JP TTS — Preview

Listening interfaces from the ongoing research. Pick a session below.

Blind A/B — 32 voices

All current candidates rendering the same JP passage. Engines · blind reveal · pipeline tabs.

Round-7 adds: japanese-parler-tts (Apache-2.0) ×2 · OuteTTS v1.0 ×1 · VoxCPM2 instruct + 8-bit clone. Plus all prior Round-6 entries.

→

Emotion sweep — instruct mode

Qwen3-TTS vs VoxCPM2 across 5 emotions (calm / sad / happy / angry / anxious).

Tests soft instruct steering · same speaker · same passage

New Round-7: jparler_* — japanese-parler-tts-mini (Apache-2.0, description-steered). outetts_en_female — OuteTTS v1.0 1B llama.cpp Metal, cross-lingual EN→JP. voxcpm_instruct + voxcpm_8bit_clone — previously missing VoxCPM2 variants now in.

Round-6: chatterbox_* — Resemble AI MIT. sarashina22_clone — SB Intuitions/SoftBank NC, JP-first.

Listening order from RESEARCH.md: Sarashina2.2 → Irodori-TTS v3 → VoxCPM2 Voice Design → Qwen3-TTS → Chatterbox → Google Chirp3-HD → Supertonic 3 → Fish S2 Pro (NC ceiling). Round-3 consensus says Irodori-TTS v3 fixes the fluent-foreigner accent that bit earlier rounds.