Proposal: URML (substrate-neutral robot intent) manifest declaration for Piper as the IPC-bounded speak substrate

[idoco2003]

Hi @OHF-Voice team,

Proposing a URML v0.1 capability-manifest mapping for Piper over OHF-Voice/piper1-gpl. URML (Apache-2.0) is a substrate-neutral spec for robot intent: typed primitive vocabulary + capability manifest + static validator.

Piper is the surviving open neural-TTS substrate after the Coqui shutdown and is the active successor to the archived rhasspy/piper. URML's integration shape with Piper is explicitly IPC-only: URML's reference adapter is a subprocess caller of the piper CLI, never an importer of the Python module. This preserves URML's Apache-2.0 license posture across the GPL-3.0 boundary.

This is proposal-only, part of URML's Move #12 outreach (16 RFCs covering speech / translation / robot-command-library substrates for URML's NL layer).

Full RFC with manifest mapping, five alternatives, and the IPC-boundary discussion: https://github.com/URML-MARS/URML/blob/main/docs/rfcs/0166-piper1-gpl-outreach.md

Questions worth maintainer input on:

1. IPC-boundary framing. Is "URML is a subprocess caller of the piper CLI, never embedding the Python module" the framing OHF-Voice would endorse, or is there language the project prefers for downstream integrations?
2. TTS-engine declaration shape. Is piper (with tts_runtime: piper1_gpl distinguishing the active repo) the right slug for URML's manifest, or do you have a preferred convention?
3. License-boundary declaration. Is tts_license_boundary: gpl_subprocess the right way to declare the GPL constraint? Useful as a downstream signal, or unnecessary friction?
4. Voice-model declaration. Is the voice-model slug (e.g., en_US-lessac-medium) the canonical identifier for downstream-manifest declarations?
5. Subprocess invocation mode. Is the piper CLI the canonical invocation surface, or are there preferred alternatives (Wyoming protocol via wyoming-piper, HTTP server, gRPC) for production?
6. Adapter home. URML-side subprocess wrapper in reference/speech-bridge/, contributed example in piper1-gpl/examples/, or external bridge repo?
7. Anything else.

Ido Yahalomi (URML maintainer, urml.dev, greenvh@gmail.com)

---

AI-assisted prose, maintainer-reviewed before posting (see VIBE.md). Human-only correspondence available on request.