The No.1 Website for Pro Audio
the other way 'round..
Old 16th March 2021
  #1
Lives for gear
 
🎧 10 years
the other way 'round..

Don't know how many of you are AES members. There's an interesting paper regarding automated mixing in the current issue of the AES journal.

Really interesting. Might it be a possibility that it changes from human-made mixes passed to a automatic mastering system to automated mixes passed to human mastering?

Here's the synopsis. Emphasis mine:
Quote:
The development of intelligent music production tools has been of growing interest in
recent years. Deep learning approaches have been shown as being a highly effective method
for approximating individual audio effects. In this work, we propose an end-to-end deep neural
network based on the Wave-U-Net to perform automatic mixing of drums. We follow an
end-to-end approach where raw audio from the individual drum recordings is the input of the
system and the waveform of the stereo mix is the output. We compare the system to existing
machine learning approaches to intelligent drum mixing. Through a subjective listening test we
explore the performance of these systems when processing various types of drum mixes. We
report that the mixes generated by our model are virtually indistinguishable from professional
human mixes
while also outperforming previous intelligent mixing approaches.
Old 16th March 2021
  #2
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by kosmokrator ➡️
Don't know how many of you are AES members. There's an interesting paper regarding automated mixing in the current issue of the AES journal.

Really interesting. Might it be a possibility that it changes from human-made mixes passed to a automatic mastering system to automated mixes passed to human mastering?

Here's the synopsis. Emphasis mine:
Not an AES member, I haven't read the paper. However I ask:

1) "virtually indistinguishable"
How indistinguishable is "virtually"? Crappy qualifying word.

2) "from professional human mixes"
Which professional? Tchad Blake? CLA? "No-name" dude like me? I mean sure the end game in 100ish years is probably a dropdown menu where you can just pick whose aesthetic you want and it happens instantly but we're not there yet...

What's the name of the paper? Are there audio examples to qualify the observational remarks?
Old 16th March 2021
  #3
Lives for gear
 
zvukofor's Avatar
 
🎧 10 years
In all such cases of development of automated decision-making systems i want to ask one question: what for? What’s the point?

Isn’t the meaning (one of meanings) of an art - people are making creative decisions, not perfect ones, but meaningful in artistic terms. How any machine could do that?... same question i have for any music production mimicking (not “inspired by”) other music production, from composing to mix to master...”i want it to be like _insert popstar name here_”. What’s the point?
Old 16th March 2021 | Show parent
  #4
DAH
Lives for gear
 
DAH's Avatar
 
🎧 10 years
Quote:
Originally Posted by zvukofor ➡️
In all such cases of development of automated decision-making systems i want to ask one question: what for? What’s the point?

Isn’t the meaning (one of meanings) of an art - people are making creative decisions, not perfect ones, but meaningful in artistic terms. How any machine could do that?... same question i have for any music production mimicking (not “inspired by”) other music production, from composing to mix to master...”i want it to be like _insert popstar name here_”. What’s the point?
Art is not needed,only content.
Old 16th March 2021 | Show parent
  #5
Lives for gear
 
Greg Reierson's Avatar
 
Verified Member
5 Reviews written
🎧 10 years
Quote:
Originally Posted by DAH ➡️
Art is not needed,only content.
"Virtually indistinguishable from music!"
Old 17th March 2021
  #6
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
What's ironic is that the only reason we originally used automation was so that we could run multiple first-generation copies of a mix. Automation was a step down in artistic quality to most people but saving a tape generation seemed worth it. Then it evolved into mix by committee.
Old 17th March 2021
  #7
Lives for gear
 
John Moran's Avatar
 
Verified Member
🎧 10 years
The computer mixing algos will never crank a preamp into overdrive distortion the way Martin did for Lennon to get "that sound" for the first time.
Algos follow parameters and instructions well but they aren't particularly creative.
I'd like to see one try to comp vocals too.
Old 17th March 2021
  #8
Gear Addict
 
🎧 10 years
For a company busting out thousands of hours of background music to a corporate brief like hold music or background music for a restaurant chain or shopping centre or airport, or any kind of library music, this is absolutely going to be used. Those jobs are all about throughput and efficiency and providing something that sounds acceptable to their clients, as quickly and extensively as possible, so the more automation the better. I'm sure a lot of this stuff is composed with input from algorithms too. Yes, it's replacing human jobs in the very near future if not already, but only those working on very high-throughput work that often doesn't require a huge amount of creative input anyway.

Even if the technologies somehow become technically advanced enough to compete with the artistic vision of human creators, there's already more music created every year by humans than anyone could listen to in a lifetime. Also, basically a lot of people get as much enjoyment from participating in the creation as others do from listening to it. Is there any benefit to increasing efficiency for this? Or to put it another way, if humans stop expressing themselves, then what's the point in anything?
Old 17th March 2021 | Show parent
  #9
Gear Guru
 
Verified Member
1 Review written
🎧 5 years
i got into music and into this business to interact and communicate with human beings and therefore still exclusively work on 'attended sessions' - is it about the process or the result? both i'd say but if the former gets automated, the latter becomes worthless, at least to me...

waiting for the computers to build drums and cymbals, set them up, select mics, position them, play the instrument, track while i'm having a coffee with the drummer/percussionist!

kinda funny we had a somewhat similar discussion some 30 years ago... - oh, and i have an alesis hr-16 for sale here :-)

Last edited by deedeeyeah; 17th March 2021 at 11:47 AM..
Old 17th March 2021
  #10
Lives for gear
 
Jerry Tubb's Avatar
 
Verified Member
🎧 15 years
We should distinguish between Automated mixing, where the automation was programmed by human a mix engineer & producer,

From A.I. “mixing” with SkyNet at the helm.

I think most of us have used Automation in mixing or even Mastering, to control fader levels.

Imagine a roomful of AI Robots sitting around SkyNet nightclub, listening to computer generated music, all samples, AI mixed and mastered,

laughing smiling and sipping cocktails filled with maintenance nanorobots....

pretending to be human. exciting isn’t it

I’m staying with Magnus, Robot Fighter!

JT
Attached Thumbnails
the other way 'round..-60bfc347-d567-4b08-a29a-b154de2c2f10.jpg   the other way 'round..-2a7bfaca-85b0-413f-ac36-d844d0500ed7.jpeg   the other way 'round..-e95a36df-08fc-4231-8df4-986df8ba0ed7.jpeg  
Old 17th March 2021
  #11
Lives for gear
 
Jerry Tubb's Avatar
 
Verified Member
🎧 15 years
Wish I had gotten in on the “why are you still using hardware” thread before it was locked!

Imagine a SkyNet robot sitting at my console, listening carefully, turning hardware knobs,

and the AI Robot producer looking over his shoulder asking

“what are you doing now?”

and being handed the placebo fader....

LoL... JT
Old 17th March 2021 | Show parent
  #12
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by Teej ➡️
Not an AES member, I haven't read the paper. However I ask:

1) "virtually indistinguishable"
How indistinguishable is "virtually"? Crappy qualifying word.
This is a scientific paper, so this was done properly, according to a modified MUSHRA test, as laid out in ITU-BS.1534. In the paper, the section detailing the quantitative and qualitative results of the listener tests take up multiple pages.

Of course I could only post the synopsis of the paper here.

Quote:
2) "from professional human mixes"
Which professional? Tchad Blake? CLA? "No-name" dude like me? I mean sure the end game in 100ish years is probably a dropdown menu where you can just pick whose aesthetic you want and it happens instantly but we're not there yet...
it says 'human professional engineer'. Adjusting for the specific style of name engineers shouldn't be that hard, as there is enough material for training the deep learning algorithms out there for each of these guys.

Quote:
What's the name of the paper? Are there audio examples to qualify the observational remarks?
"A Deep Learning Approach to Intelligent Drum Mixing With the Wave-U-Net"
Old 17th March 2021
  #13
Lives for gear
 
🎧 10 years
Interestingly some people here were referencing that algorithms don't operate under context.

Another paper in the same AES journal is named:

"Context-Aware Intelligent Mixing Systems"

Synopsis:
Quote:
Intelligent Mixing Systems (IMS) are rapidly becoming integrated into music mixing and production workflows. The intelligences of a human mixer and IMS can be distinguished by their abilities to comprehend, assess, and appreciate context. Humans will factor context into decisions, particularly concerning the use and application of technologies. The utility of an IMS depends on both its affordances and the situation in which it is to be used. The appropriate use for conventional purposes, or its utility for misappropriation, is determined by the context.
This study considers how context impacts mixing decisions and the use of technology, focusing
on how the mixer’s understanding of context can inform the use of IMS, and how the use of IMS can aid in informing a mixer of different contexts.
Old 17th March 2021 | Show parent
  #14
Gear Guru
 
Verified Member
1 Review written
🎧 5 years
Quote:
Intelligent Mixing Systems (IMS) are rapidly becoming integrated into music mixing and production workflows.
really?! - to me, this 'sounds' more like marketing bs than even a half-baked hypothesis worth being considered in a scientific paper...
Old 17th March 2021 | Show parent
  #15
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by deedeeyeah ➡️
really?! - to me, this 'sounds' more like marketing bs than even a half-baked hypothesis worth being considered in a scientific paper...
It's however true. The facts don't care about your feelings.

https://alexaudiobutler.com/
https://lawo.com/automix/

Just some examples. Are they 'intelligent'? No. Are they becoming more intelligent with each iteration? Hell yeah!
Old 17th March 2021 | Show parent
  #16
Gear Guru
 
Verified Member
1 Review written
🎧 5 years
Quote:
Originally Posted by kosmokrator ➡️
It's however true. The facts don't care about your feelings.

https://alexaudiobutler.com/
https://lawo.com/automix/

Just some examples. Are they 'intelligent'? No. Are they becoming more intelligent with each iteration? Hell yeah!
indeed just some examples (and not even very good ones imo: gain sharing isn't exactly new...) - nevermind, we're still very far from the ghost in the machine taking over, except maybe some very specific applications...

(as sports reports in the media)

...and even if that'd be the case, i couldn't bother less: i'm very confident that i'll keep interacting with human beings who don't give a thing about any smart kinda ai when it comes to tracking/mixing/mastering.
Old 18th March 2021
  #17
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
The advantage of edited Arm-Strong mixing is that happy accidents will often happen that feel lots better than left-brain dominated, over-thunk automated mixes. With digital recording and DAWs, there's no reason to not go back to how mixing was done before the '80s.
Old 23rd March 2021 | Show parent
  #18
Lives for gear
 
Jerry Tubb's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by Bob Olhsson ➡️
The advantage of edited Arm-Strong mixing is that happy accidents will often happen that feel lots better than left-brain dominated, over-thunk automated mixes. With digital recording and DAWs, there's no reason to not go back to how mixing was done before the '80s.
QFT... Words to Live By!

Cheers, JT
Old 23rd March 2021
  #19
Lives for gear
 
zinzin's Avatar
i am shure this AI mixing will be the norm in pop music in maybe 5 years.

theres even this country nashville guy doing „template mixing“. he‘s written a book about it: https://www.soundonsound.com/reviews...-and-mastering

its the business: faster! cheaper!
Old 23rd March 2021
  #20
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
Faster! cheaper! and yet most of today's sales go to old pre-'80s recordings!
Old 23rd March 2021 | Show parent
  #21
Lives for gear
 
zinzin's Avatar
Quote:
Originally Posted by Bob Olhsson ➡️
Faster! cheaper! and yet most of today's sales go to old pre-'80s recordings!
yeah, off course because they overdo it. making everything sounding the same. no wonder people go back to some „real-ness“. but i am shure, AI mixing and mastering is the future for pop.
just like cloths: t-shirt that cost 4,99 €, get used 3 times than trashed ... because they are sh*t.
Old 24th March 2021 | Show parent
  #22
Lives for gear
 
Greg Reierson's Avatar
 
Verified Member
5 Reviews written
🎧 10 years
Quote:
Originally Posted by zinzin ➡️
i am shure this AI mixing will be the norm in pop music in maybe 5 years.
Could be. Current pop music would be the ideal candidate. All sounds the same. No risks. Nothing ever done that's different than anything else. Let them have it.
Old 25th March 2021 | Show parent
  #23
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by Greg Reierson ➡️
Could be. Current pop music would be the ideal candidate. All sounds the same. No risks. Nothing ever done that's different than anything else. Let them have it.
In other words, background muzak...
Old 25th March 2021 | Show parent
  #24
DAH
Lives for gear
 
DAH's Avatar
 
🎧 10 years
Quote:
Originally Posted by Bob Olhsson ➡️
In other words, background muzak...
Or, better yet, forefront ****e, which always strives to come up from the depth to the surface.

Last edited by DAH; 25th March 2021 at 01:02 AM..
Old 25th March 2021 | Show parent
  #25
Lives for gear
 
zinzin's Avatar
Quote:
Originally Posted by Bob Olhsson ➡️
In other words, background muzak...
i‘d call it „products“, fast food music.
Old 25th March 2021 | Show parent
  #26
Motown legend
 
Bob Olhsson's Avatar
 
Verified Member
🎧 15 years
Quote:
Originally Posted by zinzin ➡️
i‘d call it „products“, fast food music.
People are willing to pay for fast food but not muzak!
Old 25th March 2021 | Show parent
  #27
Lives for gear
 
zinzin's Avatar
Quote:
Originally Posted by Bob Olhsson ➡️
People are willing to pay for fast food but not muzak!
true. thats why they are cutting down budgets and AI mixing/mastering will be the future of pop music. and streaming, where major labels get paid and a few big „artists“ ...
lets hope it all backfires and alternative music gets up again! although i am not shure if alternative music wants that.
Old 28th March 2021 | Show parent
  #28
Lives for gear
 
Adam Dempsey's Avatar
 
Verified Member
🎧 10 years
Quote:
Originally Posted by Greg Reierson ➡️
Could be. Current pop music would be the ideal candidate. All sounds the same. No risks. Nothing ever done that's different than anything else. Let them have it.
The commercials/advertising are the content, after all...
Old 29th March 2021
  #29
Lives for gear
 
Edward_Vinatea's Avatar
 
🎧 10 years
Quote:
Originally Posted by kosmokrator ➡️
Really interesting. Might it be a possibility that it changes from human-made mixes passed to a automatic mastering system to automated mixes passed to human mastering?
So, I guess we should all start selling our gear now while it's still worth something, then get a real job or retire. LOL.

Assuming we don't tragically end up in a communist-totalitarian world by next year, I don't see why this is nothing but yet another vague peek into a future none of us will ever see...

FWIW,
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 50 views: 5990
Avatar for opium89
opium89 6th February 2007
replies: 708 views: 84448
Avatar for jbuonacc
jbuonacc 3rd January 2021
Topic:
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump