This site is the home for my odd character vocal synths project, "Project Ameko"!
Basically they are characters + voicebanks made and voiced by me, for vocal synth softwares! (Currently working on DiffSinger voicebanks atm...)
Click on the icons below to lead yourself to the character/voicebank pages!
[Sorry, Ameko Kishi is still under development! Coming soon!]
NEWS:
Ameko Kero (check the character's page for more info):
27/2/2025 - Ameko Kero has been updated to ver 1.3-MbPa0.8 - Patisserie Build!
10/21/2024 - Ameko Kero has been updated to ver 1.2-Mb5.0 - Millefeuille build!
10/5/2024 - Project Ameko Tumblr Blog created!
2/5/2024 - Ameko Kero ver 1.2-Mb5.0 (Millefeuille update) teased!
5/13/2024 - Ameko Kero has been updated to ver 1.1 - Reflow build!
4/17/2024 - Ameko Kero has been publicly released for the first time, ver 1.05!
3/3/2024 - Ameko Kero's cover of "flos" by R Sound Design had been released!
30/1/2024 - Ameko Kero's first demo cover video has been released!!
TERMS OF SERVICEBy using the assets of the Ameko Project (applies to both characters and voicebanks), you are applicable and must be compliant to the terms of service written below! The section will be devided into parts as listed in the "List of Contents":
List of Contents
1. General Consensus / Intoduction
2. Depiction-based restrictions
2a. General
2b. R-18 / NSFW specific restrictions
2c. Fan derivatives
3. Commercial / Non-commercial use
4. Additional voicebank specific terms
4a. Redistribution
4b. Modification
4c. Utilization for other softwares and projects
5. Credentials
MILLEFEUILLE SPECIFIC TERMS
As of late 2024, Ameko Kero has been in collaboration with the UFR Team and their Millefeuille project. Below are the terms written in conjunction by the UFR Team regarding DiffSinger voicebanks post-ver 1.2-Mb5.0:Due to the Millefeuille model using data from many voice providers the following standardized Terms Of Service apply to all characters:All models and model assets created and trained by UFR will be referred to as "Resulting models"- Resulting models must be free upon distribution, selling or use for trades is prohibited.
- No AI generated art may be used for voice bank assets, covers, PVs, lyric videos, album covers, or other media depicting the resulting model.
- Commercial use of renders or model assets for advertisements or products not including the sell of doujin music is prohibited.
- The use of renders or model assets for creating new products, corpuses, or AI training is prohibited.
- Resulting models may not be used for illegal activity.
- Resulting models may not be used in depictions of non-consensual relations.
- Resulting models may not be used for hate speech.
- Individual character TOS may add more restrictions but cannot override these basic protections outlined in the UFR Multimodel TOS.1. General Concensus / Introduction
i. As stated above, this terms of service is applicable for the illustrative / graphical / auditory depictions (characters), and the voice database files provided (voicebanks) in this project.ii. Generally on it's own, the characters and voicebanks in this project (HoodyP's Project Ameko) falls under CC BY-NC-ND 4.0 & Commons Clause License. However, commercial use can be permissible when permission or license is given (refer to 3).iii. While the voicebanks in this project utilizes generative vocal synth softwares (at the moment, only DiffSinger), this was done in an ethical manner and should remain that way. So please DO NOT associate, use or relate ANY of the assets in this project with or for any sort of unethical softwares, technologies, and / or acts (this includes; generative AI art, ChatGPT and other similar softwares, cryptocurrencies, NFTs, etc.)iv. As with the standard rules of any given vocal synth databases, using the rendered output from a premade voicebank to be made into another sort of database / voicebank, no matter if it's for the same software or a different one, is generally forbidden.v. When publishing works that use or depict assets from this project, please ensure that the title or label states the name of the character(s) / voicebank(s) depicted or used. It's highly encouraged to also credit the author, HoodyP / HoodyPisDed if possible (though you don't have to for this as it is not too enforced).2. Depiction-based restrictions
Below covers both the illustrative and auditory depictions of the characters, and the voices from the voicebanks.2a. General
This section will be divided into both "Allowed" and "Forbidden" parts. "Allowed" is what is either generally allowed or encouraged, while "Forbidden" is what is not allowed, and goes against the terms of service.Allowed
- Any sort of LGBTQ+ representation (as it is highly encouraged and runs course with the canon of the characters in this project!).- R-18 or NSFW based depictions, this applies to both the characters and the voices. However, it does have it's own restrictions (refer to 2b).- Informing on general social issues in an inoffensive/positive manner.- Informing on events, activities, and others.- Fan derivatives are allowed, but once again, there are also specifications (refer to 2c).Forbidden
- Usage that encourage harassment to any individuals or groups.- Usage that promotes hate against any individuals or groups, such as racism, homophobia, transphobia, ableism, sexism, misogyny, misandry, etc.- Usage for and/or to promote harmful and/or illegal activities/actions.- Usage for anything extremely or explicitly political and/or religeous (this includes: relating to real life governmental parties/politicians, controversial stances on social issues, movements and/or messaging that harms, demeans and/or vilify individuals or communities).- Any sort of usage that promote or incite trolling, spamming or raiding activities.- Any depiction of romantic or sexual pairings that would be considered a "proship" (abusive / manipulative relationships, minor X adult, human/anthro x fully feral animal/creature, etc.). For the abusive / manipulative aspect specifically, this rule is exempt if the intent was for discouraging against the behavior.- As Kero and Kishi were meant to represent the likeness and characteristics of the creator/voicer (HoodyP), it is highly discouraged to depict romantic and/or sexual relationships using Kero and Kishi (the characters/visual depiction), and closely relating derivatives, that are considered "male X female". This is due to the current climate in which same sex relationships are still being frowned upon and even actively suppressed over by a lot of people and in many places (as well as the creator's personal experiences rooted with trauma/anxieties, being a gay man in one of these heavily suppressing countries).
However, this does NOT apply to the usage of the voicebanks alone, as long as the characters who were in said romantic/sexual pairings are NOT RELATED IN ANY WAY to the Kero and Kishi characters.
(For further clarification and reminder, trans, non-binary and genderswap (lesbian) depictions are generally allowed and are not in any way relating to this rule!)- Usage that promotes misinformation and/or deepfake activities (especially using the voicebank to impersonate the voicer/creator).2b. NSFW (suggestive, pornographic and graphic) specific restrictions
i. Generally: In every NSFW aspects, it is FORBIDDEN to include depictions of;
- Minors (underaged individuals, including any loli/shota expressions)
- Feral (non-anthro) creatures/animals
- Elements considered "too grotesque" (bodily wastes, fetishization/sexualization of gore, etc)
- Real life tragedies
- Any kinks/fetishes that involved harming and painBelow are not forbidden, but highly discouraged by the creator (gray area):
- Uncommon paraphillia (something like "transforming into inanimated objects", whatever this is, and anything of similar elements)
- Hypertrophilia (overexaggerated/oversized erogenous zones).(Further clarification and reminder, everything stated in 2a ALSO applies for NSFW works generally.)ii. Suggestive works: Aside from what has been listed above (2b, i), it is generally allowed, as long as it is pandered/cultivated only towards a mature audience/those who are aware and allows to see such depictions.
What falls under "suggestive":
- Non-explicit sexual/suggestive outfits (i.e. swimwear, underwear, gimp suits and gears)
- Generally harmless/internet common fetishes (i.e. soft vore, macrophilia and inflation)
- Suggestive poses and actions (humping motion, hoop + pointing finger, ahegao, moaning, etc)iii. Pornographic works: Similarly, it is also generally allowed (and still please take note of 2b, i). Though, keep in mind to "section off" said works from general viewing/other SFW works (i.e. adult-only platforms/websites/accounts, labels/hashtags, restriction-based filtering, etc). Please also comply with the rules and regulations of wherever platform the works are being posted/released on.
What falls under "pornographic";
- Exposed genitals/erogenous body parts
- Explicit/graphic depiction of sexual activities and language.iv. Graphic works: "Graphic" in this section refers to depictions of grotesque elements (gore, violence, etc). ONLY HORROR (NOT SHOCK) AND VENT BASED INTENT ARE ALLOWED.
What falls under "graphic":
- Gore; blood and organs
- Violence/brutalities/injuries2c. Fan derivatives
i. Fan derivatives are allowed to be made however, this only applies to the illustrative and auditory (rendered outputs) depictions of the characters. ANY DISTRIBUTION OF MODIFIED VOICEBANK FILES TO FIT FAN DERIVATIVES ARE FORBIDDEN.ii. When making a fan derivative, editing the official illustrations and passing it off as a fan derivative is highly discouraged. However this is laxed off if it was intended to be made as a harmless / inoffensive joke (For example: this character ).iii. Said derivative can only be considered as a "fan derivative" when it doesn't stray too much away from the main canonical characteristics of the character(s).iv. When showcasing fan derivative(s) in public posts, please state the original character that it derives from and also credit the creator (HoodyP / HoodyPisDed) for the original character.3. Commercial / Non-commercial use
- As of 6th October 2024, assets in Project Ameko are a form of donationware! This means: i. Usage of the characters and voicebanks are free to be used commercially for original songs with monetary gain (YouTube, music platforms, etc), AS LONG AS PROPER CREDIT IS GIVEN (this means including the name of the character/voicebank in the main title or body of content, and credit me, HoodyP for being the creator of the voicebank(s)!) ii. For other sorts, such as merchandises, etc, you'll have to contact me for permission and agreement! iii. Being a donationware, you can support me, the creator through Ko-fi here: HoodyP's Ko-fi! It will greatly help me and for this project I'm passionate for! iv. This however DOES NOT INCLUDE usage in cryptocurrencies, NFTs, generative AI "art", and other similar things, as I've already stated very clearly earlier in this TOS that I DO NOT ALLOW MY CHARACTERS AND ASSETS TO BE USED BY SUCH.4. Additional voicebank specific terms
Below are the additional terms applied specifically for the voicebank and the files included.4a. Redistribution
- Redistribution of the voicebank files on different hosting sites / platforms ARE NOT ALLOWED. To share the download of the voicebank files, please provide the link to this site instead (hoodypprojectameko.carrd.co). However this rule is exempt in the case if: i. Permission is given to upload / share the files on a different site by the creator (me). In that case I will also share the link in this site.
ii. Said voicebank has been discontinued and is not available to download through this site. For archival reasons, people may upload the discontinued voicebank to a different site ONLY when said voicebank is confirmed discontinued.4b. Modification
- Distribution of the entire modified voicebank's files ARE NOT ALLOWED. However, distribution of certain modified configuration files (dsdict.yaml dictionary files, and / or dsconfig.yaml files) ARE ALLOWED, and can also be utilised for the creation of another voicebank (unrelated to the Ameko Project that is). Just at least credit HoodyPisDed + UFR Team if done so.4c. Utilization for other software and projects: As stated, modifying the premade voicebank to be used in an entirely different engine is forbidden, however, you can reach out to me for permission to use Kero's (and/or Kishi's if available soon) voice corpus to create a new port/voicebank for said different software! (HoodyP's socials page)
Name(s):
Ameko Kero (雨子ケロ, character's name)
Droppledale Croaker (English name, nickname under the musical duo group, "Droppledale")Species:
Cyborg Black Spotted Wood FrogGender and Preference:
- Male (he/they)
- Gay, exclusively attracted to guysAge:
- 19 (When he became a cyborg)
- (based on however old the creator, HoodyP, is lol)Relationships:
Ameko Kishi - boyfriend / partnerOccupation:
- Lead singer, lyricist and keyboard player of the indie musical duo, "Droppledale" (nicknamed as "Droppledale Croaker")
- Works at a local museum as a tour guideLikes:
- Hot coffee and tea
- Basking in the sun
- Cooking and baking
- Sleep, being lazy, and sometimes cuddling Kishi
- Writing (books and lyrical writing)
- Lo-fi music and melancholic/sappy songs
- Eating bugs
- Collecting sewing buttons he finds on the ground (to make an odd jacket with them)
- Indie rock musicDislikes:
- Kishi bugging Kero when Kero is busy writing
- A messy bedroom
- Spaghetti-O'sExtra Lore Stuff
- Kero and Kishi lives in a amphibian + reptilian dominant universe in which robotics and mecha technology has advanced and normalised in their society.- Kero and Kishi were already childhood friends, though later reunited in their late teens through an indie band scene, and eventually started dating.- During their 3rd date, Kero and Kishi got involved in a terrible mall accident which leaves many injured and some died. Kero and Kishi survived the incident, however the incident caused them to be severly mangled and were put into a coma.- Some time has passed, and eventually both Kero and Kishi woke up in different infirmary rooms of a lab. finding themselves in mecha bodies. They are surrounded by many staffs, and also many people in similar bodies as them. One of the staffs told him that the more severe victims of the mall incident were granted approval for becoming the third batch of patients to receive "cyborg bodies", in which Kero have heard from an online news somewhere.- The cyborg body wasn't perfect and it does have flaws and limits, but Kero got used to the body quickly and loved it because it reminded him of a "cool cartoon robot guy" he watches in his childhood.- Leaving the lab after healing up for a few days, Kero sees Kishi outside and reunited with him again. After visiting family and friends again to bring forth the news regarding their current state, Kero and Kishi moved to a small studio apartment at a quite peaceful town somewhere. The cyborg bodies were definitely a reminder of such a traumatic event, but it does strengthened the bond between Kero and Kishi somehow...
Voicebank:Ameko Kero DS (DiffSinger)
Ameko Kero DS is a versatile masculine DiffSinger (in OpenUTAU) voicebank with capabilities of singing in various languages!
(English, Japanese, French, Korean, Spanish, Italian, German, Russian, Polish, Thai, Mandarin and Brazillian Portuguese)The voicebank contains 4 vocal modes!:
- Neutral (the main voice)
- Soothe (a softer/milder voice)
- Sludge (a more tense/solid/stronger voice)
- Fly (falsetto)(Recommended to use the latest available version of OpenUTAU for stability and the best quality! You can get them through the OpenUTAU Github page below!)
ADDITONAL NOTES:
If you don't know much about OpenUTAU + DiffSinger, or you want to explore more stuff about it, you may refer to the links below!
OpenUTAU Github page
Installation guide by Nebula Meadow
OpenUTAU Lunai Edition by Lunai Project Team!
Millefeuille Build specific notes:
- As of ver1.3-MbPa0.8 - Patisserie Build, autopitch functionalities has been reimplemented with a newly trained model made with Millefeuille in mind. However, "Tension" parameter functionalities are not available in Millefeuille based builds as it is deemed unstable to be trained at the current rate. A "Whisper" vocal mode is planned and in the works to be implemented in the future to circumvent the issue!
- All the phonemes used for the Millefeuille build can be found here: UFR DiffSinger Tutorial!The data used for Ameko Kero has also been used to train HifiPLN, a community based vocoder managed by Scarfmonster and PixPrucer!
You can get it here and try it out as an alternative to the default vocoder that was included in the voicebank!
Credits (post-Millefeuille)
Character, voicer, labeller: HoodyP / HoodyPisDedTools used for labelling/dataset creation:
LablMakr: tigermeat
DB creation tool: UtaUtaUtauMillefeuille update:
UTAU French Resources Team staff: Hibya (labels checking/reedit), imsupposedto (training)
Voice datas used for multitraining here: Millefeuille XLS CreditsChangelogs
7/2/2026 - Changed the main lore for Ameko Kero!
27/2/2025 - Ameko Kero has been updated to ver 1.3-MbPa0.8 (Patisserie Build)!
21/10/2024 - Ameko Kero has been updated to version 1.2-Mb5.0 (Millefeuille Build)!
5/13/2024 - Ameko Kero has been updated to version 1.1 (Reflow Build)!
4/17/2024 - Ameko Kero is publicly released, as version 1.05 (Tension and Gender update)!
4/3/2024 - Ameko Kero is first released as a DiffSinger Discord Server Beta Exclusive, version 1.0