How to compose a bass line with your voice

The idea came from here

After that post I thought to a possible audio-to midi converter…

Scaler was not an option because it recognizes chords, not single notes
The next option was EZ Bass and it worked!

The workflow is very easy

Put a Toontrack Audio Sender in the mic track so EZ Bass can listen you

Sing something, and you’ll see the audio recorded live in EZ Bass

At the end, select “Bass from bass” (maybe that “Bass from keyboard” works as well, but I didn’t test it)

Here is what Toontrack Audio Sender acquired

And here is the MIDI bass.mid (975 Bytes)

I had to edit the first and last chords that weren’t good, then I selected a cool groove and applied to the chords

The next step was dragging the EZ Bass midi directly into the EZ Keys GUI

I don’t know why the name of chords is different, but the EZ Keys tune seemed matching the EZ Bass one, so I applied a Prog groove to the chords, and eventually added a matching drum

Again, I don’t know why a Prog keys produced a jazzy tune… :thinking: :crazy_face:

I named it Voice-Bass Rulez because all started from the voice & bass

@ClaudioPorcellana , you said that so casually that Scaler does not recognize single notes in audio, that I double checked your assertion, since I could swear I had Scaler detect single notes before. But I think that must have been from MIDI source files, not from audio input. But I could indeed verify that Scaler does not accept single notes from audio input. I recorded a few single notes as sine curves (as pure of an audio signal, usually easiest to recognize), but Scaler did not “learn” any notes from it. Is that established fact? I wonder why this is, and has it been recognized as a point of improvement?

@Bernd
I think to remember that somebody, @jjfagot maybe, asked that already, and he was said Scaler recognizes chord only, and not all chords

For example the audio detection doesn’t work for arpeggios and certain kind of strums

On the other hand, I used the EZ Bass AI with old projects of mine where I lost the bass MIDI, and it recognized perfectly (sometimes less perfectly) the bassline in an MP3 containing other instruments

I suspect no, because the audio-to-midi conversion is very tough, so needing an advanced AI, and Scaler is for chords

I guess I am still baffled why single notes would be harder to recognize than chords? But I can also see how the actual sound type, even with single notes, can be complex, resonances and such. Especially with synth music, where single notes themselves can be evolving chords etc.
I suppose to mitigate the issues some Scaler users have with this, it would be helpful if Scaler returned some form of informational message that it couldn’t recognize anything instead of silently ignoring the input.

Yeah, clearly not the main focus of Scaler anyway, for now, the audio detection. That reminds me, now that I purchased deCoda, I should actually spend some quality time with it. Perhaps some learnings there can feedback into Scaler features/experiences. If us Scaler users propose the features, then the dev team doesn’t get into hot waters for mimicking other tools’ features.

1 Like

You are both right.
It is clear that the main function of Scaler is not to detect audio. With the midi it works well, although sometimes the chords it recognizes are not exactly what you hear, and you have to tweak it with a bit of musical dictation exercises.
For audio detection there are programs like Decoda that Bernd says that work better, but we are always spending money, hehe. In Ableton itself, that audio detection is often lousy. Of course, when we want it to detect the audio to convert it into midi, the first thing we have to do is try to remove all the effects that we can (reverbs, etc.). I was blown away by Claudio’s EzBass detection. I guess the EzBass AI directly removes sounds from a certain range (no treble in the bass) and simplifies the task

1 Like

not in my case because I had EZ Bass already
I can say that if you want a bass and buy it, you have the other feature as well

Aside from EZbass apparently prefiltering the input to focus only on lower bass notes, aren’t basslines mostly monophonic too? That should make it easier to unwind the actual underlying notes.
Also, coming from an AI background, it is fairly easy to establish common patterns and then infer the most likely source by similarity, without having to recognize exactly. Most humans only approximate in their perception too, so AI does not need to be perfect, just enough to fool human perception :wink:

1 Like

and I’m very happy to be fooled, as far as some nice bassline jumps out with no effort
:rofl:

I have to say, I never much concerned myself with basslines, in fact, was kinda oblivious that they even existed in music. But then, I never really understood the underlying techniques of music & related theory anyway, which got me into playing with tools like Scaler. And it wasn’t until Scaler had basslines incorporated that I ever played with that musical concept. And some of the EZbass stuff you shared here also made me learn more about this important topic. For the longest time I confused perucssion, drums, and basslines,lumping them all together under “rhythm”. Anyway, still looking which of the EZ tools I should get, and I couldn’t quite tell the difference between their different instruments. But from what I heard, EZbass is a useful tool, especially with the bassline recognition feature you describe.

:thinking: :flushed: :rofl:
I suspect it is because the music you started from
If you started from The Beatles like me, I think you had quickly got the difference

well, I would have never acquired it just for that feature
EZ Bass, as many if not all their plugins, is an easy way to have amazing basslines

Ok Boomer, LOL. I always wanted to say this to someone, and I am not even a Milennial :wink:
Anyway, there is absolutely no excuse for me not having developed earlier awareness. I do even remember the chopped up sample lyrics from “my kind of music” I listened to as a youth… “B-B-B-BASS!!!”, “Gimme a BASS Sucka!”, if not to forget Dre’s Chronic (808 anyone?) - whoop, whoop…

1 Like

I use Melodyne to convert Audio to MIDI. Sometimes I when I have difficulty getting some timing the way I want it or if working out the notes is a bit hard I’ll “da da” the bit into a microphone. Pull the audio into Melodyne and export the MIDI.

It works a treat. However you need at least the Assistant version of Melodyne. Melodyne Essential does not do Audio - MIDI conversion.

Just a thought

Tony

1 Like

Hi @teunis and what do you do then with that MIDI?
:thinking:

How I use the midi depends on what I’m doing. I usually send it to some VSTi be something like a guitar sim or maybe and orchestral part e.g. strings or brass. After all it is the midi.

One could simply convert to midi to get a melody then send it to Scaler to wrap some chords around it to get a basis for a new song that is in your head. (Not that I do this)

It really depends on what is needed.

Tony

1 Like

@teunis
Yes, I know…
But I asked because I think you’ve missed the reason why I did this post…
That is “how to compose a bass line with your voice”
:flushed: :grinning: