docs: webui now separate service; add Grigori MiniMax alias

README.md

@@ -768,7 +768,7 @@ Display the current configuration as an embed:
 
 # TtsToy
 
-Multi-engine text-to-speech cog with voice cloning, emoji-triggered sound effects, inline voice switching, a TTS queue system, and a companion web UI. Plays audio in voice channels and uploads MP3 files to chat.
+Multi-engine text-to-speech cog with voice cloning, emoji-triggered sound effects, inline voice switching, a TTS queue system, and a separate web UI service (tracked in scrapyard-ttstoy-webui). Plays audio in voice channels and uploads MP3 files to chat.
 
 ### Engines
 
@@ -1119,13 +1119,13 @@ Toggle auto-start for the bundled DECTalk server. When enabled, the server start
 
 ### `[p]ttstoy webuistatus`
 
-Check the status of the TtsToy web UI subprocess (running, reachable, URL, PID).
+Check whether the TtsToy web UI is reachable. The web UI is no longer a bot subprocess; it runs as a scrapyard service. Restart with `sudo systemctl restart scrapyard`.
 
 ---
 
 ### `[p]ttstoy webuirestart`
 
-Restart the TtsToy web UI subprocess.
+The web UI is now a scrapyard service, so this just tells you to run `sudo systemctl restart scrapyard`.
 
 ---
 
@@ -1239,12 +1239,16 @@ No voice selection needed. All text is always spoken in Morshu's voice.
 
 ## Web UI
 
-TtsToy includes a companion Flask web UI for managing voices, generating TTS, and controlling settings from a browser.
+TtsToy has a companion Flask web UI, but it no longer ships inside the cog. It is a separate scrapyard service (port 8098) tracked in its own repo: `kingston/scrapyard-ttstoy-webui` on Gitea (`http://192.168.0.200:3000/kingston/scrapyard-ttstoy-webui`).
 
 - Login via `[p]ttstoy login` (DMs you a one-time key)
-- Access at the configured URL (default: `https://ttstoy.kingstons-scrapyard.net`)
+- Access at https://ttstoy.kingstons-scrapyard.net
 - Features: voice management, TTS generation, parameter tuning, SFX mapping editor
-- TTS generated from the web UI is played in the user's current voice channel and posted to the configured Discord channel
+- The web UI talks to the cog through a shared state file under `/tmp` (a cross-process `flock`): it appends `pending_posts` and `pending_commands`, and the cog's `_webui_state_processor` loop (3 s) pops them. It is not a Discord command bridge; audio is generated by the webui service itself and posted by the cog afterward.
+
+The bot cog only:
+- issues one-time login tokens (`[p]ttstoy login`)
+- consumes `pending_posts` and posts the rendered audio to the server's webui TTS channel (and plays it in VC if the user is in one)
 
 ---
 
@@ -1263,7 +1267,7 @@ TtsToy exposes a health check HTTP endpoint at port 8097:
 | BlueGnome | - |
 | Dracafow | - |
 | Gaben | - |
-| Grigori | - |
+| Grigori / Gregori | Same voice, both spellings accepted |
 | Gnome | Default voice |
 | King | - |
 | Peppa | - |

ttstoy/ttstoy.py

@@ -60,6 +60,7 @@ MINIMAX_VOICES = {
     "Dracafow": "moss_audio_13cb66d2-12c7-11f1-b6f2-729162d0a8d2",
     "Gaben": "moss_audio_5239ca93-115b-11f1-b9c4-4ea5324904c7",
     "Grigori": "moss_audio_63a397aa-24c9-11f1-918f-5a2de67f838a",
+    "Gregori": "moss_audio_63a397aa-24c9-11f1-918f-5a2de67f838a",
     "Gnome": "moss_audio_b304011b-11b2-11f1-90fe-36953c023630",
     "King": "moss_audio_39c16b2c-1158-11f1-995d-b25e2f30d3fa",
     "Peppa": "moss_audio_d656489f-114b-11f1-995d-b25e2f30d3fa",