You know, even with the proliferation of AI and the explosion of ideas disseminated across the World Wide Web, I feel confident I'm the first to use the blog title you see above. I have written a little before about Cantai (here, here, and here), a developing software tool which "sings" notated music from directly within a notation program like Dorico or Sibelius. There have been significant improvements in the past months which have made it a very useful tool, seemingly all of a sudden. I can now produce practice audio for my church and school choirs which allow singers to hear more closely what musical lines should sound like sung. I have learned that we underestimate how challenging it can be for less experienced singers to hear a part played on piano, with its sharp attack/decay feature, and imagine how that should by sung.
To be honest, I'm also as alarmed as many about the intrusion of the artificial into human spaces (an intrusion which pre-dates the AI boom by decades, but which is perhaps accelerating), and so it gives me some comfort that Cantai, at least for now, is able to produce something useful as a means towards an end - but not really an end product that interests me. It's silly and pointless to declare that AI-plus won't eventually cross the uncanny valley to a point where the results are sublimely satisfying, but I don't think it's there yet. (I do think many people would be astonished at how much music they hear which is already generated primarily via synthetic means.)
The development of Cantai has also been hanging over me with a strange sense of obligation because I have quite a few blog projects (some almost old enough to drink) in which I relied on laughably bad virtual singers. So I realize most of those deserve a Cantai redo.
My post from a few days ago about mechanical musing inspired me to create a playlist of musical "compositions" I've created partly as a way of musing about something I'd encountered. One that came up is a quirky little duet I wrote in 2018 for two singers whose fragmented melodies end up overlapping to create a fully formed melody, with their words also flipping meaning due to the interleaving. I called it "Hocket Love," because a hocket is a musical device where two voices alternate in quick succession to create a combined melody. It's a strange project, but a pretty decent solution to a challenging puzzle. You may read all about the original here, though the singers make me cringe enough that I'll not post their efforts here.
So, as Part I of the Cantai-ization of my synthetic vocal musings, I present a new version of "Hocket Love." This is a good starting point for this new project because: 1) the song is quite short, and 2) the not-quite-human voices actually work well with the text/technique. You be the judge.
No comments:
Post a Comment