Cantamus - Real vocals from notation!

[Brad Kaenel]

As a choir director, I'm always thankful when the music publisher provides a recorded performance of the vocal music I purchase for my singers.  But I'm ecstatic when they provide practice recordings of the individual SATB parts (which, unfortunately, isn't all that often) because I won't have to create them -- spending hours recording myself plunking out notes on a piano.  For decades, I've made "practice tapes" this way, all the while hoping for some future magical tech that would one day enable me to produce usable recordings in an automated way, directly from the printed music.

 

Well, as you've guessed, we're not quite there yet, technologically -- but this year I found Cantamus (https://cantamus.app), and it has indeed produced some amazing recordings, and saved me dozens of hours.

 

Cantamus consumes only MUSICXML, so you still have the prerequisite of creating a file in Finale, Dorico, MuseScore, or any notation tool that can export MUSICXML.  OCR-like music "scanning" tools have improved greatly in their ability to produce notation from printed music, and these can still be significant time-savers if your source material is simple and laid out well -- but often I find that hunting for, and fixing-up, all of the scanner's miscues requires more time than just transcribing the printed music manually, into my notation software.  And if your music publisher prefers the paper-saving device of "divisi" scores (Soprano/Alto and Tenor/Bass, intertwined on single staves, with up/down note flags to distinguish the two parts), you will have to do this anyway, since Cantamus requires a score where each voice is presented on its own staff.

 

But once you have notation sorted -- and not just the notes, but the lyrics as well -- then magic does indeed happen.  Feed that MUSICXML file to Cantamus, and after a couple of minutes of processing it will produce a mixable, multi-channel recording of your vocal arrangement being sung by human-sounding voices, singing actual words.  Read that again; actual words -- not just vocalized "Oohs and Ahs" like you can get by hooking-up any MIDI file to a "voice" instrument in your DAW.

 

Not only is this kind of mind-blowing for me, it's also incredibly useful because I can then mix and download an audio recording for each of my SATB vocal sections, for practice.  Maybe they prefer just their part, in isolation, with nothing else; or perhaps they'd like their part quite loud, but with the other part(s) softly in the background.  With piano accompaniment; or without.  And listening to an actual voice singing is SO MUCH better than a plunked-out piano line.

 

As a demonstration, I've attached three files:

  A scanned PDF of the original printed music

  A printed PDF of the score, as I transcribed it into Dorico and exported as MUSICXML

  An MP3 recording of the default mix produced by Cantamus, from that MUSICXML

 

I'm just beginning to use Cantamus, but I think it is an incredibly cool tool, and I'll be happy to respond to any questions you have about my initial experiments and experiences with it.  The software is subscription-based, not freeware.  You cannot accomplish everything I've described here with a subscription-less "demo" account, but you can try out simple scenarios without paying.

 

Cheers,

Brad Kaenel

 

 

 

 

Hurry Now to Bethlehem.PDF Hurry Now to Bethlehem - TBAS.PDF

[analogika]

Add a General MIDI playback to those vocals, and it's finished Christmas Muzak. 

 

[Brad Kaenel]

Ha! I supposed you could, yes.  Or load the vocal tracks back into your DAW with a click-track, and then layer on your real guitar/bass/horns accompaniment.  The vocals are almost good enough that you could actually use them as "sweeteners" for the live performance (if there were some way to influence the pronunciation of those words that aren't verbalized quite correctly.)

[Anderton]

Voice synthesis software has come a long way. I use Realitone's Blue to complement an actual human singer, it's a great combination.

 

I wrote a blog post about using a Microsoft voice from their browser, loading it into Studio One, and pitching it to create a vocal. It's pretty wild    This link should take you directly to the audio example, check it out. I think it's a lot of fun.

[Notes_Norton]

I knew a girl who was in a choir, and could sight-sing music. Give her the pitch of the first note, and she could sight-sing the piece. She didn't have absolute (perfect) pitch, but endlessly worked on her relative pitch and knew notation.

 

When I was much younger, I played in a band who also played pipe organ in a Catholic Cathedral in Palm Beach, Florida.

 

He could look at a melody line, even a hymn with no chord notation, and play an accompaniment. He would play the SATB parts for the choir during practice.

 

I play sax, flute, wind synth, guitar, bass, drums, keyboard synth, and I sing, to various degrees of competency. The hardest one of all to learn was voice. I sang lead for years in bands, before I got to the point where I thought I was decent. My excuse is that I wasn't blessed with a good 'instrument'.

 

We live in an era with great musical tools. Cantamus is incredible from what I just learned. Thanks for sharing. My old organ playing buddy wouldn't have to show up for as many choir rehearsals if they had this tool.

 

My only fear is when the tools get much better, they might take my job away from me.

 

Insights and incites by Notes ♫

 

[Anderton]

6 hours ago, Notes_Norton said:

My only fear is when the tools get much better, they might take my job away from me.

 

They very well might.  

[Notes_Norton]

5 hours ago, Anderton said:

 

They very well might.  

I'm not ready to retire yet

[Brad Kaenel]

My daughter is a choir vocalist with nearly a photographic (audiographic?) memory for music -- she hears something once, even when she's not focusing on it, and it's in her head forever; songs, soundtracks, advertising jingles.  She can sing anything, even harmony parts, *if* she can hear it.  But she doesn't read music, so learning things from printed notation is still a struggle.

 

For me, as a keyboardist, 55 years of reading music has created the connection from eye to finger -- but it's still a spatial relationship (that note means press that key), not the aural relationship that I kind of wish I had (that note sounds like this).  I can sight-sing with decent competency, but only because I can recognize the intervals in the notation, not because I know what the pitches should sound like in my head.

 

I've always wondered how that works for trained vocalists -- is there a spatial or physical connection from eye to voice (that note feels like this), or is it aural (that note sounds like this)?

[Notes_Norton]

20 hours ago, Brad Kaenel said:

For me, as a keyboardist, 55 years of reading music has created the connection from eye to finger

Same for me on saxophone.

 

Like most of us, I don't think of the note names, and unless it's very complicated, I just play the rhythms, too. I don't understand how, it just happens.

 

I can't do that on the guitar, but guitar is my 7th instrument, I started late in life, and I really haven't applied myself to reading on the guitar. There are too many demands on my time for that.

 

My bro-in-law has absolute pitch, which has helped his career out for decades. Now that he is getting older, though, he says things are sounding sharp, especially in the lower pitches. He can hear a C, and it now sounds like a C#.  For someone who doesn't have perfect pitch, I can't imagine that.

 

We all have separate 'tools' that we bring to the task of playing music. There are dozens or more of these tools, and not many have all of them. So we use the ones we have, and work on the ones we don't have yet.

 

Insights and incites by Notes ♫