From ca346cc6e77d1eb34eb47c4b9aacf6c0c2522402 Mon Sep 17 00:00:00 2001 From: Adolfo Reyna Date: Sun, 9 Aug 2026 00:08:32 -0400 Subject: [PATCH] record safe voice prototype status --- projects/family_voice_context_and_speaker_id_2026.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/projects/family_voice_context_and_speaker_id_2026.md b/projects/family_voice_context_and_speaker_id_2026.md index dc54922..3de9a9e 100644 --- a/projects/family_voice_context_and_speaker_id_2026.md +++ b/projects/family_voice_context_and_speaker_id_2026.md @@ -54,3 +54,7 @@ ESP32 mic -> energy/adaptive VAD + pre-roll -> LAN clip upload ## Next action Run the Adolfo-only enrollment/scoring prototype against the clearest available local recordings, using conservative thresholds and an explicit unknown state. Do not deploy always-listening capture or modify ESP32 firmware until the offline scoring behavior is measured. + +## Prototype started 2026-08-08 + +Local project: `~/Projects/voice-speaker-id`. It uses a pretrained SpeechBrain ECAPA speaker encoder, manual clip review, and offline CLI scoring. The safe first test requires explicit approval of clips before enrollment. No HTTP listener, microphone capture, Pipecat integration, ESP32 firmware change, or automatic brain ingestion is enabled.