OpenClaw v2026.4.25: The voice reply adds 6 TTS providers in one go, and the plugin no longer performs full scans on startup

robot
Abstract generation in progress

AIMPACT announcement, April 27 (UTC+8). According to Beating monitoring by 动察, open-source personal AI assistant OpenClaw released v2026.4.25, with two biggest changes: voice replies upgraded from basic functionality to a complete TTS platform, and the plugin system migrated from full-disk scanning to a cold, persistent registry. In terms of voice replies, new commands /tts latest lets you read aloud the latest reply, and /tts chat on|off|default can toggle automatic voice output within the current session. New TTS persona (role-based voice) supports overriding global voice settings separately by agent and by account. Providers were expanded by six at once: Azure Speech, Xiaomi MiMo, local CLI TTS, Inworld, Volcano Engine Seed Speech, and ElevenLabs v3. Voice message output on channels such as WhatsApp and Feishu has also been adapted accordingly.

This release includes architectural changes to the plugin system. Previously, every startup required scanning all plugin manifest files to discover providers and capabilities. Now, installing metadata, capability queries, and provider discovery all go through a cold persistent registry (a pre-indexed read-only snapshot). \openclaw plugins list\ reads the snapshot by default and no longer triggers runtime module loading; \openclaw doctor --fix\ can refresh the index on demand. The direct result is that CLI commands like models list --all --provider run significantly faster in environments with many plugins.

Other updates: OpenTelemetry observability has been extended to model calls, token usage, tool loops, exec processes, context assembly, and memory pressure, adding Prometheus scraping routes and GenAI semantic specification metrics; browser automation adds iframe role snapshot awareness, CDP ready timeout tuning, and browser doctor --deep deep diagnostics; Control UI supports PWA installation and Web Push notifications; on first launch, it adds a fully interactive TUI onboarding (Crestodian) and a context mode selector. More than 100 bug fixes were made, covering Windows/macOS/Linux/Docker installation, WhatsApp linked-device keep-alive, session reset, log rotation, and device pairing security, among others. (Source: BlockBeats)

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned