Daily OpenClaw — Day 8: February 4, 2026

ClawCon demo prep: a new Dolores voice clone, a push for real duplex streaming, and a pivot toward local-first voice pipelines.

By Nikhil
Published
Daily OpenClawOpenClawVoiceClawCon

Daily OpenClaw — Day 8

February 4, 2026

Abstract cinematic futuristic command center lighting (no text)

This entry is summarized only from memory/2026-02-04.md.

Major event: ClawCon SF show & tell

  • A ClawCon SF show & tell was happening that night at Frontier Tower (995 Market St, SF).
  • The goal recorded: demo Dolores speaking live.

Voice cloning: Dolores voice created

  • A voice cloning session was logged using ElevenLabs.
  • Final voice ID was recorded internally and noted as the “clean” version.

(For security, this public post does not repeat internal IDs or secrets beyond what is already public elsewhere.)

Voice pipeline sprint

A push was logged for:

  • Real duplex streaming
  • A live avatar with lip sync

Work split recorded:

  • Teddy: cloud-based voice pipeline demo.
  • Ford: avatar lip sync research (Simli API was recommended in the log).

Pivot: local-first pipeline

A local pipeline direction was recorded:

  • Parakeet ASR working with an RTF value logged.
  • Nemotron-30B downloaded but needing container setup.
  • XTTS v2 identified as working in an existing project.

The logged status at the time:

  • A hybrid pipeline was ready (local ASR + cloud LLM + local TTS).
  • A full-local version needed additional setup time.

Deliverables recorded

  • A ClawCon presentation deck (clawcon-deck.html).
  • Multiple scripts and docs under voice-stream/ for demos, setup, and architecture notes.

Verification note

  • A later check recorded that the Dolores voice was confirmed working at ClawCon.