(I'm posting way too much aren't i) I want to make a vocal synth. I need a team.

[Suzuki Hoshi]

If you are good at programming, I'd love if you could help
It's going to be open-source, and the voicebank devkit will come with the synth.
It's also going to be a bit more advanced than VOCALOID, and may use AI (Google Colab for training)
If you want, you can be the voicer for the default bank

 

[Suzuki Hoshi]

bump

 

[GreenPear03]

Creating an AI vocal synth is a large task, and typically requires plenty of experience to create, if you aren't simply planning to create an edited fork of an already existing AI vocal synth such as Diffsinger or NNSVS. However, a concatenative vocal synth (similar to UTAU), is easier to make, and simply requires an audio engine, maybe a resampler (such as the commonly used WORLD, which has wrappers in the most commonly used programming languages), and a simple program that can read the notes, pitch, and length of each note. Other features, such as tuning, vibrato, and an interface can be added later in its development.

Also, what programming langauge are you planning to make the engine in, and what features would you like to add? I wish your project good luck, as someone who knows someone developing a vocal synth!

 

[heynotloid]

I would be happy to make a voicebank for the synth as soon as it comes out.

 

[SaKe]

If you are good at programming, I'd love if you could help
It's going to be open-source, and the voicebank devkit will come with the synth.
It's also going to be a bit more advanced than VOCALOID, and may use AI (Google Colab for training)
If you want, you can be the voicer for the default bank

whoah whoah whoah whoah hold up

Deep-learning vocal synths are incredibly complex, far more than something like UTAU- before you start coding you should read some academic papers on exactly how they work. I believe the papers for NNSVS and Diffsinger are readily available.
Also, it would be very hard to make an AI version of UTAU, where users can create their own voicebanks, because deep learning, unlike concatenative synthesis, requires active singing. Plus, Vocaloid devkits are MASSIVE (source: I have one) and would be really hard to incorporate into a synthesizer, let alone make it user-friendly.

Creating a GUI for NNSVS would be a good place to start if you really want to make an AI synthesizer.

 

[Suzuki Hoshi]

I would be happy to make a voicebank for the synth as soon as it comes out.

sure!

Post automatically merged: Monday at 6:14 AM

whoah whoah whoah whoah hold up

Deep-learning vocal synths are incredibly complex, far more than something like UTAU- before you start coding you should read some academic papers on exactly how they work. I believe the papers for NNSVS and Diffsinger are readily available.
Also, it would be very hard to make an AI version of UTAU, where users can create their own voicebanks, because deep learning, unlike concatenative synthesis, requires active singing. Plus, Vocaloid devkits are MASSIVE (source: I have one) and would be really hard to incorporate into a synthesizer, let alone make it user-friendly.

Creating a GUI for NNSVS would be a good place to start if you really want to make an AI synthesizer.

i mean
it doesn't have to be AI
honestly I'm still unsure

Post automatically merged: Monday at 6:14 AM

whoah whoah whoah whoah hold up

Deep-learning vocal synths are incredibly complex, far more than something like UTAU- before you start coding you should read some academic papers on exactly how they work. I believe the papers for NNSVS and Diffsinger are readily available.
Also, it would be very hard to make an AI version of UTAU, where users can create their own voicebanks, because deep learning, unlike concatenative synthesis, requires active singing. Plus, Vocaloid devkits are MASSIVE (source: I have one) and would be really hard to incorporate into a synthesizer, let alone make it user-friendly.

Creating a GUI for NNSVS would be a good place to start if you really want to make an AI synthesizer.

i mean
it doesn't have to be AI
honestly I'm still unsure

Post automatically merged: Monday at 6:17 AM

Creating an AI vocal synth is a large task, and typically requires plenty of experience to create, if you aren't simply planning to create an edited fork of an already existing AI vocal synth such as Diffsinger or NNSVS. However, a concatenative vocal synth (similar to UTAU), is easier to make, and simply requires an audio engine, maybe a resampler (such as the commonly used WORLD, which has wrappers in the most commonly used programming languages), and a simple program that can read the notes, pitch, and length of each note. Other features, such as tuning, vibrato, and an interface can be added later in its development.

Also, what programming langauge are you planning to make the engine in, and what features would you like to add? I wish your project good luck, as someone who knows someone developing a vocal synth!

i might do a non-ai vocal synth. you and SaKe's opinions are very helpful.
I was going to use python but will that work?

The stuff I want to add:
- Pitch plugin
- Vibrato editor
- Voice color
- Phrases
- Voice steadiness parameter
- Most VOCALOID parameters
- open-source voicebank editer
- open-source editor


I want to have a team for this since I'm an idiot

 

[GreenPear03]

I think Python is generally good for voice synthesizers, but you'll also need to find people with experience in languages such as C++ for Mac, or Swift/XCode for Mac for example, because they're what desktop applications are typically made in. I can help, but only if its web based, and if so, I can't help with any parts made in other programming languages.
Also, if you use WORLD, it is pretty easy to have a voice steadiness parameter (which is basically modulation in UTAU, but I don't have much experience with Vocaloid), and most parameters are easy to replicate.

I can generally help with the direction of the program, and voice a default voicebank. If you want it to be web based, I can help! Even if you want to have a team of more experienced people, its good to generally have some programming knowledge in the programming language you want to make this voice synthesizer in. I recommend prototyping your program and knowing exactly what you're going to do, and how it might be accomplished before getting programmers to work on the program.

 

[LAZARUSCROSS]

Is this going to be a free product, or will you be selling this later on once it's done?

If the latter, then you're going to have to have an idea of what features you're going to provide to make your platform stand out from the competition.

If the former, then you're likely going to have to do most of the work either yourself or somehow find close friends who are willing to help you, because programming can be very taxing and laborious work and very few people are willing to do that sort of thing for free unless you have people that are passionate about the idea.

And as others have said, unless you have a strong background in deep learning, you're probably going to want to make something more simple like a concatenative engine. But even then, making a fullblown program with no coder experience will be a quite the daunting task. If you're serious about this, I highly recommend learning C++/Python and making a few smaller programs in those languages before tackling something of this scope.

 

[Suzuki Hoshi]

I think Python is generally good for voice synthesizers, but you'll also need to find people with experience in languages such as C++ for Mac, or Swift/XCode for Mac for example, because they're what desktop applications are typically made in. I can help, but only if its web based, and if so, I can't help with any parts made in other programming languages.
Also, if you use WORLD, it is pretty easy to have a voice steadiness parameter (which is basically modulation in UTAU, but I don't have much experience with Vocaloid), and most parameters are easy to replicate.

I can generally help with the direction of the program, and voice a default voicebank. If you want it to be web based, I can help! Even if you want to have a team of more experienced people, its good to generally have some programming knowledge in the programming language you want to make this voice synthesizer in. I recommend prototyping your program and knowing exactly what you're going to do, and how it might be accomplished before getting programmers to work on the program.

I'd love your help!

Post automatically merged: Monday at 10:14 AM

Is this going to be a free product, or will you be selling this later on once it's done?

If the latter, then you're going to have to have an idea of what features you're going to provide to make your platform stand out from the competition.

If the former, then you're likely going to have to do most of the work either yourself or somehow find close friends who are willing to help you, because programming can be very taxing and laborious work and very few people are willing to do that sort of thing for free unless you have people that are passionate about the idea.

And as others have said, unless you have a strong background in deep learning, you're probably going to want to make something more simple like a concatenative engine. But even then, making a fullblown program with no coder experience will be a quite the daunting task. If you're serious about this, I highly recommend learning C++/Python and making a few smaller programs in those languages before tackling something of this scope.

it's gonna be free.

 

[SunnyWolves]

You asked for programmers to help you. Do you have the money to pay programmers skilled enough to make a vocal synth?

 

[Suzuki Hoshi]

You asked for programmers to help you. Do you have the money to pay programmers skilled enough to make a vocal synth?

I'm broke

 

[SunnyWolves]

I'd imagine you'd have trouble then. It'd be hard to imagine anyone willingly working on something as complex as a vocal synthesizer for free. You're likely going to have to do most of the work yourself if you're serious about this.

 

[Suzuki Hoshi]

I'd imagine you'd have trouble then. It'd be hard to imagine anyone willingly working on something as complex as a vocal synthesizer for free. You're likely going to have to do most of the work yourself if you're serious about this.

i could try
maybe....
you're probably right
though GreenPear03 said they'd help with the direction of the program and voicing a default bank which i must say was very nice

 

[SaKe]

i could try
maybe....
you're probably right
though GreenPear03 said they'd help with the direction of the program and voicing a default bank which i must say was very nice

I'd be willing to help with the UI design and some linguistic stuff, like writing reclists and that kind of thing.

 

[Izayahsan24]

If you are good at programming, I'd love if you could help
It's going to be open-source, and the voicebank devkit will come with the synth.
It's also going to be a bit more advanced than VOCALOID, and may use AI (Google Colab for training)
If you want, you can be the voicer for the default bank

I can make plugins (also is it going to be on mobile?)

 

[Suzuki Hoshi]

i could try
maybe....
you're probably right
though GreenPear03 said they'd help which was nice.

but it

I can make plugins (also is it going to be on mobile?)

not mobile

Post automatically merged: Tuesday at 10:27 AM

I'd be willing to help with the UI design and some linguistic stuff, like writing reclists and that kind of thing.

that'd be nice

 

[Izayahsan24]

but it

not mobile

Post automatically merged: Tuesday at 10:27 AM

that'd be nice

K just wondering if I needed to make pc and mobile plugins

 

[Suzuki Hoshi]

K just wondering if I needed to make pc and mobile plugins

pc plugins is fine

 

[Suzuki Hoshi]

I think Python is generally good for voice synthesizers, but you'll also need to find people with experience in languages such as C++ for Mac, or Swift/XCode for Mac for example, because they're what desktop applications are typically made in. I can help, but only if its web based, and if so, I can't help with any parts made in other programming languages.
Also, if you use WORLD, it is pretty easy to have a voice steadiness parameter (which is basically modulation in UTAU, but I don't have much experience with Vocaloid), and most parameters are easy to replicate.

I can generally help with the direction of the program, and voice a default voicebank. If you want it to be web based, I can help! Even if you want to have a team of more experienced people, its good to generally have some programming knowledge in the programming language you want to make this voice synthesizer in. I recommend prototyping your program and knowing exactly what you're going to do, and how it might be accomplished before getting programmers to work on the program.

maybe it'll have a website version!
I dunno

 

[GreenPear03]

For know, it'll probably be better to make it only a desktop version, however I think a web-based vocal synth people can make voices for would have higher demand.