I took the first few lines from the new demo of Saros, and compared it to the same song with the other male English voices. I added no effect or reverb to see what each sounds like on the same song in its raw form, each with its on AI generated in Sing Mode for that singer.
Maybe I’m asking stupidly, but which new demo, because I didn’t find one, but I probably wasn’t careful and didn’t search patiently. And is SVP available since the demo? Or did you create him and replace Saros with the mentioned AI singers? Otherwise, I put all the voices under myself in Studio One. Each voice has a different dynamic and different phrasing. The singing start time is also different. Asterian starts first and Hayden last. I like all of them, everyone has something in them, but the song has to be adapted to each of them, and the arrangement of any accompaniment must also be adjusted. But that’s how it works with live musicians and singers as well. Otherwise, the post was interesting for me…
I took the audio from this new video posted by Eclipsed Sounds demonstrating Saros.
They did not supply an SVP file so I just punched in the MIDI notes and lyrics for the same song, as a rough test. They say the SVP file will be included with the purchase of Saros, so someone who buys it will probably post it on a forum and we can try out the others singers on the exact file. But in the meantime this gives us an idea of the different abilities of each singer compared to Saros.
I already pre-purchased Saros so I’m waiting and thanks for your reply
That falsetto Saros could also be used on “Move Over” by Janis Joplin. Probably without competition yet. Like you, I sometimes download something from the MuseScore site and I tried it on Ninezer’s - “Move Over” and I quite liked it, but I wouldn’t put it out (publicly), because it would take a lot of time. I prefer to play with my own creations. I also tried by Enio Moriccone - “Once Upon A Time In The West” from the MuseScore sheet music and I didn’t do very well here, and Synhesizer V Pro also failed me so far.
For pop/rock/soul/blues/country you cannot get a realistic performance in Synth V by using the MIDI notes from Musescore, because that style of singing has a lot of slides and embellishments. You could START with Musescore MIDI but it would take many hours to tweak it and add all the embellishments.
All I did for the sample here (and also for the recent demos of Hayden) is to take an isolated vocal of the original recording, and use audio-to-midi conversion to get the exact notes and slides and embellishments. Then I attached the words and it is very quick to get this. That is very useful if you want to make a copy of an existing song, but not if you want to make your own songs. But doing this helps you to see how professional singers embellish vocals, so it will help you to program in the long run.
There are lots of programs that convert audio to midi (like Melodyne or Ace Studio) but if you wait for the next update, as I understand it, you will be able to do this inside Synthesizer V.
But downloading MIDI files from Musescore (or elsewhere) is great for making choirs, because you can directly use the notes as is.
Thank you for your great answer, which helped me to sort out my knowledge, where sometimes I have an idea of what it is like and sometimes I feel lost. I go to this forum because I don’t know anyone in the Czech Republic who deals with this and with whom to talk about it. I learn very useful information here. I skimmed the Ace Studio site for parts that I didn’t know. It’s interesting, but I’ll stick with Synthesizer V Pro. I have Melodyne and know it the way I need to use it. So far, it seems to me the best in converting audio to midi. I’m having a problem in midi with my voice because I use a bass and with normal vocal cords I get at least D2 and then I switch to singing with fake vocal cords which I’ve been honing since 1986 (I’m an amateur - not a real good singer) so for the pitch transition between the two it is difficult to recognize the types of vocal cords and even my singing teacher admitted this to me. The Melodyne algorithm choses me tones from false vocal chords higher than or if I use the polyphonic algorithm, it throws them into several notes above each other (as well as the RipX Deep Audio application, which also converts Audio to midi). That’s why I sing my ideas an octave higher, or write notes with the help of a musical instrument. I’d rather stop now, I don’t want to bore you, but if anyone is interested in singing with fake vocals, Eric Hollaway, who lent the voice to Asterian, can do it much better than me. And I’m learning it at the age of 73 from Asterian, mainly intonation in the octave C1-B1…
The Audio-to-midi in ACE Studio is better for vocals than either Melodyne or RipX, for some reason. I think it is because it is designed specifically for singing, so it has some kind of AI to recognize the notes better. You can try ACE Studio free for 14 days and see. But anyway, Dreamtonics has announced that soon they will also have audio-to-midi inside Synthesizer V and I expect that will work really well. There is no need to sing higher or lower than you are able, because you can sing everything in a comfortable key and then transpose the notes later in Synthesizer V or in the midi file before importing it to Synth V.
Thanks for the good informative information.