top of page

Quantize and clean it

  • Writer: davidsmith208
    davidsmith208
  • Mar 10
  • 2 min read

Automatic scores with Ai

Even in Garage Band


AI is already beginning to do this, but it is not yet perfectly reliable for complex parts like the synth solo in Tom Sawyer by Rush. However, the technology is advancing quickly.


Let’s look at where things stand.





1. What AI can already do (2026)



AI systems can already perform automatic music transcription — turning audio into notes or MIDI.


Some tools that musicians use now include:


  • AnthemScore – converts recordings into sheet music

  • Melodyne – very accurate pitch detection

  • ScoreCloud – generates notation from audio or MIDI

  • Ableton Live – can convert audio to MIDI



These work quite well for single-line melodies.


But they struggle when:


  • instruments overlap

  • heavy synthesizer effects are used

  • timing is complex (like Rush).






2. Why the Tom Sawyer synth solo is difficult



The synth solo played by Geddy Lee has several transcription challenges:


  • pitch bends

  • analog synth filter movement

  • fast runs

  • layered effects



AI must separate pitch from timbre and modulation, which is still difficult.


A human transcriber can still do it more accurately.





3. Where AI will likely be in a few years



Within 3–5 years, AI will probably be able to:


  • upload a song

  • isolate any instrument

  • generate accurate MIDI and full score



including:


  • articulation

  • pitch bends

  • modulation



This is called source separation + symbolic transcription.





4. Something you can already do with your setup



Since you mentioned using Arturia keyboards with GarageBand, there is already a practical workaround.


If you load the song into GarageBand and run:


  • audio → MIDI conversion

  • then quantize and clean it



you can often get 80–90% of the notes of a synth solo.


Then you refine it manually.





5. A fascinating historical irony



Back in 1981, when Tom Sawyer was recorded, musicians sometimes spent days transcribing solos by ear.


Now AI can do a rough version in seconds.


Even Neil Peart once joked that progressive rock required listeners to “do homework.”


AI is starting to do the homework.




✅ If you want, I can also show you something very interesting:

The synth patch used in Tom Sawyer (the Oberheim sound) and how you can recreate it on your Arturia controller in GarageBand. It’s one of the most famous synthesizer sounds ever.

 
 
 

Recent Posts

See All
Useful Ai time locations

Yes — that is exactly the recording. And I think you’ve stumbled onto an exceptionally good way for us to explore Alain. It is Jehan Alain: Complete Organ Works, performed by Jean-Baptiste Robin, rele

 
 
 
Alain pieces

Yes. Given what you are looking for—quiet, contemplative, mystical Alain rather than the spectacular Litanies/Trois Danses, and pieces that you could actually learn on a pipe organ or adapt to your sy

 
 
 
Jehan Alain

Yes — I found the program you mean, and I think the name got slightly garbled in your question. It is Jehan Alain, not “John Olan.” And Pipedreams program #2632, “Alain On Alain (Part 2),” aired Augus

 
 
 

Comments


© 2023 by Nick Martinez. Proudly created with Wix.com

© Copyright
bottom of page