Great video! The biggest difference was in the stereo image. Like you I have found the stereo balance is always to the left. I use a combination of the RX10 azimuth correction, Gullfoss and Ozone, and I think I arrived at a similar conclusion. But I am now investigating the additional plugins you used. The part that cannot be fixed is the mushy percussion, where kick drums sound like porridge and snares sound like paper bags. But very early days!
Thank you for the good, practical video. 👍 (Was the German title a coincidence? 😎) I can only completely agree with the comments on udio. 👍 For me, udio is a tool like using loops from sound packs ("loop pusher" 🤣) Actually, the process of remixing and remastering/mastering is identical for me. In addition to plugins from A1Audio and Focusrite, I use the entire iZotope range. The tip for Perception is very good (unfortunately a bit expensive 😭)
Thanks, I need to look at the A1 Audio stuff. I have translations for German, Spanish and Hindi now. I decided to provide translations for title and description as well.
@@michaelgwagner Focusrite's FastBalancer is also a very good tool (there is also a simple free version). I'm currently in the process of setting unknown German poets to music using Udio (Christmas music at the moment). This brings new honor to this density 🙂 and it's fun (title on my YT channel)
Hello Michael. You have considerable talents as a sound engineer, why not devote your time to working with human musicians and composers? Perhaps you might consider that rather than creating anything you have polluted human culture. AI is stealing from human composers and musicians and diminishes the contribution music makes to our culture.
Hi there, first of all, thanks for the talent comment. I don't think I would subscribe to that but I certainly appreciate the sentiment. This is a great question and I will most likely do a separate video about this. Short answer is that I am first and foremost a technology educator in a field that is adjacent to music production. And that means a large potion of what I do is figuring out how the production workflows will be like 10-15 years in the future so that we can educate students properly. There is a much longer explanation that deals with common misconception about how this transformers actually work.
@@michaelgwagner You seem to be holding the misconception that the way in which the AI produces an output is relevant to any discussion about how it diminishes human culture. You also seem to lack any understanding about the actual processes of musical composition and performance. Perhaps it is time to consider the social consequences of this technology rather than continuing as if it exists in a vacuum. Your continued assertion that this technology is allowing people to create music is insulting to musicians and composers and a dangerous misconception to spread.
Where exactly do you draw the line? The AI that drives the Sonible EQ, the AI that enables stem separation for remixing, the AI behind Ozone or LANDR, the AI behind Logic's Session Drummer,...? Just wondering.
@@michaelgwagner All these AI you mentioned are not generative, they are an aid to musicians and producers (and those are welcome). AI should not create anything per se, especially if the pre-training/fine-tuning data is copied (temporarily or not) without a licence. Intellectual Property stands on the framework that the intellect is human since it grants an incentive to keep creating and get the economy going. Otherwise you would see copyrighted works or patents assigned to animals or AI entities (as opposed to humans/human companies).
@micheler.3439 Fair point. We do not know what Udio is trained on. Some of the main investors have access to very large libraries of their own. The licensing issue is one that will be resolved. I think it is necessary to looked at generative AI under that assumption.
Wow, that's an easy $1000 worth of plugins trying to clean up that mess. These sort of challenges are so interesting, precisely because you are having to do quite extreme processing with sub-optimal source material. I think you did a pretty good job with just a stereo track. If I had the budget for your setup and was trying to obtain the very best output , I would run the stereo track through an AI splitter first, to obtain, bass, drums, vocals and music busses. I would then sort out the weak transients on the drums, and try and get some additional processing on the individual busses, dependent on their frequency ranges and content. You could then recombine in a mix thru a similar mastering chain to get (hopefully) more optimum results. Taking it a step further, re- sequencing and replacing/augmenting parts would then be another route to achieving a better output at the mixing stage.
Thanks! I played around with stem separation. The results are mixed to be honest. The best I found is working with something like RipX DAW which operates on the principle of stem separation but goes a bit beyond that. I did that in a couple of cases where I felt it would be good to change a few notes here and there. Since they introduced audio in-painting, I primarily deal with that through in-painting though.
nice to see you using Decibel!
Yes, that’s a good one.
great video. and the track was actually gorgeous
Thanks! 😊
Great video! The biggest difference was in the stereo image. Like you I have found the stereo balance is always to the left. I use a combination of the RX10 azimuth correction, Gullfoss and Ozone, and I think I arrived at a similar conclusion. But I am now investigating the additional plugins you used.
The part that cannot be fixed is the mushy percussion, where kick drums sound like porridge and snares sound like paper bags. But very early days!
Thanks! Need to look at RX for that purpose. Just picked up version 11.
Thank you for the good, practical video. 👍 (Was the German title a coincidence? 😎)
I can only completely agree with the comments on udio. 👍 For me, udio is a tool like using loops from sound packs ("loop pusher" 🤣)
Actually, the process of remixing and remastering/mastering is identical for me. In addition to plugins from A1Audio and Focusrite, I use the entire iZotope range. The tip for Perception is very good (unfortunately a bit expensive 😭)
Thanks, I need to look at the A1 Audio stuff. I have translations for German, Spanish and Hindi now. I decided to provide translations for title and description as well.
@@michaelgwagner Focusrite's FastBalancer is also a very good tool (there is also a simple free version). I'm currently in the process of setting unknown German poets to music using Udio (Christmas music at the moment). This brings new honor to this density 🙂 and it's fun (title on my YT channel)
You can polish a turd all day, but at the end of the day, you just end up with a shinny turd.
I’m old enough to remember that people said the exact same thing about the first digital cameras. 😉
Hello Michael. You have considerable talents as a sound engineer, why not devote your time to working with human musicians and composers? Perhaps you might consider that rather than creating anything you have polluted human culture. AI is stealing from human composers and musicians and diminishes the contribution music makes to our culture.
Hi there, first of all, thanks for the talent comment. I don't think I would subscribe to that but I certainly appreciate the sentiment. This is a great question and I will most likely do a separate video about this. Short answer is that I am first and foremost a technology educator in a field that is adjacent to music production. And that means a large potion of what I do is figuring out how the production workflows will be like 10-15 years in the future so that we can educate students properly. There is a much longer explanation that deals with common misconception about how this transformers actually work.
@@michaelgwagner You seem to be holding the misconception that the way in which the AI produces an output is relevant to any discussion about how it diminishes human culture. You also seem to lack any understanding about the actual processes of musical composition and performance.
Perhaps it is time to consider the social consequences of this technology rather than continuing as if it exists in a vacuum.
Your continued assertion that this technology is allowing people to create music is insulting to musicians and composers and a dangerous misconception to spread.
Where exactly do you draw the line? The AI that drives the Sonible EQ, the AI that enables stem separation for remixing, the AI behind Ozone or LANDR, the AI behind Logic's Session Drummer,...? Just wondering.
@@michaelgwagner All these AI you mentioned are not generative, they are an aid to musicians and producers (and those are welcome). AI should not create anything per se, especially if the pre-training/fine-tuning data is copied (temporarily or not) without a licence. Intellectual Property stands on the framework that the intellect is human since it grants an incentive to keep creating and get the economy going. Otherwise you would see copyrighted works or patents assigned to animals or AI entities (as opposed to humans/human companies).
@micheler.3439 Fair point. We do not know what Udio is trained on. Some of the main investors have access to very large libraries of their own. The licensing issue is one that will be resolved. I think it is necessary to looked at generative AI under that assumption.
Wow, that's an easy $1000 worth of plugins trying to clean up that mess. These sort of challenges are so interesting, precisely because you are having to do quite extreme processing with sub-optimal source material. I think you did a pretty good job with just a stereo track. If I had the budget for your setup and was trying to obtain the very best output , I would run the stereo track through an AI splitter first, to obtain, bass, drums, vocals and music busses. I would then sort out the weak transients on the drums, and try and get some additional processing on the individual busses, dependent on their frequency ranges and content. You could then recombine in a mix thru a similar mastering chain to get (hopefully) more optimum results. Taking it a step further, re- sequencing and replacing/augmenting parts would then be another route to achieving a better output at the mixing stage.
Thanks! I played around with stem separation. The results are mixed to be honest. The best I found is working with something like RipX DAW which operates on the principle of stem separation but goes a bit beyond that. I did that in a couple of cases where I felt it would be good to change a few notes here and there. Since they introduced audio in-painting, I primarily deal with that through in-painting though.