@@progressionspod Thanks for putting it up here. Great work! He's one of the few I can listen to for hours without an end. Just priceless information to absorb and such a humble guy himself in an industry where most are just rocking with their cocks out not being able to get over their own bloody egos. ❤
25 years ago I used to try to make MY record every single session. I realised my mistake when I’d made what I thought was an awesome record for a local band. I was so happy with it. Polished, slick. Punchy and radio friendly. The band absolutely hated it. I redid the whole session for free. It ended up raw, scruffy, never to be played on a radio anywhere. The band thought it was the greatest thing they’d ever heard. I ended up working on four records for them over the years. It’s a difficult skill to throw your ego in the bin and make whatever it is that the client wants!
I think it's actually part of the learning process. You kind of have to over do it for years to learn how have your own sound and still support the artist. That's why the great mixers are the great mixers. They've found a way to balance both.
The Shore ~ The Shore 2004 Maverick records. Best record that was never heard ~ Tracked on Trident 80 series console ~ Mixed In the box by Rick Parker on Pro tools mix plus. Beige g3
I worked at a cheer leader music production studio for a few months, and it was mind numbing to say the least. Hearing Andrew talk about editing vocals and tuning boy bands makes me laugh and take me back. Someone else already said this, but it’s true, “these interviews keep us sane”.
Haha. That’s a pretty niche production house! I did a lot of kids music for Disney and Nick. That stuff would get so embedded in my head it would make me crazy.
@@progressionspod yeah there’s a whole world of cheer competition that most don’t know about. Every song they cheer/dance to in those competitions are original pieces. The routines are given to the studio written out on paper, and then we’d make the songs to go to the routine. It was literally some paint by numbers crap. Completely soul sucking.
Just a remark on the AI-trainingdatasets: while I obviously don't know what data they used to train Spike Stent's AI thing with (I don't even know what it does, only saw the headline, never looked into it), the notion that AI can only be successful when being trained with enormous datasets is not true. There have been some researches into this lately which suggest that depending on context, small data sets are highly preferable and deliver better outcomes. I suppose the point is that it depends on how specific and narrow a task is. A LLM like GPT needs sh*tloads of data. But I can totally imagine that if the Spike AI tries to make things sound like Spike mixed it, it has been trained on just his mixes and a set of other mixes for distinction and that that could be sufficient for this task. After all, the AI is not really learning how to operate gear and push faders, in the end it's lots of spectral shaping to get to a desired end state so it goes more by sonic fingerprinting.
Yes this is true. After chatting with Andrew, I brought it up to a friend of mine that makes LLMs and he said that you could build something with significantly less data as long as you treated that data the right way. (he used bigger words, but that's what I took away 😂)
If you create an Atmos recording will the height related sounds be pushed back down until a Dolby Digital mix if Atmos is not available, or will some of it be lost in the translation from Atmos down to non-atmos systems?
The genius of the Atmos tech is that playback is decoded for the system so that things playback as closely as possible to the full mix. I don’t know to what extent top channels are blended in but the tech should handle it for you. You can pretty easily check various rerenders when you are mixing. So you could check a 7.1 if you wanted to.
Great presentation! I'm excited about AI, it's already pretty mindblowing. However I see AI as a tool and that music made by humans will still remain the leader, at least for the top productions. It's somewhat like over a century ago when photography replaced many things painters did back then, photography didn't replace all art painted by humans.
I think as AI becomes more common (and better) there will be a surge of people who will gravitate towards human made art, etc. I hope at least. Thanks for watching!
Was confused about the intellectual property session conversation. To me asking for a mixed session is a hard no, but raw multitrack is definitely fine. Were you guys agreeing with that?
Yeah, we were agreeing. I used to be happy to send a session to a label for archiving, but after talking to Andrew I think I’ll probably just print a very deep set of stems because it is so easy. That would be better for long term anyway. A session with a bunch of old plug ins is going to be worthless 20 years down the line. If you think about it like that sending a mix session is pretty pointless regardless of the IP convo.
@@progressionspod in my opinion even printing processed stems should be out. IMO you get the multitracks and the usual deliverables and that’s it. But I rarely win that battle. Especially if atmos is being done by someone else, and I’ve done half their work for em’. Haha
"ANYTIME I REALLY LIKE IT, THEY HATE IT." I just take it from there. I'm feeling so much better now.
haha. yeah that one really resonated with me. Thanks for watching!
@@progressionspod Thanks for putting it up here. Great work! He's one of the few I can listen to for hours without an end. Just priceless information to absorb and such a humble guy himself in an industry where most are just rocking with their cocks out not being able to get over their own bloody egos. ❤
25 years ago I used to try to make MY record every single session. I realised my mistake when I’d made what I thought was an awesome record for a local band. I was so happy with it. Polished, slick. Punchy and radio friendly.
The band absolutely hated it.
I redid the whole session for free. It ended up raw, scruffy, never to be played on a radio anywhere.
The band thought it was the greatest thing they’d ever heard.
I ended up working on four records for them over the years.
It’s a difficult skill to throw your ego in the bin and make whatever it is that the client wants!
I think it's actually part of the learning process. You kind of have to over do it for years to learn how have your own sound and still support the artist. That's why the great mixers are the great mixers. They've found a way to balance both.
Every few months a new Scheps interview drops and there’s always gold in there. Great questions too!
He's so gracious with his time and knowledge. Thanks for watching!
@@progressionspod Yes, I've always enjoyed interviews with Andrew and he seems like such a genuine, sweet guy.
@@robertfoshizzle Such an easy person to talk to
I watched this while lifting weights - I feel I have gotten stronger as a mix engineer watching this.
Always remember to practice gym safety 😂
@@progressionspod 💪🎧💪
So happy with an engineer like Andrew, he loves talking and has a lot of knowledge to share. Perfect combination 🤟🏼
He's so good at putting his thoughts into words. Thanks for watching!
This was easily my favorite episode, amazing work!
Nice! Glad you enjoyed it! One of my personal favorite conversations. Andrew is a joy to chat with.
The Shore ~ The Shore 2004 Maverick records. Best record that was never heard ~ Tracked on Trident 80 series console ~ Mixed In the box by Rick Parker on Pro tools mix plus. Beige g3
watching these interviews are ways to stay sane
I’m glad you dug it! Thanks for being a supporter of the show.
love this interview! thank you!!!
Thanks for checking it out!
loved this conversation. keep it up!
Glad you enjoyed it! Thanks for watching!
I love Scheps’ hot takes! He’s so often right too!
He basically speaks in clips. Haha. Thanks for watching!
This was the best episode by far and a fantastic interview. Thank you!
I agree. haha. A personal favorirte conversation for me. Glad you enjoyed it!
Great interview, really grateful for andrew's insight and honesty. here's to more gigs on music we love!
He shares so much with the our community. And is an absolute lovely hang!
Fucking brilliant. So well done. Thank you both!
Thanks for watching! Glad you dug it!
I worked at a cheer leader music production studio for a few months, and it was mind numbing to say the least. Hearing Andrew talk about editing vocals and tuning boy bands makes me laugh and take me back. Someone else already said this, but it’s true, “these interviews keep us sane”.
Haha. That’s a pretty niche production house! I did a lot of kids music for Disney and Nick. That stuff would get so embedded in my head it would make me crazy.
@@progressionspod yeah there’s a whole world of cheer competition that most don’t know about. Every song they cheer/dance to in those competitions are original pieces. The routines are given to the studio written out on paper, and then we’d make the songs to go to the routine. It was literally some paint by numbers crap. Completely soul sucking.
Luckily this stuff is anything but catchy so it wouldn’t get stuck in your head. It just sounded awful and was not pleasant to listen to.
Excellent!
It's an epic hang. Hope you dig it!
Great video and interview like always! Deffinatly resinated with a lot of the topics on this one!
Thanks! Andrew is such a wealth of knowledge and one of the best at conveying his ideas.
Salvador Dali said about rules, or I should say learning. "First learn to paint like a master and then do whatever you want."
And he definitely did that! Great quote. Thanks for sharing it!
Just a remark on the AI-trainingdatasets: while I obviously don't know what data they used to train Spike Stent's AI thing with (I don't even know what it does, only saw the headline, never looked into it), the notion that AI can only be successful when being trained with enormous datasets is not true. There have been some researches into this lately which suggest that depending on context, small data sets are highly preferable and deliver better outcomes. I suppose the point is that it depends on how specific and narrow a task is. A LLM like GPT needs sh*tloads of data. But I can totally imagine that if the Spike AI tries to make things sound like Spike mixed it, it has been trained on just his mixes and a set of other mixes for distinction and that that could be sufficient for this task. After all, the AI is not really learning how to operate gear and push faders, in the end it's lots of spectral shaping to get to a desired end state so it goes more by sonic fingerprinting.
Yes this is true. After chatting with Andrew, I brought it up to a friend of mine that makes LLMs and he said that you could build something with significantly less data as long as you treated that data the right way. (he used bigger words, but that's what I took away 😂)
If you create an Atmos recording will the height related sounds be pushed back down until a Dolby Digital mix if Atmos is not available, or will some of it be lost in the translation from Atmos down to non-atmos systems?
Dosent matter. Atmos is a money grab fad no one wants, the general public don’t even want to bother with having two speakers to listen to music.
The genius of the Atmos tech is that playback is decoded for the system so that things playback as closely as possible to the full mix. I don’t know to what extent top channels are blended in but the tech should handle it for you. You can pretty easily check various rerenders when you are mixing. So you could check a 7.1 if you wanted to.
Great presentation! I'm excited about AI, it's already pretty mindblowing. However I see AI as a tool and that music made by humans will still remain the leader, at least for the top productions. It's somewhat like over a century ago when photography replaced many things painters did back then, photography didn't replace all art painted by humans.
I think as AI becomes more common (and better) there will be a surge of people who will gravitate towards human made art, etc. I hope at least. Thanks for watching!
Yeah ,nah!
Was confused about the intellectual property session conversation. To me asking for a mixed session is a hard no, but raw multitrack is definitely fine. Were you guys agreeing with that?
Yeah, we were agreeing. I used to be happy to send a session to a label for archiving, but after talking to Andrew I think I’ll probably just print a very deep set of stems because it is so easy. That would be better for long term anyway. A session with a bunch of old plug ins is going to be worthless 20 years down the line. If you think about it like that sending a mix session is pretty pointless regardless of the IP convo.
@@progressionspod in my opinion even printing processed stems should be out. IMO you get the multitracks and the usual deliverables and that’s it. But I rarely win that battle. Especially if atmos is being done by someone else, and I’ve done half their work for em’. Haha
i need AI mixing asap
I think AI Mixing only exist due to it being more proccessed productions instead of actual live bands that track music