Menu
What's new
By:
Filters

Saros - new voice from eclipsed sounds

There are two versions available for download - one for the current version of SynthV, and one for the beta. If you try to load the beta version into the current editor, you'll get a warning that it needs the beta.

So if you don't want to upgrade, just grab the older version of Saros.
 
They are here, woohoo!

Since I wanted to try out the Spanish language mode as well as the ARA integration, I used the beta version of SynthV. Oh boy, that caused some serious issues and data loss in the end :(
So yeah, currently there is no Spanish support for the audio to pitch feature, so it produced gibberish, but the notes it created were mostly usable. I just ended up correcting those where needed and putting in all the lyrics by hand. I only had it create notes, not pitch. The pitch is by Saros. Especially for English parts, I totally see myself singing them in in the future instead of writing everything by hand 😎
However, the ARA plugin itself hosted in Studio One 6.5.1 (and possibly also the standalone version? I don't know) had serious issues. I wanted to use the European Spanish mode, and it would always switch itself off again, reverting to Latin American Spanish. Then, sometimes selecting a note would select all notes instead until clicking a couple of times. Using undo with ARA plugins is also a dangerous game, since at some point the plugin view must have lost focus and instead of undoing an operation in SynthV, it undid adding the plugin. At least I was able to redo and the data was still there. But then finally, when I was still in the middle of tuning the vocals, I closed the plugin for a second to adjust the mix. When I reopened it, all parameter automation I had drawn before was gone 😭

So about Saros, I really like their voice. They are very expressive and the vocal modes are useful. The Spanish works well. I'm not a native speaker, only learned a bit of Spanish for fun. But even so, I perceived the pronunciation as not quite correct at times and struggling with fast syllables. There was one case where they refused to pronounce a "rr" phoneme and would always say "l" instead.

So here's my first attempt. This is as far as I got before the parameters were wiped and I didn't feel like redoing it. It's Electricidad by VALANTAIN.

View attachment electricidad-saros-acapella.mp3
 
Last edited:
FlamencoJazz said:
That is fantastic! Flamenco is one of my main interests and I was not sure that any of the voicebanks could produce an authentic flamenco-singing voice. Is this from a recording, or did you sing this yourself?
Click to expand...
It was a quick test using the first male flamenco clean acapella that I found on YouTube:


I used the resounding & full vocal modes combined. For greater realism and timbral contrast, would be better to apply soft vocal mode in the lower parts.

In Saros Spanish sound "j" is phoneme /x/ which is aspirated and equivalent (or very similar) to English /hh/. Perfect for the southern Spanish accent, but does not have the raspy "j" sound used in the rest of Spain.

The conversion of words to phonemes has some shortcomings, for example in words with the grapheme "z" you have to replace the /s/ with /C/ or edit the notes to include it.
 
cloudrunner92 said:
So yeah, currently there is no Spanish support for the audio to pitch feature, so it produced gibberish, but the notes it created were mostly usable. I just ended up correcting those where needed and putting in all the lyrics by hand.
Click to expand...
What language are you using in the settings? English? Do you have a spanish dictionary installed? I’m just wondering how to make synthv pronounce how I want, I’m trying to write Finnish, so japanese is very close in pronounciation, but it can’t say the letter u like finns pronounce it (like in the word you), and always pronounces it like japanese do, as ü, but if I switch to English, no English word I know has the letter ü in it (ew is somewhat close but not the same or the j sound in you) and in English all the words are filled with unnecessary h-sounds
 
Last edited:
AceAudioHQ said:
What language are you using in the settings? English? Do you have a spanish dictionary installed? I’m just wondering how to make synthv pronounce how I want, I’m trying to write Finnish, so japanese is very close in pronounciation, but it can’t say the letter u like finns pronounce it (like in the word you), and always pronounces it like japanese do, as ü, but if I switch to English, no English word I know has the letter ü in it (ew is somewhat close but not the same or the j sound in you) and in English all the words are filled with unnecessary h-sounds
Click to expand...
Speaking of Finnish, have you solved “r”? We have it as alveoral trill and haven’t figured out how it is done in SynthV.
It’d be nice to make SynthV singing in Finnish. There’d be some foreign accent most likely, but it wouldn’t bother me and no one else but another Finn would notice anyway.
 
AceAudioHQ said:
What language are you using in the settings? English? Do you have a spanish dictionary installed? I’m just wondering how to make synthv pronounce how I want, I’m trying to write Finnish, so japanese is very close in pronounciation, but it can’t say the letter u like finns pronounce it (like in the word you), and always pronounces it like japanese do, as ü, but if I switch to English, no English word I know has the letter ü in it (ew is somewhat close but not the same or the j sound in you) and in English all the words are filled with unnecessary h-sounds
Click to expand...
Oh, I guess there's a bit of a misunderstanding. I used the official Spanish mode that is now available in beta and only with Saros for synthesizing the vocals. Only the lyrics transcription from audio feature does currently not support Spanish. It's not possible to access all recorded phonemes from all languages at the same time, I think, but I haven't tried using unsupported languages yet.
 
cloudrunner92 said:
Oh, I guess there's a bit of a misunderstanding. I used the official Spanish mode that is now available in beta and only with Saros
Click to expand...
Oh ok, I didn't notice that as I had to remove the beta as it complained constantly about a debugger. I read on the synthv forum that you can now use different language setting per word, before I had to make another track If a song was bilingual, so maybe I'll just mix and match english and japanese to get what I want
 
Blancanegra said:
Just a quick test converting audio to notes, very cool!

Flamenco - Audio to MIDI - Saros
View attachment Calabozo test.mp3
Click to expand...
Cóoooojones! Wow! Thanks for sharing, I NEED this voice. As soon as I noticed Dreamtonics had something in Spanish I looked for your name and.... convinced 1 minute later!

EDITED: Is this made with just the pitch copied from a YT song + lyrics entered manually (plus some usual phonetic corrections) plus playing a little bit with S-V modes and parameters? And the audio you are posting is THIS? Wow!
 
Last edited:
Anyway... I bought it!

At the moment I have a problem. I tried to installed the "beta" voice with Spanish capabilities (saros-installer-v101b1.svpk) after updating Synth-V to the latest version. (Synthesizer V Studio Pro 1.10.1).

But when I double click the voice bank file it says this:

1701977497199.png

Do I need to find and install a new beta version directly from Dreamtonics server to install this beta voice? I never used a beta before and from comments in this thread I assumed that installing the last official version should be enough.


--> Too excited to wait. Google took me here:

index - powered by h5ai v0.29.2 (https://larsjung.de/h5ai/)

index - powered by h5ai v0.29.2 (https://larsjung.de/h5ai/)
resource.dreamtonics.com resource.dreamtonics.com

And then I could install saros beta version v101b1. Just in case someone is in the same situation.
 
Last edited:
parapentep70 said:
Cóoooojones! Wow! Thanks for sharing, I NEED this voice. As soon as I noticed Dreamtonics had something in Spanish I looked for your name and.... convinced 1 minute later!

EDITED: Is this made with just the pitch copied from a YT song + lyrics entered manually (plus some usual phonetic corrections) plus playing a little bit with S-V modes and parameters? And the audio you are posting is THIS? Wow!
Click to expand...
Yes, was done using audio to MIDI, typing lyrics and manually adding some breath between phrases, with Resonant vocal mode applied. No parameters editing.

The better and clean the input, the better the conversion. Phonetic output was English by approximation, then I chose Spanish as languague and typed the proper lyrics.
 
I feel a bit lost... Solaria impressed me so much that I also bought the bundle with Solaria. But it's difficult with male voices. I like Asterian a lot but it's very special. I don't know Kevin... Is Saros possibly the male voice of the same quality as Solaria?
 
You must log in or register to reply here.
Top Bottom