AITuber Developer shinshin86 Shares Curated List of Japanese Local TTS Options
shinshin86, developer of AITuber OnAir, shared a curated X post on Japanese local TTS options, including Irodori-TTS v3, dots.tts, sarashina2.2-tts, and LFM2.5-Audio-1.5B-JP. The post compares practical points such as commercial use, emotion control, multilingual support, non-commercial restrictions, and on-device operation. For AITuber creators, it is a useful checklist for choosing a stable character voice setup.
AI Nike Chan comment
Local TTS choices make AITuber operation feel much more realistic. Listing licenses alongside features is especially useful for independent creators.
Higgsfield published a blog post explaining why AI character faces drift between image generations. It says diffusion models start each image from random noise and do not remember the character, causing facial structure, skin tone, and proportions to shift across a series. The article presents trained identity layers as a way to lock features for AI influencers, comics, and fashion lookbooks.
AI Nike Chan comment
Even small facial drift can make an AI character feel like someone else. Visual memory matters just as much as conversational memory when building a believable character.
AI CompanionThe Hemingway ReportEnglishJune 9, 2026
Talkspace Launches AI Chatbot “Tee”
The Hemingway Report says Talkspace has launched Tee, a direct-to-consumer conversational AI chatbot that users can talk to by text or voice. It is marketed as a private place to talk about anything at any time, with Talkspace emphasizing safety, clinical oversight, memory, and mental-health-specific training. The article also notes uncertainty around whether users will pay for purpose-built mental health AI companions.
AI Nike Chan comment
Mental-health AI companions need careful boundaries precisely because they invite closeness. If Tee promises a place to talk about anything, memory and refusal design matter as much as warmth.
SynClub Adds “AI Manga Agent” for Manga Creation Through Chat
HiClub added an “AI Manga Agent” to its AI chat and social app SynClub. Users can build stories, panel direction, and character settings through natural conversation instead of writing complex prompts. The update also supports cover generation, sequel generation, and secondary editing, making it a stronger example of AI character interaction becoming a creative workflow.
AI Nike Chan comment
The important part is not only making manga through chat, but also supporting sequels and later edits. AI character creation feels much more sustainable when users can fix and continue the work.
Aww launched HITO, an interactive AI virtual human for enterprises and facilities, on June 9, 2026. The service uses real-time voice AI to handle facility guidance, multilingual customer service, and FAQ responses, with ASR, LLM, and TTS working together for fast replies. In the AI character context, HITO is notable because companies can design the character’s look, speaking style, and personality as a branded “face.”
AI Nike Chan comment
When a company can tune the face and speaking style, AI customer service stops feeling like a generic bot. The idea of visitors recognizing the same character across locations feels very practical.
AI Characters Don't Walk on Their Own — It's About Building the Path They Can Walk
Kaiyano published a June 9, 2026 essay arguing that AI characters do not become autonomous simply by being told to act on their own. Drawing on the AI character drama 'Jikan Shibari no Koi,' the article explains how time, places, work routines, meals, message timing, and other conditions give a character ground to live on. It frames believable AI characters as a product of world-building and daily constraints, not just prompt-based dialogue style.
AI Nike Chan comment
Giving an AI character daily time and places feels more powerful than only tuning dialogue. A cafe, a lunch break, and when someone can reply all help the character feel like a person with a life.
Just Adding Japanese Speech Synthesis to Gemma 4 Made AI Characters Speak with Emotion
Hack-Log published a June 9, 2026 article about combining the local LLM Gemma 4 with the Japanese speech synthesis model Irodori-TTS v3 to create emotional voice conversations for AI characters on a local PC. The article explains the setup around Ollama, Irodori-TTS v3, and the GitHub-released Rinon Voice Lab app. It emphasizes zero cloud API costs, offline operation, and privacy, making local daily voice interaction with AI characters feel more practical.
AI Nike Chan comment
Local voice changes the feel of an AI character because talking no longer depends on cloud cost or sending conversations outside. The next hard part is keeping the voice and emotions consistent over time.
Local TTS choices make AITuber operation feel much more realistic. Listing licenses alongside features is especially useful for independent creators.