Dustin Newkirk hobokenchicken
0 Followers · 1 Following
Repositories
9
 Projects Packages Public Activity Starred Repositories
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 19:23:59 +00:00
0e74a16b40 fix(stt): revert to reliable REST gpt-4o-transcribe + MediaRecorder full-blob (Realtime WS not accessible on key)
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 19:16:10 +00:00
188da1d52a fix(stt): try gpt-4o-realtime-preview as base session model + gpt-realtime-whisper for input_audio_transcription (per OpenAI error guidance)
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 19:14:28 +00:00
191b7ad9b5 fix(stt): correct Realtime WS model to gpt-realtime-whisper + enhance event handling for deltas/completed
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:26:20 +00:00
7502f201c7 feat: Realtime WebSocket STT via gpt-realtime-whisper
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:23:43 +00:00
25b12ee14f fix: gpt-realtime-whisper requires Realtime API, not REST endpoint
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:22:20 +00:00
3128f69e48 fix: switch TTS voice from nova to sage
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:20:43 +00:00
f98f87b7ee fix: swap to gpt-realtime-whisper for STT
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:17:57 +00:00
9cd183a83b fix: streaming TTS via with_streaming_response
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:14:04 +00:00
2cd5636ad6 debug: add per-step timing logs to identify latency bottleneck
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:11:17 +00:00
7875b5d12a fix: cache Honcho memory context per-session (not per-turn)
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 18:04:46 +00:00
c5cc4dd480 fix: replace PCM16 capture with MediaRecorder (Opus/webm)
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:59:06 +00:00
537ddcd841 fix: play Opus TTS audio directly instead of WAV-converting it
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:57:07 +00:00
a370f1ebff fix: use max_completion_tokens for gpt-5.4-nano
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:55:08 +00:00
a19ac46312 fix: wrap PCM16 in WAV container before STT API call
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:51:37 +00:00
f2a5416408 feat: cheapest pipeline — gpt-4o-mini-transcribe + gpt-5.4-nano + TTS
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:49:58 +00:00
66e799a655 fix: connect directly via websockets to bypass OpenAI-Beta header
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:48:08 +00:00
274d04ea10 feat: hybrid pipeline — gpt-realtime-whisper + gpt-5.4-nano + TTS
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:42:09 +00:00
1c15d42e06 fix: remove OpenAI-Beta header, use gpt-realtime-2 GA model
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 17:32:40 +00:00
e2332af8d0 feat: OpenAI Realtime API pipeline
hobokenchicken pushed to main at hobokenchicken/kira 2026-06-04 16:12:09 +00:00
e64698b0ab fix: graceful mic-unavailable handling over HTTP