Just tried a V1 project that was about ready to roll, but I held back to see if V2 would improve the sound before starting to mix the song.
I tried both Full export/import and a MIDI export/import and both times the result was frankly horrible.
Solaria was transformed from a sweet, softly spoken pop vocalist to a warbling karaoke pub singer! What happened to the promised improved realism? To my ears is sounds like Dreamtonics has rolled back the years to the ugly Vocaloid offerings!
Maybe V2 will be prove to be worth it once the numerous bugs get ironed out and I try building a project with Solaria II from scratch, but as it stands Iām pleased to have an extra voice to use (Thank you for Liam, Dreamtonics) but Iām back to V1 for now.
Maybe this rant will prove to be misplaced and I hope Iām wrong and will grow to love V2, but my initial impressions are of a huge disappointment, and an opportunity missed.
I was sure this was exaggerated so I took an SVP from SVS1 and put it side by side with it in SVS2. Solaria really does sound incredibly warbly and weird. Changing the Expression slider from the centre all the way to Rigid helped a little, but there are some very clear pronunciation differences too.
The .uw phoneme has turned into the .uh phoneme, for example, so āmoonā turns into āmun.ā
This was my exact same experience. Solaria was my favorite voice because when you put the notes and words in, it was 85% of the way there. Bringing her V1 track into V2 was pretty disappointing, because it wasnāt just different - it was actually kind of bad. Then I brought over a Saros track, and the vibrato was too fast, so I looked for a way to slow down the vibrato - easy in V1, and as far as I can tell, impossible in V2. V2 does have a higher fidelity sound, but removing the manual vibrato settings probably made it useless for me. Iām hopeful that Dreamtonics will eventually fix these things, and then I can use it. Iām fine with waiting until things are better, because V1 is very, very good.
Same here, back to V1. Between this and the fact that they removed the global vibrato control, I prefer to continue working with V1. I hope itās a bug and they fix it soon.
Solaria II has great flexibility and thereās ample control over vibrato, pitch, timbre, etc on individual notes as well as much better control over phonemes.
Hereās my quick and dirty first test with the voice.
I havenāt tried Solaria yet, but the other V2 voices have been generally better than V1 for me. A couple of weird pronunciations but I didnāt find any of that when making a new song.
I suspect V1 voices will not work great with SV2. Iāve had few issues with the new voices. After I saw that compatibility for SV1 voices was limited, I decided not even bother.
Control over vibrato is better? Please tell me how to slow the vibrato down or speed it up. I donāt see any way to do that, other than redrawing the lines by hand.
Same here.
I am using Yuma with 130% gentle, 70% airy. ~0.2 breathy voice, and -0.155 gender. Itās to make a cool tomboy-ish sound I like.
But in V2, it sounds too sweet often, and way way too mumbling when singing in English.
Thatās the way to do it - but remember you donāt need to draw the complete sine way just a slope here and there and the line is automatically filled in for you. Itās much quicker and produces a better result - try it.
Youāre talking about the Pitch Deviation tool, right? I could see where, with some practice, I might be able to get pretty good at drawing in the type of vibrato I want for an individual word. But hereās my point. Some vocal databases like Saros and Liam have fast vibratos. Too fast for slower styles of music. In V1, you could select all of the notes and slow down the vibrato speed with one slider. You could also control how long the singer would sing a note before the vibrato kicked in with one value. Thatās all gone. So in a track with Saros, to fix their ātoo fastā vibrato, Iām expected to go into each note and hand-draw a new vibrato. What used to take seconds could now take an hour and depend on my fine motor skills (which are not great - my hands are kind of shaky). Or they could just re-add the manual settings and fix this. Like others, I am grateful I can still use V1. And thank you for the sample - I do appreciate you trying to help me, honestly. I think that they didnāt consider vibrato speed. Not everyone likes a fast vibrato. I hope they fix it.
Itās the fact that the manual tuning panel is missing, for me. How the heck am I supposed to cover songs that are meant to have that keropitch type sound now XD SV2 is looking like a downgrade.
I was about to write about manual tuning missing in V2, but updated to 2.0.1 just now and apparently pitch control curves mode now works again mostly like manual tuning used to be in V1. Patch notes said that it was bugged if the curve extended over multiple notes in initial release, though it felt like there was some larger issue since when I first tested, it wasnāt very effective esp. in rap mode. But nonetheless seems to work properly now in that regard.
But still gotta test Solaria, first day I spent with Mai 2 which at least sounds pretty good.
Solaria II sounds like she came out of retirement after 10 years, having taken up smoking and hitting the bottle a little too hard
Seriously, a common gripe over at the Cubase forum is that early adopters are treated like unpaid beta testers. Given the limited resources of Dreamtonics, in hindsight it might have been better to give those of us that upgraded six weeks ago the opportunity to download the beta version āat your own riskā as a soft launch and gather feedback before the full launch.
Synth V2 has way too many gaps and problems to be classed as a full product. Hereās hoping the Dreamtonics staff can wade through the tsunami of complaints and put a workable V 2.5 together for us.
SV1 files are processed very differently on SV2 than on SV1, and so when working with old files Iāve found it critical to clear all the old tuning off of it and starting from scratch, especially pitch deviation on the bottom panel. If you hit Ctrl+A and the delete key, you can clear any pitch deviation and any other automation that may exist on the track pretty quickly.
Make sure you save it as a different file name if you want to retain your SV1 tuning for use in SV1. Once you do that, Solaria and any other bank youāre using in SV2 is going to sound a lot better in it.
Iāve been playing around with Yuma a lot, and the sweetness is most likely coming from the gentle VM. Vocal modes are a lot more potent in SV2 than in SV1, and Iāve noticed Gentle having that sweetness effect, while Airy is similar in softness without being as sweet. Try 150% airy, 55% gentle, and tweaking gender + breathy as needed.
Cross-language synthesis has degraded a lot in this version, though you might benefit from the phoneme timing and voicing panels. Thereās a bit more control over pronunciation on Version 2 voicebanks (such as the vocal modes having a pronunciation slider so you can get, say, the pronunciation of Solid without the sound/tone of Solid, as well as a mouth opening parameter) and upgrades are on sale for the next month or so (you can get Yuma 2 for $19 at the moment,) but it might also be something thatās sorted out in later releases as SV2 is in a very beta-like state at the moment. Whether thatās something worth trying now or if youād rather wait for updates is up to you and your use case.
I did my own comparison matching the setup from SV1 to SV2 before the crash of SV2. Hereās the youtube of what I shared. Iām back to SV1 waiting on bug fixes.
The second passage is awful, but also the first sounds too unnatural. This was indeed way better in V1, drawing the slider slightly to the left and getting a good result.
This is very funny. Iām still laughing writing this reply.
These two demos were just to show that itās possible to change the vibrato rate. They are not finished, beautiful renders thatās why I called them āvery dirty.ā
But within V1 there was no need to differentiate between āvery dirtyā and āfinishedā, cause in most cases the vibrato was ok for most of the song after adjusting one slider.
