# Changelog ## v0.5.1 ### Features - Added `esp_capture_service_ai_audio_src_trigger_wakeup()`, `trigger_sleep()`, and `keep_awake()` for button and hold-to-talk control of the compact AFE wakeup state ### Breaking Changes - `esp_capture_service_ai_audio_src_wn_cb_t` now reports wakeup start (`state=1`) and end (`state=0`) plus the wake-word index ## v0.5.0~1 ### Changes - Updated depended components `gmf_ai_audio` to ~1.1, `esp-sr` to ~2.5, `esp_audio_codec` to ~2.6 ## v0.5.0 ### Features - Initial version of `esp_audio_capture_service` - Board-aware audio recorder built on `esp_capture_service`, discovering the audio ADC through `esp_board_manager` - Automatic selection of codec-device source or AI audio front-end source when AI features are requested - Up to two output streams with independent codecs, enable state, and muxers - Fixed source sample-rate pinning for deterministic multi-stream negotiation - Reuses `esp_capture_service` runtime features: multi-stream output, storage muxers, manual record, frame pull, and service linking - Optional AI front-end features: AEC, NS, VAD, WakeNet, DOA, plus callbacks and PCM dump - Optional MCP tool support for remote setup and control - Example `audio_record` demonstrating streaming, storage, dual-stream, and AI audio cases