Hey… lend me your ear?
55 whispered lines, best heard with headphones. Direction and distance are added through audio post-processing.
Open video in a new tab — Hey… lend me your ear?
Let’s make
something fun.
Made with curiosity.
AI, animation, and the occasional strange idea.
Small tools, technical experiments, and works in progress.
Work and experiments
Iroha-TTS
Controllable speech synthesis
Control emotion, whispering, and vocal events.
All four Japanese samples use the same line.
Listen by emotionJapanese samples
V1
[happy]
あっ、つい話し込んじゃってたわね。
Oh, I got carried away chatting.
0:00
Samples are from the Iroha-TTS V1 research blog. For each control’s capabilities and limits, see the research notes.
Demo videos
Short stories and a feature tour, voiced with Iroha-TTS. Japanese audio with Japanese and Chinese subtitles.
English subtitles
What Iroha-TTS can do
Emotion, intensity, laughter, whispering, and pronunciation, with side-by-side comparisons of the same lines.
Open video in a new tab — What Iroha-TTS can doAfter-School Melody
Three characters, changing emotions, and a short after-school story. All dialogue is generated with Iroha-TTS.
Open video in a new tab — After-School MelodyStory and audio credits
3D audio uses TH Köln’s KU100 near-field HRIRs, applied through convolution to the ASMR and feature-tour audio. HRIR data license: CC BY-SA 3.0.
Creative journal
Iroha-TTS-V1
Choose how a line sounds.
Emotion and intensity, whispers, laughter, and pronunciation for Japanese dialogue. Explore the seven controls, compare audio, and read the evaluation and setup notes.
Read the release article
About Iroha Studios
We build anime-related AI and AI agents out of personal interest.
These are hobby projects, with no aim to make a profit. Iroha-TTS is one of them. Here we share what we make, how we build it, and what we learn along the way.