【Eleanor Forte】Binary Star - SawanoHiroyuki[nZk]:Uru【Synth V Cover】


Hi all! I’ve been playing around with SynthV and wanted to share. :grinning:

1 Like

Sounds like you put a lot of work into this!

I’ve got some comments - feel free to ignore if you weren’t looking for feedback! :smiley:

The vocal sounds flat in a lot of places, and if you look at the pitch line, it is. Did you apply AI singing to this? A little bit of “humanization” goes a long way, and I think in this case it might have gone a touch overbaked.

Might I suggest that you look at using alternate ways to pronounce some of the words?

For example, “as the pendulum swings” has a harsh /ch/ sound on pendulum. Replacing it with /p eh n - d y uh - l eh m/ would soften it a bit.

Putting small breaks between some words, such as “can’t believe” might also help.

For example, in the phrase “You know blood is thicker than water”, the emphasis is on the downbeat of “is”.

But because there’s no break between the words, it comes out as “you know blood diz thicker than water”, which isn’t what you want. Putting a break between blood and is would fix this.

There’s a build at 1:04, but the vocal sort of gets buried in the mix - especially at the start of the phrase at 1:16.

The phrase “decide you’re not a friend” at 1:51 is the sort of phase where you generally want to ignore how other people sing it, and adjust the melody so it sounds natural when Eleanor sings it.

At 2:40, the phrase “pushing us” is buried in the mix. Automating the gain in your DAW (or in SynthV) would help.

I can’t make out the lyrics at 2:44, but this seems another place where you probably want to ignore the written/sung notes and tweak it until it sounds right.

You can think of speech as sustained vowel sounds interrupted by consonants, and singing is doubly so. Held vowels give sung music their beauty, and consonants clarity. But Eleanor tends to over-articulate consonants, which gives clear speech, but at the expense of vowels.

The Note Properties panel has an option to adjust the relative duration of phonemes, and if you’ve got the Studio version, you have the option of choosing alternate (ALT1, ALT2) versions of phonemes instead of the default (DFLT). Note that adjusting the phoneme before the one you want changed will typically change the following phoneme as well.

At 2:44, it sounds like the consonants are taking most of the duration of the syllables and leaving no room for the vowels, so you might first try reducing the duration of the consonants and see what happens.

The vocal again is buried at 2:56.

At 3:22, the vocal starts off too soft. The volume of the vocal will automatically adjust as the pitch goes up, so for low notes you may need to go into the Parameters and dynamically adjust the Tension and Gender, or even the Loudness (although I find that parameter is generally less effective, because it sounds less natural). Perhaps adding a bit of Breathiness would help keep it soft?

You’ve got a really nice arc building, and at 3:27 it really shines.

The words “to sadness” sounds a bit off - a combination of being a bit too “straight” with the rhythm, as well as Eleanor making an… interesting vowel choice on “sadness”. Using alternate phonemes here can help, or just tweaking the sound with Tension and Gender.

Then again, if it sounds good to you, leave it as-is.

The phrase “we didn’t learn” is another example of what happens if you let Eleanor and the built-in dictionary sing. There are just too many consonants, and it sounds unnatural. Substituting “we din learn” and putting a small (say, 32nd note rest) between “din” and “learn” will give you the impression of a /t/. Or use /d ih n dx/ for a softer /t/ substitute.

At 3:33, in the phrase “For the broken hearted” has sounds glitchy on the /b/ of “broken”. Putting a short space before “broken” might solve this. The pitch change on “hearted” can probably be fixed by going to the Note Properties panel and adjusting the Duration Right of the first syllable and Duration Left of the second syllable to make the transition smoother.

Also, consider replacing the hard /t/ with a /d/ or /dx/ phoneme.

On sustained long vowels, like on “binary”, you might also consider adding an arc using Tension, Gender and even Breathiness. You’ve got it on the vibrato - why not pull out all the stops?

At 4:21, the vocal is again buried under the backing track. Contrast that with 4:24, where "binary" jumps out.

Getting Eleanor to sing *“binary” is a bit hard - the second syllable needs to sound more like “air”, and she doesn’t quite want to make that vowel sound. Adjusting the consonant duration can help a bit.

At 4:30, the lyric is again buried by the backing track.

Anyway, nice work, and hopefully something above will be helpful.

1 Like

Wow! Thank you for all the feedback!!! I really appreciate it. :smiley: :smiley:

I did apply Tsurumaki Maki’s English AI pitch tuning to it and then adjusted it from there.
I’m not very good at doing the pitch tuning from scratch. :grimacing:

I did notice a few of the issues you mentioned - I’ll definitely give your tips a go!
Thanks again for your feedback! :grinning:

Have the feeling of singing well, thank you for your creation ~

Thank you for listening!! :grinning:

1 Like

@dcuny I changed the pronunciation & tuning as you suggested and it does sound much better! :sweat_smile: Thanks again! :grinning: :grinning:

1 Like