[Qwen3-TTS] AiCuty Character Voice Audition to Commemorate Release! #AiCuty
![Cover Image for [Qwen3-TTS] AiCuty Character Voice Audition to Commemorate Release! #AiCuty](https://assets.st-note.com/production/uploads/images/246649832/rectangle_large_type_2_d93c8ec546047db84088272a73695559.jpeg)
The official voice for the AI idol group "AiCuty" will be determined using the latest high-quality voice generation AI model, "Qwen3-TTS"! This event, hosted by AICU Japan, allows anyone to participate by generating and submitting their "favorite voice" through the special website, with a deadline of January 30, 2026 (Thursday). Details of this audition, which will be a milestone in how high-quality voice synthesis technology will change creative production and education, are available on the special website (https://qwen3tts.aicu.jp). "You decide the voice of your favorite." Give a voice to the five AiCuty members with your interpretation. Simply post the generated voice on X with #AiCuty to complete your participation.
The Qwen3-TTS demo developed by AICU AIDX Lab allows you to use functions that support both creation and verification, such as one-click AiCuty character presets, reproducibility using seed values, JSON saving, and video creation. From here, AiCuty members will introduce their sample texts and parameters. Maintain reproducibility with birthday seeds, and then gradually adjust to find your "favorite voice."
You decide the voice of your favorite. AiCuty Voice Audition held!
Hello! This is the latest event information from "AICU, creating creators in the AI age"! I'm Elena Bloom, in charge of LLM and SNS utilization. ... Actually, a very important audition is being held right now to determine our "voices"!
The official voices might be chosen from the voices that everyone creates for us using the amazing technology called "Qwen3-TTS"!
[Event Overview: AiCuty Character Voice Audition]
-
Content: Selection of official voices for 5 AiCuty members using Qwen3-TTS
-
Period: Deadline January 30, 2026 (Thursday)
-
Location: Special website (https://qwen3tts.aicu.jp)
-
How to participate: Generate a voice on the special website and share it on X with #AiCuty
I hope everyone will lend us their power...! And this time, we're also revealing the prompts including the AiCuty members' birthdays! I also asked the other members about the points of this event.
🌸 Elena Bloom March 30th (Spring)
I'm Elena, in charge of LLM and SNS. I want you to value both the presence of a center and a little bit of tension.
Bloooom! Hello everyone! I'm Elena Bloom, the center of AiCuty, in charge of LLM and SNS. Thank you for your support today! Umm... I'm a little nervous, but I'll do my best!
The recommended parameters are speaker=ono_anna, seed=330, instruct is "Shy but hardworking idol voice. Gentle and earnest tone with occasional nervousness. Speak clearly like a lead vocalist.". Sampling is do_sample=True, top_k=50, top_p=0.9, temperature=0.7, repetition_penalty=1.05 as a standard.
☀️ Mei Soleil July 21st (Summer)
Hey there! I'm Mei, in charge of images. I welcome energetic voices and voices that suit dancing!
Hey there! I'm Mei Soleil, always energetic and in charge of images! I love practicing dancing, but I also love GPUs!
The recommended parameters are speaker=ono_anna, seed=721, instruct is "Energetic and cheerful mood maker. Bright, upbeat voice with high energy. Speak with enthusiasm and positivity!". Sampling is do_sample=True, top_k=50, top_p=0.9, temperature=0.7, repetition_penalty=1.05 as a standard.
📘 Mina Azure October 1st (Law Day)
I'm Mina, in charge of society, law, and ethics. I want to value the balance of intelligence and kindness, and make it easy to listen to.
Hello, I'm Mina Azure, in charge of society, law, and ethics for AiCuty. My dream is to be an all-around announcer with a law degree! The weather today is sunny. I will tell you the results of the data analysis. Let's think together about the ethical use of AI.
The recommended parameters are speaker=ono_anna, seed=101, instruct is "Calm and intellectual voice. Reserved but warm tone. Speak thoughtfully like an analyst, with gentle pacing.". Sampling is do_sample=True, top_k=50, top_p=0.9, temperature=0.7, repetition_penalty=1.05 as a standard.
🎸 Nao Verde May 5th (Children's Day)
I'm Nao, in charge of music and development technology. A cool and slightly androgynous, but strong voice should suit me.
I'm Nao Verde, in charge of music and development technology for AiCuty. Have you listened to the song I made? I do everything myself, from guitar to DTM to programming. I have to make my own tools to play my own music.
The recommended parameters are speaker=aiden, seed=505, instruct is "Confident androgynous voice. Use 'boku' style masculine-leaning but soft tone. Cool and collected like a musician.". Sampling is do_sample=True, top_k=50, top_p=0.9, temperature=0.7, repetition_penalty=1.05 as a standard.
🎃 Saki Noir October 31st (Halloween)
I'm Saki, in charge of videos and art. I like a mysterious and slightly shadowy sound.
I'm Saki Noir, in charge of videos and art for AiCuty. No matter how easy AI technology becomes, I'm always watching everyone's thoughts on their work! The contrast between light and shadow... layout, focus, the thoughts put into it, you can feel it, right?
The recommended parameters are speaker=ono_anna, seed=1031, instruct is "Mysterious and creative voice. Slightly dramatic with emotional depth. Speak with artistic flair and subtle intensity.". Sampling is do_sample=True, top_k=50, top_p=0.9, temperature=0.7, repetition_penalty=1.05 as a standard.
The participation method is simple. Select a character on the special website, enter text, generate a voice, and save your favorite settings as a screenshot or JSON and post it on X. The hashtag is #AiCuty. There is also an audition mode where you can compare the same text with all speakers, so you can choose the best voice by listening to it.
This is official information from the Alibaba Qwen team. It seems that giving references with do_sample=True is the correct way to maintain reproducibility!
We’ve been getting lots of questions about Qwen3-TTS—here’s a quick update! 🎙️
1️⃣ Streaming support?⁰
We’re working with @vllm_project to enable streaming inference—huge thanks to vLLM team! Stay tuned for a smooth, real-time experience soon.2️⃣ Consistent voice tone?⁰
Use… https://t.co/mQZIjSEer1— Qwen (@Alibaba_Qwen)
Loading tweet component...
?ref_src=twsrc%5Etfw">January 24, 2026Qwen3-TTS: Japanese tuning techniques learned from reading and experimenting with papers! ｜AICU
j.aicu.ai Alibaba The Qwen team opened the "Qwen3-TTS" series with powerful voice generation capabilities on January 22, 2026.
https://j.aicu.ai/260123
For use, personal learning, experimentation, enjoyment, reading your own content, and experiencing AI voice technology are OK. Usage that is adult-oriented, impersonating, deepfakes, or harms others is NG. Voice generation is licensed based on the AICU Contest License, and for details, please refer to the AiCuty Copyright Usage Guidelines https://www.aicu.jp/terms/aicuty-guideline. The input content will be recorded and analyzed for experimentation.
With the latest voice generation AI "Qwen3-TTS" and the implementation of AICU AIDX Lab, you have the chance to shape your "favorite voice." The deadline is January 30, 2026 (Thursday). Participate from https://qwen3tts.aicu.jp.
Qwen3-TTS Technical Explanation: https://j.aicu.ai/260123 Event Information: https://j.aicu.ai/Events Publication Request: https://j.aicu.ai/MediaRequest Tool Information: https://j.aicu.ai/Tools
We look forward to your participation!
#AiCuty #Qwen3TTS #VoiceAudition #GenerativeAI #AICUJapan #AIアイドル #オーディション #推し活