Alan2Z/dsh-speak
Voice-announce the final reply on Windows (SAPI5 natural voices) and macOS (system voice); skips reasoning and tool calls, one-line npm install.
dsh-speak reads the final assistant reply aloud through system speech synthesis when a long task completes — no more staring at the screen waiting for the agent. On Windows, uses SAPI5 with Windows 11 built-in natural voices or NaturalVoiceSAPIAdapter (Windows 10). On macOS, uses the built-in say command and can follow a Siri natural voice. Structured so any Cordis-compatible harness can plug in.
Install
dsh plugin --profile web add dsh-speaknpm package dsh-speak 1.5.0 (registry-verified 2026-08-24). On Windows (PowerShell): dsh plugin --profile web add dsh-speak or npm install --prefix "$env:USERPROFILE\.dsh\profiles\web" dsh-speak. On macOS (bash): npm install --prefix "$HOME/.dsh/profiles/web" dsh-speak. After npm install, append to ~/.dsh/profiles/web/cordis.patch.yml: - insert:\n - id: speech-hook\n name: 'dsh-speak'. Then restart DSH web. On Windows 10, install NaturalVoiceSAPIAdapter for higher-quality voices; Windows 11 built-in voices are sufficient.
Compatibility
Windows (SAPI5 natural voices; Windows 11 built-in or NaturalVoiceSAPIAdapter on Windows 10) and macOS (say command, can follow Siri natural voices). Any harness that supports Cordis plugins. Project status: stable, not updated further (existing feature set is complete and verified).
Details
- Repo: Alan2Z/dsh-speak
- Category: Notifications & Integrations
- Stars: 5
- Version: npm dsh-speak 1.5.0 (registry-verified 2026-08-24)
- Last push: 2026-08-16
- First seen: 2026-08-14
Recent updates
1.5.0: macOS say (Siri natural voice aware); Windows SAPI5 (Windows 11 built-in voices; NaturalVoiceSAPIAdapter fallback for Windows 10); graceful fallback to stock voices; Cordis plugin structure; project marked stable (no further updates planned).
FAQ
- Does this work on macOS?
- Yes — on macOS the plugin uses the built-in say command. It can follow a Siri natural voice if one is configured in System Settings → Accessibility → Spoken Content.
- How do I get natural-sounding voices on Windows 10?
- Install NaturalVoiceSAPIAdapter (linked in the README) to make Windows 10 use neural TTS voices via SAPI5. On Windows 11, the built-in natural voices work without any extra installation.
- Which agent reply does it read aloud?
- Only the final assistant reply is read, not intermediate tool calls or streaming text. This means it announces when the agent is done working, which is the primary use case — step away and come back when you hear the announcement.
Alternatives
1624318455/dsh-plugin-tts · omdsh-dev/dsh-notification · haytham818/dsh-notify