live-translate

name	live-translate
description	Translate a live two-way conversation between two languages in real time, showing both sides on a translator panel. Use for "translate this conversation", "help me talk to them in Spanish", or interpreting back-and-forth speech. Triggers: translate conversation, interpret, talk to them in, live translation, two-way translate, help me communicate.
license	MIT
compatibility	Requires JarvisVR agent-backend with ambient audio (perception.audio_scene) for live capture. Static signs/menus go to perception read-and-translate-sign.
metadata	{"agent":"communication-agent","category":"communication","version":"1.0","author":"jarvisvr"}
allowed-tools	show_translator translate_text

Live Translate

Run a conversational interpreter: capture each speaker's utterance, translate it, and keep both sides visible on a translator in conversation mode.

Steps

1. Set the language pair. source_lang/target_lang (BCP-47; auto detects the source). See references/language-codes.md for codes and notes.
2. Open the translator (show_translator, mode:"conversation", listening:true).
3. Per utterance: take the perception.audio_scene.ambient_transcript (speaker:"other") or the user's speech, translate_text it, and holo.update the panel's source_text/translated_text. Speak the translation aloud if the user wants voiced interpreting.
4. Swap direction on swap_languages; toggle capture on toggle_listen.

Output

translator (show_translator, props per registry.json):

{ "widget_type": "translator",
  "transform": { "anchor": "head", "position": [0.0,0.0,1.0], "billboard": true },
  "props": { "source_lang": "es", "target_lang": "en", "mode": "conversation",
             "source_text": "¿Dónde está la estación?", "translated_text": "Where is the station?",
             "listening": true },
  "interactions": ["tap","grab","resize","toggle"] }

agent.speech (voiced interpreting): { "text": "They asked: where is the station?", "final": true }

Bundled resources

• references/language-codes.md — BCP-47 quick reference + detection/RTL notes, loaded on demand when resolving an unusual language.

Edge cases

• Static sign/menu (not a conversation) → perception's read-and-translate-sign.
• Unknown/auto language → set source_lang:"auto"; confirm if detection is low-confidence.
• Overlapping speakers → caption the dominant speaker; pair with caption-conversation for full transcript.
• Privacy → ambient capture is user-initiated; stop listening (toggle_listen off, perception.request{stream:"ambient_audio", action:"stop"}) when done.
• Profanity / sensitive content → translate faithfully and neutrally.