Bind Frame right X to normal-priority hold-to-talk

CMakeLists.txt

@@ -138,9 +138,11 @@ if(BUILD_TESTING AND NOT CMAKE_CROSSCOMPILING)
     endif()
   endif()
   if(Python3_FOUND)
+    add_test(NAME frameyap.action_bindings COMMAND "${Python3_EXECUTABLE}" -m unittest discover -s tests -p test_action_bindings.py)
     add_test(NAME frameyap.python_worker COMMAND "${Python3_EXECUTABLE}" -m unittest discover -s tests -p test_worker.py)
     add_test(NAME frameyap.installer COMMAND "${Python3_EXECUTABLE}" -m unittest discover -s tests -p test_installer.py)
-    set_tests_properties(frameyap.python_worker frameyap.installer PROPERTIES WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" TIMEOUT 60)
+    set_tests_properties(frameyap.action_bindings frameyap.python_worker frameyap.installer
+      PROPERTIES WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}" TIMEOUT 60)
     if(TARGET frameyap_worker)
       add_executable(frameyap_worker_test tests/worker_test.cpp)
       target_link_libraries(frameyap_worker_test PRIVATE frameyap_worker)

README.md

@@ -2,7 +2,7 @@
 
 Standalone, on-device voice typing POC for Steam Frame. **MIT licensed.**
 
-Implemented: native OpenVR overlay, remappable grip controls, bounded SDL3 capture,
+Implemented: native OpenVR overlay, remappable controller actions, bounded SDL3 capture,
 persistent local Parakeet Redux worker, preview/explicit insertion through Gamescope,
 and an idempotent user-local installer. No Steam store AppID, sudo, desktop ASR
 server, cloud fallback or unrelated application dependency.
@@ -19,9 +19,10 @@ See [third-party notes](docs/third-party.md). No GitHub release is published yet
 
 ## Controls
 
-- **Right grip:** short tap, then hold the second squeeze to record; release to
+- **Right X (default Frame binding):** hold to record; release to
-  transcribe. Remappable through SteamVR bindings; separate hold-to-talk action too.
+  transcribe. This PTT action is remappable through SteamVR bindings; delivery
-- **Left grip:** double-tap to explicitly send Enter. Never inferred from speech.
+  in the dashboard still requires an on-device test.
+- **Left grip (when active):** double-tap to explicitly send Enter. Never inferred from speech.
 - **Overlay:** Record/Stop, Cancel, paginated preview, Insert, Enter and Quit.
   Review and settings share one Inconsolata/neon-framed surface. World-space by
   default; settings offer left wrist, right wrist and head mounting, plus recenter.

assets/actions.json

@@ -19,7 +19,7 @@
     "/actions/frameyap": "FrameYap controls",
     "/actions/frameyap/in/left_grip": "Double tap left grip: Enter (enabled only)",
     "/actions/frameyap/in/right_grip": "Tap right grip briefly, then squeeze and hold to record; release to finish",
-    "/actions/frameyap/in/ptt": "Hold to record",
+    "/actions/frameyap/in/ptt": "Hold right X to record (default; remappable)",
     "/actions/frameyap/in/cancel": "Cancel recording or discard",
     "/actions/frameyap/in/insert": "Insert reviewed text",
     "/actions/frameyap/in/enter": "Explicit Enter"

assets/bindings_frame_controller.json

@@ -1,7 +1,7 @@
 {
   "controller_type": "frame_controller",
-  "description": "Right grip: tap then hold to dictate. Left grip: double tap Enter. Remap other actions in SteamVR.",
+  "description": "Hold right X to dictate; release to review. Grip gestures remain remappable in SteamVR.",
-  "name": "FrameYap Steam Frame grips",
+  "name": "FrameYap Steam Frame controls",
   "bindings": {
     "/actions/frameyap": {
       "sources": [
@@ -14,6 +14,11 @@
           "path": "/user/hand/right/input/grip",
           "mode": "button",
           "inputs": { "click": { "output": "/actions/frameyap/in/right_grip" } }
+        },
+        {
+          "path": "/user/hand/right/input/x",
+          "mode": "button",
+          "inputs": { "click": { "output": "/actions/frameyap/in/ptt" } }
         }
       ]
     }

docs/overlay.md

@@ -92,9 +92,11 @@ texture replacements; it does not establish that ordinary state-changing UI
 updates are flicker-free.
 
 `assets/actions.json` names six actions: left/right grip, PTT, cancel, insert,
-Enter. `bindings_frame_controller.json` uses the observed Frame profile's grip
+Enter. `bindings_frame_controller.json` maps right X click to hold-to-talk PTT;
-click paths; both bound actions were reported tracked/active in the device check.
+the grip bindings remain for optional remapping/diagnosis. Both grip actions
-This does not prove physical gesture delivery. `bindings_knuckles.json` is an
+were **inactive** during the dashboard controls check despite a loaded binding
+file; this does not prove whether right X will deliver until an on-device test.
+`bindings_knuckles.json` is an
 additional **Index/knuckles example only**. Collisions with scene actions require
 separate on-device validation. Left grip double tap
 (releases <=250 ms, second press within 350 ms) requests explicit Enter only

docs/poc.md

@@ -7,10 +7,13 @@ initialize OpenVR, open a microphone, run ASR, download files or inject input.
 
 - Opt-in OpenVR RGBA overlay with world-space default and selectable wrist/head
   mounts, status, recording timer, paginated UTF-8 preview and explicit controls.
-- Remappable SteamVR actions. Default Steam Frame grip bindings use the observed
+- Remappable SteamVR actions. The default Steam Frame binding maps right X
-  `frame_controller` profile. Right: short tap, then hold the second squeeze to
+  (hold to record, release to transcribe) to the existing PTT action using the
-  record; release to transcribe. Left: two short taps request explicit Enter.
+  observed `frame_controller` profile. Grip bindings remain, but both grip
-  First squeeze <=250 ms; second squeeze begins <=350 ms after first release.
+  actions were inactive in the observed dashboard check; do not rely on them.
+  If left grip becomes active, two short taps request explicit Enter.
+  For grip gestures, first squeeze <=250 ms; second squeeze begins <=350 ms
+  after first release.
   Activity/tracking loss cancels a held recording and requires neutral rearm.
 - SDL3 default recording device, mono float32 conversion at 16 kHz, 200 ms minimum,
   20 second maximum. Microphone is closed outside actual capture. Other apps may
@@ -90,10 +93,11 @@ That is a producer workflow, not an end-user compiler requirement.
 
 The first acquires/releases an IME without text/actions. The second displays a
 five-second inert panel. The third displays a 30-second diagnostic panel and
-reports pointer actions, SteamVR action activity, tracking and read-only legacy
+reports pointer actions, SteamVR grip and PTT action activity, tracking and
-grip state, **without microphone or input injection**. Mount clicks in check
+read-only legacy grip state, **without microphone or input injection**. Mount
-modes do not save a preference. A SteamVR error code or inactive grip means the
+clicks in check modes do not save a preference. A SteamVR error code or inactive
-gesture cannot be accepted; raw grip reads do not authorize a fallback binding.
+action means the gesture cannot be accepted; raw grip reads do not authorize a
+fallback binding.
 Action handles alone are not proof that gestures were delivered. Checks must be explicitly launched
 while the user expects the panel. Normal CLI/help/version remain inert.
 

src/main.cpp

@@ -24,8 +24,9 @@ void help() {
                  "Mount: world (first-launch default), left-wrist, right-wrist, head.\n"
                  "Settings save the mount; --mount overrides it for this launch. --head is an alias.\n"
                  "Font defaults to bundled Inconsolata in the assets directory.\n"
-                 "Right grip: double-tap, hold second squeeze to speak, release to review.\n"
+                 "Right X: hold to speak, release to review (default Frame binding).\n"
-                 "Left grip: double-tap for explicit Enter. Bindings are remappable.\n"
+                 "Grip gestures are remappable but may be unavailable in the dashboard.\n"
+                 "Left grip: double-tap for explicit Enter when active.\n"
                  "Review first: Insert approves current focus; never automatic Enter.\n"
                  "No device access unless an explicit runtime/check/registration mode is used.\n";
 #ifndef FRAMEYAP_NATIVE

src/overlay.cpp

@@ -341,6 +341,15 @@ std::string Overlay::controls_status() {
             " origin=" + (origin_ok ? "Y" : "N") +
             " raw=" + (raw_ok ? ((raw.ulButtonPressed & vr::ButtonMaskFromId(vr::k_EButton_Grip)) ? "down" : "up") : "n/a");
     }
+    // The default Frame binding maps right X to the named hold-to-talk action.
+    // Report the action's own activity separately from the stricter pose gate;
+    // remapped bindings may intentionally have a different origin.
+    vr::InputDigitalActionData_t ptt{};
+    const auto ptt_error = impl_->input->GetDigitalActionData(impl_->actions[2], &ptt, sizeof(ptt), vr::k_ulInvalidInputValueHandle);
+    const auto [accepted, down] = impl_->digital(2);
+    result += "\nPTT (default right X): err=" + std::to_string(int(ptt_error)) +
+              " active=" + (ptt.bActive ? "Y" : "N") + " down=" + (ptt.bState ? "Y" : "N") +
+              " accepted=" + (accepted ? "Y" : "N") + " gated-down=" + (down ? "Y" : "N");
     return result;
 }
 std::string Overlay::pointer_status() const {

src/runtime.cpp

@@ -24,7 +24,7 @@ void signal_stop(int) { interrupted = 1; }
 std::string state_label(State state) {
     switch (state) {
     case State::Warming: return "Warming - on-device Redux CPU";
-    case State::Ready: return "Ready - double-tap + hold right grip";
+    case State::Ready: return "Ready - hold X or click Record";
     case State::Recording: return "RECORDING";
     case State::Transcribing: return "Transcribing - on device";
     case State::Review: return "Review - Insert approves CURRENT focus";
@@ -66,7 +66,7 @@ int run(const Options& options) {
         if (session.state() == State::Error) session.cancel();
         if (!session.record()) return;
         audio.start();
-        detail = "Release grip to finish. Cancel discards. Maximum 20 seconds.";
+        detail = "Release to finish. Cancel discards. Maximum 20 seconds.";
     };
     try { warm(); }
     catch (const std::exception& e) { session.fail(); detail = e.what(); }

tests/test_action_bindings.py

@@ -0,0 +1,38 @@
+"""Offline contract for the default Frame controller shortcut (no OpenVR)."""
+import json
+from pathlib import Path
+import unittest
+
+
+ROOT = Path(__file__).resolve().parents[1]
+
+
+class ActionBindingTests(unittest.TestCase):
+    def test_right_x_is_hold_to_talk(self):
+        manifest = json.loads((ROOT / "assets/actions.json").read_text())
+        binding = json.loads((ROOT / "assets/bindings_frame_controller.json").read_text())
+        self.assertIn(
+            {"controller_type": "frame_controller", "binding_url": "bindings_frame_controller.json"},
+            manifest["default_bindings"],
+        )
+        self.assertIn(
+            {"name": "/actions/frameyap/in/ptt", "type": "boolean"},
+            manifest["actions"],
+        )
+        sources = binding["bindings"]["/actions/frameyap"]["sources"]
+        right_x = [source for source in sources if source["path"] == "/user/hand/right/input/x"]
+        self.assertEqual(right_x, [{
+            "path": "/user/hand/right/input/x",
+            "mode": "button",
+            "inputs": {"click": {"output": "/actions/frameyap/in/ptt"}},
+        }])
+        # Do not map a second PTT source or replace the dashboard's pointer trigger.
+        self.assertEqual(sum(
+            source["inputs"].get("click", {}).get("output") == "/actions/frameyap/in/ptt"
+            for source in sources
+        ), 1)
+        self.assertFalse(any(source["path"].endswith("/input/trigger") for source in sources))
+
+
+if __name__ == "__main__":
+    unittest.main()