Well Kept Secret

[Jimbo 88]

I have around 50-100 songs I wrote (using the basis of underscores I composed for TV Docs) that I did almost 30 years ago.  I always used words to help create my melodies, would write the lyrics in my down time  and  thought they might be useful to have in a Library someday.   I never completed production of these songs because getting a Pro singer to do them was too $$$ and too involved...and I was not skilled enough as a mixer. I had my daughter sing a lot of scratch tracks when she was a teen and the ones I did produce where a disaster.  

But now things have changed!  I've learned how to open up Counsel View of CbB!  And I now have an arsenal of AI singers to throw at projects... And I can use AI images to create videos that compensate for my lack of editing skills.

The video is ruff at this point.  I will replace all images with 4K versions.  My computer gets bogged down trying to edit hi res images.

 

**for some reason it appears the only way to watch this is to click on the  "Watch On YouTube"  button on the bottom.

 

Let me know what you think.  Should I keep the notation score on the bottom?

 

Edited June 26, 2023 by Jimbo 88

[treesha]

I enjoyed your work, to me it seems great and I have no idea how you used AI to do what you did! I think the notation looks good on the bottom, I don,t read music but just as an artistic feature to me it looks good moving along. The song itself is very good.

[Jimbo 88]

11 hours ago, treesha said:

I enjoyed your work, to me it seems great and I have no idea how you used AI to do what you did! I think the notation looks good on the bottom, I don,t read music but just as an artistic feature to me it looks good moving along. The song itself is very good.

Thanks treesha!

So the short story for how I use AI to create vocals... I use a program called Synthesizer V, I purchased that and 3 different voices that where programed to work in it.  Synth V works much like a DAW and similar to Cakewalk's Piano Roll View.  You export your vocal as a .wav file and import it into Cake.  After that it's no different than any vocal track you or I might have recorded (except it's clean of noises and in-tune!  :) ). 

The long story...Synthesizer V is a great program, but composing in it can be painful.  You draw your notes/melody in Piano Roll View then type in lyrics. Which is all fine and good until you start tweaking and experimenting.  Then you have to wait for new audio every stinking time to be generated and becomes time consuming.  The easiest, best way is use a notation program to compose, export a midi file,  import that into Synth V.  Then 80%  of your work is done and you need to adjust pronunciations etc.    It's kinda funny 'cause I speak with a Chicago accent and try to get the singer to say the words the way I hear them.  They seem to have more of a British accent so like the word "Dance" sounds very prim and proper.  I have to spell that word "Daaance" to get my Chicago-ism.   Even if you don't read music, you should be able to create your melody in Cake's Piano Roll View, then look at it in Staff View and input your Lyrics.

Here is where the rub comes.  Cakewalk will export lyrics with midi, but the words don't line up  correctly all the time. So I end up exporting a XLM file out of Cake and into Sibelius (a notation program and the midi/lyrics work well).  I tweak a little more, just things that I know will work better in Synth V.  Then export midi out and into Synth V.

If you are going to do multiple tracks, say with background vocals, I would highly recommend that you use a notation app to start the process.   

[Glenn Stanton]

+1 -- yeah, the Synth V is very cool. i took one of my songs, imported the "choir lead" track MIDI, then added the lyrics (took some adjustments but not too bad as Jim noted some tweaks), then cloned and re-arranged a few parts for better harmonies across 4 voices. one cool thing (depending) is it constantly re-compiles as you're working, for me this is transparent, but on a slower machine would not be fun. then import as you would any other vocal tracks. as a note, before you start - setup the project with tempo, key, time signature, etc. one handy thing is the ability to import an audio track so you can position everything to the rough mix.

i think some of the voices are still a bit more synthetic than i would like but it's likely i don't have enough time in to make it better yet (only did 1 hour so far 🙂 )

Edited June 27, 2023 by Glenn Stanton

[Bajan Blue]

Really interesting - these are some of the best AI voices I have heard

They really are getting better and better

What program did you use for the video?

Cool stuff

Nigel

 

[daryl1968]

AI........scary . Great song and video. 

The future is gonne be weird

 

[Jimbo 88]

1 hour ago, Bajan Blue said:

What program did you use for the video?

I purchased a year's subscription (really 12,000 credits) from Kaiber.ai to generate the images.      I actually have a well paying project that needs images (I'm teaching middle school kids how to score to picture) and it more than justifies the $120.  So I'm going to take advantage of that and use the subscription for projects like this also.  The video editing program is a lite version of Vegas Pro I purchased many years ago.  I imagine that AI will have drastically changed by the time (one year) my credits for Kaiber.ai run out.

 

1 hour ago, daryl1968 said:

AI........scary . Great song and video. 

The future is gonne be weird

 

Thanks for the compliment.  The future IS gonna be weird for those who don't jump in.  I've learned that the hard way a couple of times over.  I saw it in the 80s when I midi allowed me to grab lots of work from the old guys in the biz.  I saw it again in 2000s when computers, social media and the internet took work away from me when I did not embrace the new technology.  And so I'm not going to fall behind this time.  You shouldn't either.  Jump in and hand on! 

Edited June 28, 2023 by Jimbo 88

[daryl1968]

3 minutes ago, Jimbo 88 said:

Jump in and hand on! 

I have a cheaper alternative  - the lovely current MrsDaryl1968 has a great voice and Melodyne to fix my wailing