chat: fix and complete voice chat (STT/TTS) wiring — full endpoint URLs, live gateway key, engine defaults

The speech section shipped 6 Nov was doubly broken: LibreChat POSTs to
speech.*.openai.url VERBATIM in 0.8.8, so the base '/v1' URL 404'd, and the
STT_API_KEY/TTS_API_KEY env values held a key that no longer exists on the
gateway (401 token_not_found). Also:
- STT: tinfoil/voxtral-small-24b (token-priced; gateway bills ~$0.00008/request,
  125x under the flat $0.01/request whisper; same TEE provider, verified equal
  quality on dictation clips). Whisper remains allowed on the key as fallback.
- TTS voices: full provider speaker list (an unknown voice 400s at the gateway).
- speechTab: object form seeds Settings->Speech defaults (engineSTT/engineTTS
  external, voice casual_female) for members who never saved their own.
- new gateway key: alias librechat-speech, $5/30d, audio models only, injected
  as STT_API_KEY/TTS_API_KEY app env.

Verified on a throwaway librechat-test install then live (chat.inference.coop):
config/get serves the defaults; /api/files/speech/stt returns a correct
transcript; /api/files/speech/tts/manual returns audio. Known gap (gateway-side,
NOT this repo): aspeech rows log $0 spend — voxtral-tts is unmetered on the
gateway's aspeech path while the co-op pays ~$0.002/request.

CloudronManifest.json

@@ -1,6 +1,6 @@
 {
   "id": "ai.librechat.cloudronapp",
-    "version": "1.4.8",
+  "version": "1.4.5",
   "title": "LibreChat",
   "upstreamVersion": "0.8.8",
   "author": "Danny Avila",
@@ -55,5 +55,5 @@
     "https://screenshots.cloudron.io/ai.librechat.cloudronapp/librechat-01.png"
   ],
   "icon": "file://logo.png",
-    "dockerImage": "registry.inference.coop/co-op/librechat-custom:20261106-c2aec74"
+  "dockerImage": "registry.inference.coop/co-op/librechat-custom:20261031-2a98892"
 }

Dockerfile

@@ -20,7 +20,7 @@
 #      the manifest's scheduler task). All other behavior lives on the
 #      persistent volume (/app/data/librechat.yaml, /app/data/env,
 #      /app/data/mcp/) and survives image updates unchanged.
-FROM cloudron/ai.librechat.cloudronapp:202611030321030000
+FROM cloudron/ai.librechat.cloudronapp:202610230221180000
 
 # Login-page + header logo (the client bundle references assets/logo.svg)
 COPY assets/logo.svg /app/code/librechat/client/dist/assets/logo.svg

librechat.yaml

@@ -166,12 +166,6 @@ speech:
     advancedMode: false
     speechToText:
       engineSTT: external
-      # Pin the transcription LANGUAGE default to English. Without it the
-      # client sends no language param and voxtral-small-24b auto-detects —
-      # which misfired once as Spanish on a short first mic press (reported
-      # 2026-11-07). Members who save their own language in Settings -> Speech
-      # always override this default.
-      languageSTT: "en"
     textToSpeech:
       engineTTS: external
       voice: casual_female

release.sh

@@ -28,7 +28,7 @@ set -euo pipefail
 APP=${APP:-4023ccbf-d57a-4500-97df-14de9cf3516e}
 REGISTRY_REPO=${REGISTRY_REPO:-registry.inference.coop/co-op/librechat-custom}
 SERVER=${SERVER:-my.inference.coop}
-BASE_IMAGE=${BASE_IMAGE:-cloudron/ai.librechat.cloudronapp:202611030321030000}
+BASE_IMAGE=${BASE_IMAGE:-cloudron/ai.librechat.cloudronapp:202610150224320000}
 MANIFEST=CloudronManifest.json
 
 die() { printf 'release.sh: %s\n' "$*" >&2; exit 1; }