PG Music Home Forums General Discussion User Showcase Time

Hi, floyd jane.
You certainly know how to make a songwriter feel good about their song!

Now that's my dream team... Although Karla Bonoff would be nice, too.

I mentioned in my notes that I had higher hopes for it, but my brain wasn't putting out anything particularly distinctive. So there are a some pleasant "gestures" between the parts of the string section, but I really need to spend some more time working on getting my individual string parts to feel more cohesive. I'm happy with how it came out, but I listen to arrangements done by George Martin for The Beatles or America, and think That's how it's done!.

That part was played using my EWI. It's the same sort of wind controller that Mario uses.

As an aside - breath controlled instruments (using a wind controller or breath controller) are a fantastic development. I can play all sorts of instruments - strings, brass, double reeds - that I'd never be able to play in real life with a degree of control that is staggering. Even if you don't have a breath controller, you can still control the instruments via the mod wheel and play then strictly with a keyboard. But as a singer, I find the wind controller very natural.

I originally played a Sample Modeling cornet on the part. I've got a very schmaltzy playing style, with lots of breath controlled vibrato. I thought it sounded good and provided a good contrast, but my wife (and Deej) thought it was too out of character with the rest of the song, so I changed it to a Sample Modeling violin. I was able to use the same MIDI data, and just swapped out the instruments. I did make a few minor tweaks to the melody to better lead into the vocal entry on the second half.

I sent a link to my old choir director, who asked if it was synthetic, or and actual singer. I figure that if a singer knows it's probably synthetic but still has to ask, I must be doing something right.

One more tangent. Right now, there are still a limited number of singers available, and it's fairly easy to recognize specific voices.

There's a technology called RVC (retrieval voice cloning) will make it possible to train an neural network to replace an existing voice with another voice. The performance remains, but the timbre of the voice is changed. It's expensive to record all the material needed to create a new voice, but it's fairly cheap to train a neural network using RVC. So I expect in the near future there will be a wider variety of voices available using this sort of technology.

Thanks again, I really appreciate your kind praise!