I am totally not crying over this incredible update.
She sounds and looks fantastic. This is exactly what she and her creators deserved.
She sounds and looks fantastic. This is exactly what she and her creators deserved.
SynthV Kasane Teto for Synthesizer V AI
- Thread starter PearlStarLight5
- Start date
I don't know if I just missed it maybe, but I don't think anyone directly linked her Amazon JP physical preorder link. Here it is!
I have already partaken
I have already partaken
OSTER Project on the demo song:
Google Translate:
Google Translate:
I guess my linking of Teto Territory on April 1st was a bit premature, or very appropriate lol. This is a great example of one of my favorite things about the vocal synth community in general: people make things and create something bigger than all of them, purely by force of creativity. Miku, as a character, was basically memed into existence, and so was Teto.I was in charge of the official demo song "April Star" for Synthesizer V AI Teto, which was released on 4/27
This is the story of how a false diva born from April Fool's Day becomes a real diva
this is SO exciting and she looks and sounds so good!! I read on twitter people who own AHS products get a discount on her and I have miki hmm....but I havent bought voice synths in awhile cause i always just use the same two and i hardly ever use jpn vocals. i'm so torn
The discount is only on the physical version if you order from AHS, not the download.
At least if you have SV Pro she can also sing in English and Chinese. Haven't heard her crosslang yet, so I dunno how good it is, but...
You can hear it a bit in her mellow demo. It sounds similar to the crosslang of other JP native banks, which means accented, but leagues more understandable than her English CVVC ever was by default.
oooh interesting, i havent been keeping up with synth v news so i didnt know about this feature! sadly i do not have sv proAt least if you have SV Pro she can also sing in English and Chinese. Haven't heard her crosslang yet, so I dunno how good it is, but...
but this is good to know
I like the fact the UTAU and SynthV AI banks can be complimentary to each other
I messed around with her English CVVC, and honestly, I don't hear much of a difference as far as comprehensibility goes unless used by someone who doesn't know how to optimize CVVC. That said, I'm gonna hold off on further direct comparisons between the two until we get a proper English demo; it doesn't feel right declaring one superior to the other just yet when we've heard her AI sing exactly five words in English.
The english crosslingual is something I'm most curious about myself, so I get the excitement, but I share the sentiment that this isn't enough to go off of to make any fair judgements, especially when the original doesn't struggle with the two lines either:
This is pretty nice, and those demos are impressive so far. I'll definitely be looking into Teto's SynthV voice bank, especially if we can hear more of the english crosslingual.
I think my use of the word "better" was a mistake. I apologize for that. There are many reasons an end user could prefer the CVVC over a SynthV XLS system, but from my perspective, as someone that used her CVVC voicebank extensively, the reason for this would be "personal preference" and "I like the sound" for the same kind of reasons I still prefer listening to concatenative synthesis in most of my music. It's an instrument that can be played.
Let me clarify my intended reading. Her english sounds like (from the little we've heard) it will be comparable, if not basically the same, to the other Japanese VBs on this engine cross-synthesized into English. I feel confident saying that because I have not heard yet a Japanese Synthesizer V bank that differs from the group in this area yet. Even though we only heard 5 words-- she sounds a whole lot like the TOKYO6 characters in those 5 words.
I've discussed this on these forums before, but I think the reason AI synthesis has made such a splash in the anglosphere is the fact that AI synthesis eliminates a lot of the compromises we made with concatenative synthesis and reclists like CVVC that do not exist to the same extent in Japanese.
The XLS has not been perfect, but it has done wonders on wider perspectives toward English synthesis in these contexts. In my opinion, XLS has been more successful at synthesizing English than traditional concatenative synthesis (not restricted to UTAU), even if I have a personal taste for listening to voices with these reclists for what they are (instruments). "Better" in a linguistic sense, not necessarily what will make the best art for you personally. CVVC will have its unique parts that XLS simply cannot have because everything is rendered out of the same cross-language model, which is why it's so consistent (but also samey).
Her CVVC is full of charm but to use it effectively, you need to know a lot about UTAU configuration. I don't think it's as much "optimization" as knowing how to mix her voicebanks together outside of the English CVVC VB. Which can be learned, but... I've never heard her English described in any context other than being a nightmare to work with, by the people I know that are loyal fans of that VB.
This is why I feel confident saying, "it will be better than CVVC English" to a person that is wondering whether or not they will be able to use it to sing in English, because so far, every XLS JPN->ENG has had better luck synthesizing english, as a language, than Teto CVVC. I am sorry. This comes from a place of love X(
But yes, nothing is a guarantee so we still don't know and I spoke too quickly. We'll have to wait and see.
We are all gonna make cool-ass things. With UTAU Teto and SV Teto! Together! Fusion!
Let me clarify my intended reading. Her english sounds like (from the little we've heard) it will be comparable, if not basically the same, to the other Japanese VBs on this engine cross-synthesized into English. I feel confident saying that because I have not heard yet a Japanese Synthesizer V bank that differs from the group in this area yet. Even though we only heard 5 words-- she sounds a whole lot like the TOKYO6 characters in those 5 words.
I've discussed this on these forums before, but I think the reason AI synthesis has made such a splash in the anglosphere is the fact that AI synthesis eliminates a lot of the compromises we made with concatenative synthesis and reclists like CVVC that do not exist to the same extent in Japanese.
The XLS has not been perfect, but it has done wonders on wider perspectives toward English synthesis in these contexts. In my opinion, XLS has been more successful at synthesizing English than traditional concatenative synthesis (not restricted to UTAU), even if I have a personal taste for listening to voices with these reclists for what they are (instruments). "Better" in a linguistic sense, not necessarily what will make the best art for you personally. CVVC will have its unique parts that XLS simply cannot have because everything is rendered out of the same cross-language model, which is why it's so consistent (but also samey).
Her CVVC is full of charm but to use it effectively, you need to know a lot about UTAU configuration. I don't think it's as much "optimization" as knowing how to mix her voicebanks together outside of the English CVVC VB. Which can be learned, but... I've never heard her English described in any context other than being a nightmare to work with, by the people I know that are loyal fans of that VB.
This is why I feel confident saying, "it will be better than CVVC English" to a person that is wondering whether or not they will be able to use it to sing in English, because so far, every XLS JPN->ENG has had better luck synthesizing english, as a language, than Teto CVVC. I am sorry. This comes from a place of love X(
But yes, nothing is a guarantee so we still don't know and I spoke too quickly. We'll have to wait and see.
We are all gonna make cool-ass things. With UTAU Teto and SV Teto! Together! Fusion!
I apologize, but I'm not sure I'm fully following
What I'm getting from this is that she's going to be more streamlined to use, and if that's the case, I agree! But tonally, pronunciation-wise, etc., I don't wanna pass judgements yet until we hear a full sample with her, that's all. And if that's not what you're getting at, I apologize. Blame my cats. They decided to play Who Can Be the Loudest Fucker in Town last night so I haven't slept much, heh...
I'm making an assumption, based on what we've heard in the song, and based on the track record of all of the other Japanese VBs on the engine, that Kasane Teto SV will sound similar/have a similar range of effectiveness in English, because they are all very similar, and have effectively the same pitfalls.
English is a language that has benefitted greatly from this kind of synthesis specifically. So Kasane Teto SV would be an outlier if it is a lot different and struggles or succeeds in unique ways. Which, to be fair, is completely possible. But it would be surprising.
Tonally we definitely can't know how it will compare to Teto CVVC yet or how it will appeal to people that want the very specific sound from Teto CVVC instead of broadly being interested in a "Teto that can speak English". But pronunciation wise we can have a good idea from the rest of the products on this engine.
But like I said, waiting for a full sample, is still a good idea. You are right.
English is a language that has benefitted greatly from this kind of synthesis specifically. So Kasane Teto SV would be an outlier if it is a lot different and struggles or succeeds in unique ways. Which, to be fair, is completely possible. But it would be surprising.
Tonally we definitely can't know how it will compare to Teto CVVC yet or how it will appeal to people that want the very specific sound from Teto CVVC instead of broadly being interested in a "Teto that can speak English". But pronunciation wise we can have a good idea from the rest of the products on this engine.
But like I said, waiting for a full sample, is still a good idea. You are right.
Acapellas of teto ai were shown on the live stream along with a bigger sample o fher english cross lang, no chinese though. will edit in the time codes for when shes shown later.
https://www.youtube.com/AHSoftware
edit:
1:26:52-1:29:40
1:30:10-1:30:40
1:31:45-1:32:30
https://www.youtube.com/AHSoftware
edit:
1:26:52-1:29:40
1:30:10-1:30:40
1:31:45-1:32:30
Last edited:
cute little english demo, it sounds good but i feel like i had no idea it was teto singing until the "fwaa". still getting used to the synth v bank i s'pose!
Currently manifesting a longer English demo over a rock instrumental, c'mon Wrong + Right sequel from CircusP 
Ahhh it is the 27th in Japan!! Boxes haven't shipped yet (sad) but a buddy who got her digitally has sent some renders and she is gorgeous. She sounds so much like herself still! A very faithful port.
There are already a few covers on YT too! This one is quite nice.
There are already a few covers on YT too! This one is quite nice.