and TV shows I've ever worked on (about 700, last time I checked) have been 48kHz / 24-bit. Not one of them has been 96kHz or above, including some major concert projects like the Rolling Stones Shine a Light film.

David Gilmour: Live At Pompeii.
Sting: Live in Berlin
Pink: The Truth About Love Tour, Live from Melbourne

Majority of Concerts are still 48kHz though, including those from U2, the Rolling Stones, et cetera.

But, Blu-ray format in its current form provides enough capacity for higher sampling rates, and the Blu-ray Disc Association intends on increasing capacity even further to 128GB to accommodate 8K video streams, NHK22.2 arrays, and still with available capacity for 24/192kHz audio streams.

I think you can make a good case for 96kHz for classical concert work going to SACD (at least the few thousand people who have an SACD player or a high-res streaming device)

Majority of classical recordings for SACD are recorded at the native DSD64 rate, with a minority recording in DXD (aka 352.8kHz).

Increasingly for DSD downloads, DSD128 and DSD256 are preferred, such as the many albums available from https://www.nativedsd.com

We all know 96kHz output will be added to Resolve at some point. Is this a big thing to add?
I would need this on 2 projects I'm working on right now.

Kays Alatrakchi wrote:Don't quote me on this, but I was under the impression that the Dolby ATMOS spec required audio at 96khz.

This statement is now true.

Dolby have just announced 96kHz support for Atmos Masters at NAB 2018.

Count me in for wanting support up to 192khz. I'm working with original dsd files that im converting to 176 and would love to keep the high resolution for my own archives. It's a shame so many people here think that because they don't need it that nobody else should. Well, some of us need it and would love to use it. Im afraid I may have to use another NLE to work with some of these high resolution files.

New forum member here but long time user of Resolve.

I would also like this feature as I'm also working pretty much exclusively 96Khz and would prefer to keep that pipeline as long as possible.

Being able to export a master video file with the audio as 24/96 that can be used as the mezzanine file from which all the various encodes can be done would be ideal and the most future proof.

It's really surprising that with the integration of Fairlight as a complete audio production suite designed to replace Protools or Nuendo, there would be the inability to work at sample rates other than 48Khz.

The well respected Danish mastering engineer Lagerfeldt did extensive testing and found for music videos being uploaded to Youtube, the best audio sample rates for uploading were 44Khz & 96Khz as in his analysis of the audio AFTER encoding by Youtube, the native 48Khz audio had aliasing.

siddhu wrote:The well respected Danish mastering engineer Lagerfeldt did extensive testing and found for music videos being uploaded to Youtube, the best audio sample rates for uploading were 44Khz & 96Khz as in his analysis of the audio AFTER encoding by Youtube, the native 48Khz audio had aliasing.

Why would 96kHz audio have less aliasing than 48kHz audio on YouTube? I think this is a specious argument, especially considering the massive amount of compression YouTube does in general. That's a very, very low bar in terms of sound & picture quality.

Three years is a lifetime in software development.

Still no 96khz file output in V16?  

Again, don't care about being able to monitor in 96khz, just need to pass though the 96khz source to dcp/imf pcm or at least wav for delivery.  Its a big pain to have to allocate time to rewrap everything when your under deadlines.

Thanks for adding 96khz DCP/IMF import, at least I can play them back now.

Did you read the release notes?

How to actually export it, I don't know. Maybe it's yet to be implemented in the beta, or it's only for certain types of export.

Edit: It wasn't the release notes, but the New Features Guide for v16:

Increased Sample Rate Support
The project settings of DaVinci Resolve 16 now support “high resolution” audio with 96k and 192K sample rates for recording, mixing, and output through supported audio interfaces.

NOTE: While your project can be set to use and render high resolution audio, DeckLink hardware always downsamples the audio output to 48K because of the hardware it uses.

Hmm, the way that's worded it seems it might not be possible to render, "output through supported audio interfaces". But in the note below it it does say "render".

I read the release notes, didnt see the feature guide.

There is no Project setting or deliver page option to set sample rate.

Does not seem to be in the Studio version beta1 as far as I can tell?

Rendered a 96khz source, ended up with a 48khz IMF.

If you read this note it says nothing about export. It's rather about monitoring.
Resolve is the only NLE locked to 48KHz export. It really needs to be sorted.

Andrew Kolakowski wrote:If you read this note it says nothing about export. It's rather about monitoring.
Resolve is the only NLE locked to 48KHz export. It really needs to be sorted.

We record everything in 96kHz so would also really appreciated if we had export possibility in 96kHz

As I understand this has been sorted now in 16.1, right?
At least Fairlight can work with higher frequencies and I would sure hope that its possible to export with those as well.

Best,
Andreas

AndreasOberg wrote:As I understand this has been sorted now in 16.1, right?

Correct.

Resolve 16.1 Studio includes the setting of the Timeline rate (i.e. 48k, 96k or 192k) in Project Settings.

That is fantastic news. Very useful for projects with more complex audio processing.
Version 16 of Davinci has got tons of features that I have missed, very promising.
/Andreas

hardware must accept 96k or 192k because attention to POP, breath or strange sounds at the start of reading ...