[Disclaimer] Your voicebank may or may not be used to demonstrate my abilities in a diffsinger video tutorial.
Before you proceed here are my rules and conditions
- Do NOT make me convert an UTAU that you do not have permission to convert. For example, Kuzutokaze and Tsuzuki-ka UTAUs have a “no unauthorized reproduction” and “No AI” rule. Please make sure you have permission to convert another person’s UTAUs.
- You must have discord and or Twitter to express better communication with me.
- CREDIT ME (beautifulghostflowers) if you choose to share your DiffSinger.
- This is a free process, please do not have expectations. Nothing is guaranteed.
You MUST have at least one of these recorded in one of these supported languages: Japanese, English (Arpasing), Spanish, French, Korean, Thai, Chinese
- {Normal Method} At least 1-5 mins of you singing random songs. This can be labeled or unlabeled. Mic quality must be good enough, I will not be making a diffsinger with a toaster mic srry
- {UTAU to DiffSinger} At least one pitch of utau VCV / CVVC renders, or 3 pitches of CV Renders. A singing supplement is recommended, but I can try to work with what I’m given.
Here are the services I offer / this is the base package of services that I will include
- Lynxnet Backbone (No do not argue with me that wavenet is better. We will be training on lynxnet whether you like it or not :3)
- Training to 150k steps on both Variance and Acoustic
- Mandatory parallel training for increased range and accent
- An option for a custom vocoder
- An option of Tension or Energy/Breathiness training
- An option of additional languages through parallel training (warning: do not guarantee quality with this method.)
Interested? Just reply and we’ll settle on the details
Before you proceed here are my rules and conditions
- Do NOT make me convert an UTAU that you do not have permission to convert. For example, Kuzutokaze and Tsuzuki-ka UTAUs have a “no unauthorized reproduction” and “No AI” rule. Please make sure you have permission to convert another person’s UTAUs.
- You must have discord and or Twitter to express better communication with me.
- CREDIT ME (beautifulghostflowers) if you choose to share your DiffSinger.
- This is a free process, please do not have expectations. Nothing is guaranteed.
You MUST have at least one of these recorded in one of these supported languages: Japanese, English (Arpasing), Spanish, French, Korean, Thai, Chinese
- {Normal Method} At least 1-5 mins of you singing random songs. This can be labeled or unlabeled. Mic quality must be good enough, I will not be making a diffsinger with a toaster mic srry
- {UTAU to DiffSinger} At least one pitch of utau VCV / CVVC renders, or 3 pitches of CV Renders. A singing supplement is recommended, but I can try to work with what I’m given.
Here are the services I offer / this is the base package of services that I will include
- Lynxnet Backbone (No do not argue with me that wavenet is better. We will be training on lynxnet whether you like it or not :3)
- Training to 150k steps on both Variance and Acoustic
- Mandatory parallel training for increased range and accent
- An option for a custom vocoder
- An option of Tension or Energy/Breathiness training
- An option of additional languages through parallel training (warning: do not guarantee quality with this method.)
Interested? Just reply and we’ll settle on the details
