Update voice list to the last available (#1)

Fix: Increase audio loop mute buffer to 2.5s for browser latency

Enhance Audio Loop prevention: Use audio duration tracking for precise muting

Fix: Audio Echo loop via self-mute + Silence motion tool vocalizations

Restrict Twitter tool to explicit user requests only

Implement Twitter Search fallback to User Timeline for Free Tier robustness

Allow 10 tweets and instruct prompt to summarize concisely

Fix Twitter API limits: fetch min allowed, return user max

Optimize Twitter tool: Limit tweet retrieval to 3 for Free Tier conservation

Implement real Twitter API integration using Tweepy

Add mock GetTweetsTool

Fix race condition: use task ownership check in start_up finally block

Add session debugging logs and fix start_up finally block interference

Add tool_choice:none to kickstart response to prevent tool cascades

Complete session restart rewrite with proper locking and _restarting flag

Fix session restart: add connection validity check and unit tests

Implement manual Restart Session command

Fix voice switch silence by adding kickstart message trigger

Fix syntax error and properly catch ConnectionClosed exception

Refactor session restart logic: fix task tracking and suppress closure errors

Add unit tests for realtime package and fix resampling bug

Refactor RealtimeVoiceHandler into modular package and add Grok support

feat: integrate Grok Voice API

feat: Add weather tool using Open-Meteo API

151 create fun and diverse profiles (#162)

fixing shutdown (#156)

removing head tracking until we solve the mediapipe dependency

enabling head tracking for non wireless, and tuning head trackign a little bit

fix initial setup (#147)

remove bad move (#146)

Persist key (#145)

Change key name (#144)

improve the startup with openai (#143)

62 appify the demo (#122)

Update stop function to reset to neutral position (#131)

Add wireless version support (#133)

fix(audio channels): making the play_loop and receive methods robust to audio inputs/outputs shapes (#132)

Merge pull request #126 from pollen-robotics/fix-bug-where-reachy-appears-deaf

Properly make mono

fix(camera tool): removing the temporary file creation which was not compatible with Windows. The current solution encodes the frame in the JPG format but stays in the RAM. (#123)

Merge branch 'develop' into 114-improve-transcript-accuracy

Merge pull request #111 from pollen-robotics/fix/sample_rate_consistency

Fix mypy and enhance code clarity

fix(mypy): fixing mypy errors

Fix comments

Include debounce logic and switch transcript model

Fixed BGR2RGB when sending image to gradio

chore(ruff)

fix(samplerate): removing hardcoded sample rates values and replacing with actual values from the robot, uniformizing processing (always on the receiving side), nuking librosa out

Created a V0 of the silent emotion imitator