Listen to a native speaker and an AI voice, at normal or slow speed, and record yourself to compare.
Native recordings by a professional Japanese voice actor, commissioned for this site. AI voices synthesized locally with AivisSpeech.
Recording analysis runs in your browser with Whisper (OpenAI, MIT) and japanese-hubert-base-phoneme-ctc-v4 (prj-beatrice, built on rinna's japanese-hubert-base, Apache 2.0) via transformers.js; notices in /asr-rt/NOTICE.