Another new song composed with the assistance of AI.

I'll post the details tomorrow when I get to the studio.

Pass it on

Thanks for listening.

Howdy Herb ...

Wow ... this is just so good! You managed to really put some emotion into the AI voice. I was impressed with the higher notes the AI vocal rendered. It didn't get thin or strained. And the slight bit of vibrato on the voice was quite effective. I'm totally impressed with this!

The band did a super job. What caught my ear was the mandolin. For me, it was an unexpected band member. But gosh dang, it really added a nice flavor to everything else. I guess you could say the band was "delicious"! I'll buy that.

Everything was amazing: the band, the arrangement, the mix, the vocal, the lyric, the message. But most of all, the genuine enjoyment of a song sounding like it should with no apologies needed.

Amazing!

Alan


BTW... thanks for the front row and center seat! Send me the bill!

Alan

It's a brave new world, wow. It's really hard to tell this isn't a human. It all sounds nice but the best thing in your song was the live guitar riff LOL. I'm not sure AI will replace human emotion anytime soon but that was quite impressive. As an aside, what program did you use to generate the AI vocal?

Bleeding edge stuff to be sure.

Best of luck

Rick

Herb,

Extremely impressive! This particular voice sounds very natural and has some nice nuances. Great track selection - can't wait to see where this all goes!

Bob

Al david: Your comments made my day when I opened this site this morning. THanks

Rick: It is indeed a brave new world. THere's currently no live guitar in this but that might change because I'm not able to find something in the real tracks close to this tempo that has the tone I really want. I'll likely learn the solo note for note or perhaps create my own.... but with a tone I can live with. This one is just too brittle sounding on the top end. The voice is Synth V Pro with Solaria AI database.

Bob: Yes.. I agree. These vocal synths have come a long way in the past year.

Here's the updated info.

The lyric idea was input into Lyriclab and the result was heavily edited and revised.

This is, I think, the third complete recording. I started with other styles and didn't like them. I deleted everything and started each one again from square one using only the SGU as a starting point on each. Including taking it into Cake and deleting that too. Never be afraid to start over. Lots of edits to the structure and chords as it progressed through the various stages.

Style is MANPMGA which is pop country ballad.

Tracks currently include:

Drums: Modern pop groove
Bass: 1827
Mando: 1947
Guitars: 1596, 4021, 4022, 4023, 4026, 3505, 3348
Piano: 1839

Things are still in a condition of flux. As I mentioned, I'm not satisfied with the tone of the solo guitar quite yet. So I might do this live.

Synth V voice Solaria AI database. There's a lot to learn about this synth to get realistic sounding vocals. Dreamtonics is the company that makes this. If you want to play with this technology, you can download the Basic version and some of their demo voices for free. They tend to be synthy sounding but it gives you a feel for the synth. Obviously, it's not everyone's cuppa java.

Mixed in Cakewalk.


So... that hopefully answers some questions..... Now.... I have a few of my own questions.

At a songwriters meeting last night... this song was presented. One of the comments was "yeah, I could tell this was not a real singer from the autotune". Does anyone hear artifacts (obvious or subtle) that would indicate autotune or something else? If so, point out exactly where you hear it please. I might be too close to this to hear it.
Also, I'd appreciate any observations on production nits or advice on the mix/mastering. Levels.... EQ.....Reverb......instruments..... I have pulled the mando down on a few errant notes that were in a place of relative silence that kinda stuck out at me....

Anyway... thanks for your time and comments.

From experience: you should never say beforehand that it's not a real singer, because a) everyone will focus on the singing and not the song, and b) everyone will say it's obvious.
If they don't know, they might say that the song sucks but the vocals are great.
And ask who the singer is.
Been there, done that.

The Autotune™ remark could be about the first 2 words of your song, "we all" does indeed sound pitch-corrected. The rest of the song is fine.

Production-wise it's good, I wouldn't change that. Overall it's a good song and I like it!

The vocals sound a bit throaty, did you change the "gender" settings to make the voice a bit darker?

good advice and yep.... gender a bit to the huskier side.... not too much....

I'll have a closer look at the first two words....

Thanks

Yeah the 1st two syllables stick out as does the seemingly very precise diction.
Most of the larger intervals sound a little suss. It has a classical singer doing country feel to it.
The vocal is strident - some EQ to soften it would help it sit a little.
Yeah the solo tone is not a keeper in my book either. It something I struggle with in BIAB...that & that I really don't seem to like Brent Mason's stuff - and there's so much of him.

Guitarhacker,

First off, the song itself is stellar, really cool feel and vibe to it. The music bed is terrific and overall arrangement is excellent. Had I not known it was AI generated vocal . . . I don’t think I would have noticed at first . . . the bridge is where it doesn’t quite sound right to me . . . and on some of the high notes throughout . . but not enough to make me think it is an artificial vocal by that itself.

My bigger focus is that the vocal doesn’t quite fit with the feel of the tune . . . kind of what rayc said above. So, a good suggestion from him to perhaps look to soften in spots and bring down some of the high end.

I also agree with ray regarding the Brent Mason RTs in general . . . they rarely seem to work for me. But that’s a different issue . . .

Good stuff, and impressive use of AI for this.

All my best to you,

Deej

Interested to find out if any of the people at the meeting commented on the A.I.assisted lyrics?

Great song. I think that the A.I voice has got great potential. You are at the cutting edge indeed, Herb.

I would also be interested in learning how long this song took from start to finish and if it is getting any easier.

Herb,

A lot to like about this.
The BAND is solid. Good RT choices and well mixed.
I Like Brent's tone (I listened to the song before reading any comments - and thought "nice to have a bit of grit to the solo").

The vocal is surprisingly good.
It does have a quality to it that pegs it as "generated".
I think it has more to do with the fact that those generated vocals lack some of the dynamics of a "real" singer (breathes and volume nuances). They have a sort of "sameness" no matter where they are in the natural range of a singer. I'd guess in time they will get better at that. (less sameness).
That is not necessarily a bad thing. A lot of Pop sounds like that these days.

As ray said, there is a lot of high EQ to the vocal. Rolling that off (considerably) would help with much of the impression of "generated" - that "sound" is accentuated in the highs. (And, Country vocals these days have MUCH less highs to them)

All of the AI generated lyrics I've hear (so far) have a "generic" feel to them (big time).
THAT really seems to work for this - a We-Are-The-World or Buy-The-World-A-Coke kind of thing.

Good work.

Ray: yep... so I did a regen on those 2 words. Better I think... I'll post the revised track in a few days

Deej56: Still working on it.... revamping the tracks and muting things to see what works..... and levels....

JC: No... the lyrics were not called into question. I'm loving that app BTW. It's a great idea starter. Total time start to finish.... hummmm... I'll let you know when I get it done.

FJ: That high end is puzzling me.... I might have to go back to the synth and play with the settings to see what can be done there. Right now, I'm burned out a bit from listening and trying to figure things out, so I am going to take some time and come back fresh in a day or so.... I'm still playing with the band to get something that I like and sounds good.... I'm drifting into a more acoustic band to start and bring things in later in the song. my next version will be up sometime next week and you'll see what I'm thinking.

I certainly appreciate all the comments and advice. Taking it to heart and seeing how it applies and working towards the final result. Stay tuned. This is a work in progress. And I'm not finished.

Herb... I am still amazed at what AI can do...this is such a great song...melodically, lyrically and with a positive message...it's so great to hear a song that's not about losing love or a cheating lover, etc...I love the track...well constructed and the modulation was so appropriate to lift even higher an up-lifting song...superior effort indeed!!

Uplifting rock ballad with nice melody, + for modulation. The band sounds smooth and the mix is very balanced. The guitars and the drums sound good and the mandolin is nice addition.
Great work with Solaria, because I couldn't recognize her. Very natural and powerful sounding, although I would like to hear her sounding calmer and softer in verses.

Janne

Headed in a different direction.... what do you think? Yes or no? heavier or the slightly more acoustic version?
I'm actually looking for a more country flavor than a rock/pop version. Hence the changes

Pass It On ( final version)

Herb,
Big sound ballad. Sounds very... Eurovision Not in a bad way, just that is how I hear it. Very cool production. Arrangement and mix is very strong. Vocal synth tech seem to progress significantly. Good effort and thank you for sharing.

P.S. "Pass it on (acoustic based)" seems to be broken.

Misha.

Firesong & Janne

Thank you for listening and commenting. I've changed it up a bit.... I wanted a more country slanted version.

Again thanks for listening

Thanks... Try the link again. Might have been swapping the file in box.

I prefer the country version to the first one. Softer, warmer - and the lyrics register clearer, here. But a great song with a great message - well done!

Good production good use of real tracks and I like the positive message in the lyrics.

Musician 17: Thanks... I like the latest version myself.... Aside from minor production nit fixes, I think this is the one.

Elliot: Thanks. I appreciate you spending the time to listen and comment.

First of all, I really like this whole thread where everyone is pitching in (pi). A lot to learn hear. Personally I like the gender swap of the singer. She does sound a bit too perfect it pitch. But I only have questions about the synth voices because I have no knowledge. Can you have them off-pitch a bit? Can you make them glide into the next note and waver a bit in vibrato? Still, it's such a good song, and not even a classical singer could make it sound bad.

Had to listen again for the production stuff. To my ears the mix is fine. I'm sure I can hear each piece of the production and I enjoy how you separate them. I really like the mando cutting through all those guitars. The volume of the lead could be raised to at least the volume of the vocal, but I think the tone is fine. Perhaps a bit cleaner tone would be better for the song. The song. Yeah, I really like it. One tiny nit, but there is the last stick count-in just before the song begins. I'd either drop it or include the whole count-in. Hey, I really enjoy the lyrics. Fun stuff here, Herb.

Hello Herb
A really good song
Great melody and sound
But the vocals are extremely good
Never heard such a good AI singing before
I am really impressed
Many greetings
Andi ;-)

Very interesting experiments in technology Herb.

I look forward to seeing where this takes you.

Wow! This sounds pretty realistic to me. Amazing! Sounds pretty close to Freddy Mercury to me.

Mike B.

Babu: This voice has the singer singing slightly off pitch by a variable amount... well, not perfectly on. And, you can edit it to be on or off as you want. I only fixed the obvious stuff. If you listen in the bridge, you can hear she's not "on" the entire time. The last version is edited to no intro.

Andi: I was impressed too which is one of the reasons I hired this singer. Chances are good she'll be hired again.

David: Me too.

Mike: I didn't catch a Freddy Mercury vibe but hey... FM was pretty darn good so I'll take that as a compliment.



Just a heads UP.... this is the latest and final version


Pass It On ( Final)

I cut the intro out because, well, it wasn't actually serving a useful purpose. Why beat around the bush... times a wasting.... lets hear the song. Waaa Laaa, it's done, I'm moving on to the next project.

Thank you all who commented with advice and observations.

This is REALLY good!!! What was the role of the AI?? I listened to this and the other.....vocals are great in both......I like the first one better but it could be a bit less bright or something. VERY COOL!!! Take care. Greg

AI was used to sketch out the lyrics and the voice is a synth that uses AI.

I had to edit the lyrics to get what I needed without the cliches and lazy rhymes it currently likes using. The vocal synth required a good amount of time and effort to finesse the singer.
And of course band in a box is a bunch of black box voodoo magic.

Hope that helps

Thanks for listening.

All three versions are impressive and each got better. Her vocals sound less harsh in the second and third versions and the harmonies sound good too. Actually her voice reminded us of Sonia Isaac of the Isaac Family.

For us her voice in the higher register sounds more unnatural.

The band on all sounds great.

Uplifting message too.

Well done! And no doubt a helluva lot of thought and effort.

J&B

I really like the final version.

Cool arrangement. I wasn't expecting a slide guitar!

Yes, the vibrato in the singer made it feel real, along with some slide ups to some notes.

Good melody that has me singing along. Band sounds good! Majestic ending.

thanks.... it is and has been a learning curve. Trying to figure out the various parameters and getting them right, because, there's currently not a lot of info in the manual on the details in English. A fair bit of lets punch this button and see what it does, going on. Part of what I'm learning is the vocal range that sounds the best, and what doesn't because it's either too low or too high. I actually transposed and reworked the melody lower in a few places. Just like us, she strains to hit those higher notes and while she will hit the pitch, the formants are not sufficient to make it sound good.

Considering that there were several versions created and deleted previous to any of these.... yeah, there was a lot of work involved and thinking as well as a lot of second guessing.....


I had the opportunity to see and meet the Isaacs a number of years ago when they played a concert at our church. Amazing vocalists, all.

I figured I'd throw in a dobro for the break. Thanks for all your help.



Vibrato is built in but controllable. As are the legato between the notes. I'm glad you listened and liked it ...Thanks