I remembered that technically they're already at 7 years dev time, as Maghni was in development for 3 years before it was announced.
It took Vocaloid 1 4 years to be developed and released, and that was with a big company like Yamaha backing it. I don't think a long development time is necessarily bad, but communication with people (especially if they've already preordered some of your stuff) is very important, which I'm glad they're at least communicating at all.
I honestly think the issue of maghni AI is they probably tried to scope too hard, like wanting to support every language under the sun etc for a small team, that's quite the goal
Hey, everyone!
I really want to open by saying: thanks so much for all the constructive criticism! I think I'd be lying if I said we haven't had a few rough patches this last year, but I'm really optimistic to see so much we can still improve on - especially when it comes to communication and establishing trust.
This restructuring has come with a fair deal of challenges, but I'm really happy to report that they're mostly ironed out! Communication, however, is something that we've continuously dropped the ball on. Be that as it may, I'm doing everything I can to pick up the reigns a bit and provide updates as often as I can on our process, now that I've gone from a Developer -> Co-Director.
As things are now: I fully run our email support line, oversee the main project and all third-party project discussion/planning, work closely with and strategize with our overseas outreach team, and am also the owner and manager for Soundcake Studio, an upcoming third-party for the engine. Needless to say, if there's ever any questions you may have, please don't hesitate to reach out to our support line on the maghni.ai website. It will be my pleasure to assist in any way that I can!
If I may, I'd like to address some of the things mentioned the last week and try my best to provide some helpful insight in regards to the project as a whole! Starting with...
• How long HAS Maghni AI been in development?
While early concatenative plans and development stretch back to over half a decade, the actual AI engine version of Maghni as we know it today has only just started development in the last year or two! Hopefully this explains where we're at currently with development.
• Another delay? When can we expect Maghni to release?
We're a small team, but with our previous structuring, certain people in charge set goals that were... lofty, at best. As such, a lot of the previously promised demos or (what were planned to be) character reveals fell completely flat, despite the development team's best efforts. For that reason, it's difficult to give a firm timeframe beyond "Sometime in 2025". However, base model retrain 3.0 is right around the corner, we've finished implementing all the core languages we've planned to initially launch with (The number eludes me, but I believe it's around 12?), backer rewards are prepped to be shipped out in January (VocaTone still hasn't announced it yet, but the comic is still in development, as well as the digital album.), and we've enormously increased the number of voices and characters ready for development on the engine.
Fun fact: the languages, while plentiful, aren't the main cause of delay; actually, the work on those was mostly done months ago!
• Isn't there a better place all this can be going?
Actually...
We're currently implementing a blog section for our website as we speak; a proper, more organized place for me to share future updates for everyone interested. Since VocaTone isn't intrinsically tied to the project anymore (in fairness, they were never really developing the engine or vocals for it anyway), it seems I've been tasked with figuring out the best way to convey our progress to everyone. Hopefully this works, for the time being - but, when the blog is finished, I hope you'll read along as we further develop Maghni AI into the new year!
That's about it!
If there's any comments, questions, or concerns, I would be overjoyed to address them! Thanks again for being kind enough to offer some really important criticism. It seriously helps, and gives me a very strong sense of direction on what more I can be doing with my newfound position.
Looking forward to what we can show off in 2025! Also, Happy Holidays!
I really want to open by saying: thanks so much for all the constructive criticism! I think I'd be lying if I said we haven't had a few rough patches this last year, but I'm really optimistic to see so much we can still improve on - especially when it comes to communication and establishing trust.
This restructuring has come with a fair deal of challenges, but I'm really happy to report that they're mostly ironed out! Communication, however, is something that we've continuously dropped the ball on. Be that as it may, I'm doing everything I can to pick up the reigns a bit and provide updates as often as I can on our process, now that I've gone from a Developer -> Co-Director.
As things are now: I fully run our email support line, oversee the main project and all third-party project discussion/planning, work closely with and strategize with our overseas outreach team, and am also the owner and manager for Soundcake Studio, an upcoming third-party for the engine. Needless to say, if there's ever any questions you may have, please don't hesitate to reach out to our support line on the maghni.ai website. It will be my pleasure to assist in any way that I can!
If I may, I'd like to address some of the things mentioned the last week and try my best to provide some helpful insight in regards to the project as a whole! Starting with...
• How long HAS Maghni AI been in development?
While early concatenative plans and development stretch back to over half a decade, the actual AI engine version of Maghni as we know it today has only just started development in the last year or two! Hopefully this explains where we're at currently with development.
• Another delay? When can we expect Maghni to release?
We're a small team, but with our previous structuring, certain people in charge set goals that were... lofty, at best. As such, a lot of the previously promised demos or (what were planned to be) character reveals fell completely flat, despite the development team's best efforts. For that reason, it's difficult to give a firm timeframe beyond "Sometime in 2025". However, base model retrain 3.0 is right around the corner, we've finished implementing all the core languages we've planned to initially launch with (The number eludes me, but I believe it's around 12?), backer rewards are prepped to be shipped out in January (VocaTone still hasn't announced it yet, but the comic is still in development, as well as the digital album.), and we've enormously increased the number of voices and characters ready for development on the engine.
Fun fact: the languages, while plentiful, aren't the main cause of delay; actually, the work on those was mostly done months ago!
• Isn't there a better place all this can be going?
Actually...
We're currently implementing a blog section for our website as we speak; a proper, more organized place for me to share future updates for everyone interested. Since VocaTone isn't intrinsically tied to the project anymore (in fairness, they were never really developing the engine or vocals for it anyway), it seems I've been tasked with figuring out the best way to convey our progress to everyone. Hopefully this works, for the time being - but, when the blog is finished, I hope you'll read along as we further develop Maghni AI into the new year!
That's about it!
If there's any comments, questions, or concerns, I would be overjoyed to address them! Thanks again for being kind enough to offer some really important criticism. It seriously helps, and gives me a very strong sense of direction on what more I can be doing with my newfound position.
Looking forward to what we can show off in 2025! Also, Happy Holidays!
Last edited:
Thank you for stopping by to give an update, Cinna! I'm looking forward to hearing more progress next year.
Thanks for stopping by!
As long as you're here, a couple of clarifying questions about some things you've mentioned:
I imagine you have a lot on your plate so sorry for the long questions and sincerely good luck with everything!
As long as you're here, a couple of clarifying questions about some things you've mentioned:
- The "Maghni Moment 1" video, where the revised timeline was shown, promised 5 languages (English, Spanish, Mandarin, Japanese, and French) + 5 unannounced languages developed in contingent with 3rd parties. So that would be 10 languages promised to the public, not 12. Is this still what investors should be expecting at launch, both in quantity and in actual language selection?
I've harped on the IndieGoGo campaign far too much but I think the reason people are so confused is because the IndieGoGo campaign states very explicitly that Maghni AI "is capable of singing in over 40 languages", between the campaign copy and the original campaign video.
Yes, the exact number at launch did end up getting clarified in a follow-up video, but the "Maghni Moment 1" video was shared months after the campaign already closed. That means the "originally promised 5 languages" must have been specified in the "Maghni AI Q&A" video, (or maybe an unlinked twit post?) which while being a good method for connecting with the community, shouldn't have been the mode of conveying integral information. It was 2 hrs long and had plenty of discussion that didn't have much to do with the product itself, which while OK in a vacuum, makes it difficult to expect every backer to sit through all of that content to learn that the engine was only supposed to support 5 languages at launch rather than 40+. This Q&A in and of itself also complicates the picture because IIRC this is where Natalia discussed the wordnet-inspired phonetic architecture which implied seemingly intentionally that the selection of usable languages would have the capacity to expand rapidly. - "Last year or two" for AI engine development is a bit of a confusing statement. If it's only been "about a year of development", that would mean the IndieGoGo campaign video Oliver demo didn't actually come from Maghni AI's technology since the IndieGoGo was launched almost a year ago to the day. I'm sure that isn't true since there have been offhanded comments about that transition from the old Misbah concatenative technology to the AI synth happening in 2022, but the uncertainty there I imagine could be a bit confusing to a passerby
am I completely off-base here?
I imagine you have a lot on your plate so sorry for the long questions and sincerely good luck with everything!
No worries! I'll do my best to answer as many of these as I can, Pico.
#1 ) Yes, that's correct - the number of languages would be the ones you listed, plus the unannounced few. While I can't elaborate more on what those languages are at this time, I can say that since originally joining the development team a few years back, my priority has been both supporting these languages, but more so working with native speakers to give these languages the proper respect they deserve. As cool as it would be to launch out the gate with previously unsupported languages for a character like OLIVER, I feel we owe it to the communities that speak these aforementioned languages to give them a unique voice and character before it is appropriate for us to attribute said language to an unrelated voice library universally.
All that being said: yes - these ten languages are fully implemented, and when cross-lang is finished development, it will be possible to utilize any of them for any of our vocals. However, as we are working with quite a few native speakers for said languages, I feel it will be better to let these native voices release first, then roll out cross-language support for the applicable language for everyone else (Oliver, unfortunately, breaks this rule, but he'll be the only one going forward). It's our goal to have plenty of content to release post-launch, and I feel this method is more ethical and more feasible for both our developers and those interested in trying Maghni AI.
(To risk being personal, the idea of developing and essentially selling these languages without hiring and supporting native speakers sort of feels... not amazing, and is my main inspiration for this mindset. Hopefully you can see where I'm coming from!)
#2 ) Also, correct! I didn't really expect to have my words taken so literally, my apologies. "a year or two" is just my dialect, but very much was leaning towards "two". If the Kickstarter is now a year old, then Maghni AI's transition towards an AI-based synthesis engine is now officially past the two-year mark.
Also, a quick segway:
I can't speak for what was said in the past, but I'd like to reassure everyone that I'm striving for major change in how Maghni handled... everything, publically. It's no secret that the previous management wasn't as professional as it should have been (and by no means do I ever intend to turn the project into a cold, calculated robot with zero personality), but with new changes comes new opportunities!
If there is ever a moment where I accidentally get a detail from 2021(tweeted by Natalia at 3:42am on a Thursday) wrong, or what I say is inconsistent, please let me know! I am working with totally new information over here, and was not in this position until very, very recently. I may be casual and (attempt to be) funny, at times, but I pray that's never confused for malicious intent. Just wanted to clear that up, just in case.
With that...
I believe that's mostly everything! What users can expect at launch is still uncertain as we are actively in the middle of development, but future blog posts on our website should be far more consistent, helping shed some light on those topics as we further transition into the new year. If there's anything else I can assist with, do let me know!
#1 ) Yes, that's correct - the number of languages would be the ones you listed, plus the unannounced few. While I can't elaborate more on what those languages are at this time, I can say that since originally joining the development team a few years back, my priority has been both supporting these languages, but more so working with native speakers to give these languages the proper respect they deserve. As cool as it would be to launch out the gate with previously unsupported languages for a character like OLIVER, I feel we owe it to the communities that speak these aforementioned languages to give them a unique voice and character before it is appropriate for us to attribute said language to an unrelated voice library universally.
All that being said: yes - these ten languages are fully implemented, and when cross-lang is finished development, it will be possible to utilize any of them for any of our vocals. However, as we are working with quite a few native speakers for said languages, I feel it will be better to let these native voices release first, then roll out cross-language support for the applicable language for everyone else (Oliver, unfortunately, breaks this rule, but he'll be the only one going forward). It's our goal to have plenty of content to release post-launch, and I feel this method is more ethical and more feasible for both our developers and those interested in trying Maghni AI.
(To risk being personal, the idea of developing and essentially selling these languages without hiring and supporting native speakers sort of feels... not amazing, and is my main inspiration for this mindset. Hopefully you can see where I'm coming from!)
#2 ) Also, correct! I didn't really expect to have my words taken so literally, my apologies. "a year or two" is just my dialect, but very much was leaning towards "two". If the Kickstarter is now a year old, then Maghni AI's transition towards an AI-based synthesis engine is now officially past the two-year mark.
Also, a quick segway:
I can't speak for what was said in the past, but I'd like to reassure everyone that I'm striving for major change in how Maghni handled... everything, publically. It's no secret that the previous management wasn't as professional as it should have been (and by no means do I ever intend to turn the project into a cold, calculated robot with zero personality), but with new changes comes new opportunities!
If there is ever a moment where I accidentally get a detail from 2021(tweeted by Natalia at 3:42am on a Thursday) wrong, or what I say is inconsistent, please let me know! I am working with totally new information over here, and was not in this position until very, very recently. I may be casual and (attempt to be) funny, at times, but I pray that's never confused for malicious intent. Just wanted to clear that up, just in case.
With that...
I believe that's mostly everything! What users can expect at launch is still uncertain as we are actively in the middle of development, but future blog posts on our website should be far more consistent, helping shed some light on those topics as we further transition into the new year. If there's anything else I can assist with, do let me know!
Last edited:
Totally makes sense.
I understand the implicit frustration here of past information and public announcements being so disparate and random, but that’s precisely why I thought it was important to ask for clarity as to where things stand today from the new management’s perspective.
Thank you!
I understand the implicit frustration here of past information and public announcements being so disparate and random, but that’s precisely why I thought it was important to ask for clarity as to where things stand today from the new management’s perspective.
Thank you!