v3.0.0

3.0.0 (2026-07-21)

⚠ BREAKING CHANGES

• asr: sherpa-onnx-node removed, transcribe-cpp is now the STT engine

Features

• asr: replace sherpa-onnx with transcribe-cpp + LLM recommendation system (b9e33ac)

v2.5.0

2.5.0 (2026-06-09)

Features

• add 10 UX and performance improvements (#76) (1f1bdfb)
• sessions: add WAV recording and offline reprocess (#74) (8bac8f8)

v2.4.0

2.4.0 (2026-06-09)

Features

• sessions: add VTT-based transcript sessions with latency fix (#72) (5893630)

v2.3.1

2.3.1 (2026-06-08)

Bug Fixes

• build: disable hardenedRuntime to fix adhoc signing crash (#70) (e50ee66)

v2.3.0

2.3.0 (2026-06-08)

Features

• permissions: add macOS microphone permission handling (#67) (c264cb6)

v2.2.1

2.2.1 (2026-04-29)

Bug Fixes

• ci: resolve pnpm version conflict in build-dmg (#64) (942655e)

v2.2.0

2.2.0 (2026-04-28)

Features

• asr: add Voxtral-Mini-4B-Realtime as streaming STT engine (#60) (b871f90)

v2.1.1

2.1.1 (2026-03-17)

Bug Fixes

• build: remove asyncify.wasm reference dropped by onnxruntime-web (fb1ea18)

v2.1.0

2.1.0 (2026-03-17)

Features

• asr: add Whisper medium/large-v3 model support with Settings UI (#50) (ec35357)

Performance Improvements

• autopilot: reduce SFX auto-trigger latency — Phase A (#49) (8fbe726)

v2.0.1

2.0.1 (2026-03-16)

Bug Fixes

• e2e: tighten ASR skip guard to avoid false-skipping during reranker download (#47) (e009d8c)