Risks of Automated IA-Assisted Mastering

Question: What risks does a musician incur by using low-cost IA-assisted automated online mastering services and apps?

Illustration

Advice from the Experts

  • Illustration

    Maor Appelbaum(Faith No More, Sepultura, Dream Theater ecc.)● Awarded with Platinum and Gold records● Grammy-nominated● Juno-winning Mastering Engineer for The Music Of The World - Sounds With Impact.

"The Human Factor"

When working with artificial intelligence algorithms, you're dealing with a tool that can be programmed by the user with a set of information.
For example, you can specify:
● Song style● Desired compression level● Type of stereo image● Final loudness level
The algorithm will process the provided information and apply it to the track, but it can never make decisions based on personal taste.
I'm not saying it can't work. Some things might turn out fine, but others may not be as unique and interesting.
On the other hand, when the work is done with a mastering engineer, you can receive feedback, if needed, on what to improve and enhance in the mix.
You also have the opportunity to provide directions on what you'd like to achieve through mastering, allowing the mastering engineer to use this information to give the song the most suitable sound profile.
So, the human factor is crucial in establishing a connection with the music, interpreting it, and, on an emotional level, guiding it through the most fitting expressive manner.

  • Illustration

    Luigi Barone (Gigi Barocco)(Bloody Beetroots, Steve Aoki, Merk & Kremont, Gabry Ponte, Marracash, Gianna Nannini, Fedez, Achille Lauro, Frah Quintale ecc.)● 9+ Gold Records● 10+ Platinum Records

"Lack of Sensitivity"

The issue it's not that there are risks.
The issue, in my opinion, with these automated low-cost services, especially in the professional realm (speaking of more prominent artists), is that the human element in the mastering process is missing.
By human element, I mean the sensitivity of the person doing the job.
This, you see, can never be replaced by anything else, and it's why an artist chooses one mastering technician over another. Because they have a sensitivity that leads to a unique sound and style.
These low-cost automatic services can't have this human component because they're essentially based on algorithms.

  • Illustration

    Sabino Cannone(David Kahne, Martin Gore, Paul McCartney, Tom Jones, Corrado Rustici, Jessie J, Pino Daniele, Tiromancino ecc)● 40+ Platinum Records

"An Accurate Analysis Is Needed"

The risks associated with using IA-assisted and semi-automatic mastering services stem from the lack of a thorough analysis of the mix characteristics.
In my opinion, in musical genres with more predictable sonic aspects, like certain types of electronic music, automatic services might surprisingly work quite well.
The effectiveness of such services depends on the predictability of the material.
The main risks include the possibility that critical aspects of the mix aren't identified as such and are instead processed according to a predefined standard.
In my view, a semi-automatic process might achieve only about 50% of the results that could be attained by a human expert. This is because a person who can identify the nuances and criticalities of the mix can develop a set of strategies to address them, maximizing the potential of the mastering.

  • Illustration

    Federico Pelle(Tommaso Paradiso, Michele Bravi, Noemi, Grazia Di Michele ecc)● Professor at the "A. Steffani" Conservatory in Castelfranco Veneto (Italy)

"Discriminating Ability"

The mastering process can't and shouldn't be the same in all cases and contexts. Mastering requires evaluative and critical skills, and these skills depend on the individual sensitivity of the mastering technician.
For instance, I've encountered projects in the past that sounded so good already that the best decision I could make was simply to increase the track levels in the mix.
I communicated this to the client, who candidly told me that I was the third mastering studio he had approached because the previous ones had intervened excessively, and he wasn't satisfied with a sound he no longer recognized.
When he asked me what I had done, I told him that all I did was raise the volume of the files because I felt everything was already perfect.
Mastering is a phase that requires evaluation, context, and the ability to discern between a) what could be done and b) what really needs to be done.
From my perspective, Artificial Intelligence isn't yet capable of making these kinds of decisions, and I'm not saying it will be impossible, but it is certainly going to be a challenging thing for AI and other automated services to match the quality of a human-directed music master.

  • Illustration

    Alex Trecarichi(Cassius, Kanye West, Francesco Gabbani, Max Pezzali, Fedez, Elodie, Ghemon ecc.)● Served as a lecturer at S.A.E Milan, I.E.D, and numerous other professional schools in Italy

"Low Care"

There aren't any particular risks in mastering with online automatic services, except for the possibility of releasing a track that isn't perfectly focused and professionally polished.
I'm confident that in a little while, machine learning will reach very high levels that will enable Artificial Intelligence to master most songs correctly.
Today, the technology isn't quite there yet, but the result obtained through an AI-based service can still be a good starting point for comparing your work and gaining another perspective.

Conclusions

The use of automated online mastering offers advantages such as affordable prices, ease of use and immediate results, but may have some limitations in terms of human communication, creativity and work customization.
Nevertheless, AI-based mastering may be the right choice for those who have a limited budget or do not need a professional mastering engineer, as it offers an affordable and super-fast alternative to achieve a good sound master.
On the other hand, the majority of the experts agree that the use of automatic mastering services based on AI does carry some risks. 
In fact, while there are professionals who believe that AI may work just fine in music genres that are more predictable, the majority believes that these services may lack sensitivity to specific needs and contexts as well as the ability to assess the nuances and criticalities in personalizing a music mix. 
Personally, I believe that AI can be an interesting assistant in the mastering process if used in synergy with the unique experience of a music pro. 
AI can be a source of inspiration and comparison for sound engineers, enabling them to improve their skills and expertise.
However, in my view, today the human element remains an important component to achieve truly original, high-quality results in professional audio mastering.

Pros and Cons

of using low-cost automated AI-based mastering services.



PROS
● SimplicityThe use of AI mastering services is generally intuitive and requires little or no technical know-how.
● AffordabilityAI-based mastering often offers inexpensive solutions, ideal for those on a tight budget.
 SpeedYou get almost instant results without having to schedule and wait for long mastering sessions.
● Subscription OptionsMany services offer subscription plans that allow unlimited mastering sessions at very affordable rates.
● AccessibilityAI Mastering services can be used from anywhere with an internet connection.
● Consistent resultsAI services offer consistent and predictable results, which can be useful in maintaining uniformity in your discography.
● Initial resultsAI-generated results can provide a starting point for subsequent improvements or comparisons with the assistance of a pro audio engineer.
 Wide range of stylesAvailability of a wide variety of presets to suit different musical genres.


CONS
● Lack of Human CommunicationWhen using automated services, there is no direct contact with a human mastering engineer. This means that you cannot communicate your wishes or make customized changes to the mastering process.
● Creative LimitationsAlthough AI can produce quality results, it will always lack the creative and artistic approach of a human professional. This may affect the ability to adapt the master to a specific musical style or track.
● Limited customizationAutomated mastering services generally work on stereo tracks and do not offer the possibility to intervene on specific instruments or parts of a track. This lack of detail limits how the work can be customized.
● Limited RevisionsAlthough some services offer revision options, these revisions are often limited, and it is not possible to check all parameters as one would in a traditional mastering setup assisted by a human engineer.
● Lack of Human EarSubtle musical nuances and decisions based on human listening may be lost as the AI relies on predefined algorithms.
● Not Suitable for All Music GenresAI may work better in some music genres than others, depending on the complexity of the mix and the structure of the song.However, in my view, today the human element remains an important component to achieve truly original, high-quality results in professional audio mastering.

Illustration

Discover resources, techniques and tools to record, mix and produce music effectively

Subscribe to "Music Producer" Alessandro Ciuffetti's monthly newsletter.
Receive every month in your inbox:
● Useful tools and instruments● Techniques and practical advice● Tips from the experts● Videos and tutorials● Examples, templates, checklists● Success stories