my first commit

The 'uv sync' next-step is redundant when the LiveKit CLI already ran the
install task during `lk agent init --install`. Move it into a guarded subtask
that is skipped when LIVEKIT_DEPS_INSTALLED is set (the CLI sets it on a
successful install), matching the existing status-guard pattern used for
LIVEKIT_AGENT_NAME and LIVEKIT_SANDBOX_ID.

README.md

@@ -9,10 +9,14 @@ A complete starter project for building voice AI apps with [LiveKit Agents for P
 The starter project includes:
 
 - A simple voice AI assistant, ready for extension and customization
-- A voice AI pipeline built on [LiveKit Inference](https://docs.livekit.io/agents/models/inference)
+- A voice AI pipeline built on [LiveKit Inference](https://docs.livekit.io/agents/models/inference), providing zero-configuration access to [models](https://docs.livekit.io/agents/models) from top labs
-  with [models](https://docs.livekit.io/agents/models) from OpenAI, Cartesia, and Deepgram. More than 50 other model providers are supported, including [Realtime models](https://docs.livekit.io/agents/models/realtime)
+  - Uses the fast, open-weight Gemma 4 31B model, [hosted by LiveKit](https://docs.livekit.io/agents/models/llm/livekit/) and tuned for optimal performance in voice AI, as the default LLM
+  - Uses Fish Audio S2.1 Pro for TTS, which renders the inline delivery markup that expressive mode relies on
+  - Supports more than 50 models from OpenAI, Cartesia, Deepgram, and other providers
+  - Access to a wide range of other models, including [Realtime models](https://docs.livekit.io/agents/models/realtime), through extensive plugin ecosystem
+- Expressive mode, enabled by default: the framework injects the TTS provider's markup guide into the LLM prompt, so the model emits inline delivery tags (emotion, pacing, non-verbal sounds) that the TTS renders and the transcript never shows
 - Eval suite based on the LiveKit Agents [testing & evaluation framework](https://docs.livekit.io/agents/start/testing/)
-- [LiveKit Turn Detector](https://docs.livekit.io/agents/logic/turns/turn-detector/) for contextually-aware speaker detection, with multilingual support
+- [LiveKit Turn Detector](https://docs.livekit.io/agents/logic/turns/turn-detector/), an end-of-turn model that listens to the user's audio directly, combining semantic understanding with acoustic cues for state-of-the-art accuracy across 14 languages
 - [Background voice cancellation](https://docs.livekit.io/transport/media/noise-cancellation/)
 - Deep session insights from LiveKit [Agent Observability](https://docs.livekit.io/deploy/observability/)
 - A Dockerfile ready for [production deployment to LiveKit Cloud](https://docs.livekit.io/deploy/agents/)
@@ -85,20 +89,14 @@ You can load the LiveKit environment automatically using the [LiveKit CLI](https
 
 ```bash
 lk cloud auth
-lk app env -w -d .env.local
+lk app env --write --destination .env.local
 ```
 
 </details>
 
 ## Run the agent
 
-Before your first run, you must download certain models such as [Silero VAD](https://docs.livekit.io/agents/logic/turns/vad/) and the [LiveKit turn detector](https://docs.livekit.io/agents/logic/turns/turn-detector/):
+Run this command to speak to your agent directly in your terminal:
-
-```console
-uv run python src/agent.py download-files
-```
-
-Next, run this command to speak to your agent directly in your terminal:
 
 ```console
 uv run python src/agent.py console

pyproject.toml

@@ -9,7 +9,7 @@ description = "Simple voice AI assistant built with LiveKit Agents for Python"
 requires-python = ">=3.10, <3.15"
 
 dependencies = [
-    "livekit-agents[silero,turn-detector]==1.5.16",
+    "livekit-agents>=1.6.9",
     "livekit-plugins-ai-coustics~=0.2",
     "python-dotenv",
 ]

src/agent.py

@@ -7,13 +7,12 @@ from livekit.agents import (
     AgentServer,
     AgentSession,
     JobContext,
-    JobProcess,
+    TurnHandlingOptions,
     cli,
     inference,
     room_io,
 )
-from livekit.plugins import ai_coustics, silero
+from livekit.plugins import ai_coustics
-from livekit.plugins.turn_detector.multilingual import MultilingualModel
 
 logger = logging.getLogger("agent")
 
@@ -25,7 +24,7 @@ class Assistant(Agent):
         super().__init__(
             # A Large Language Model (LLM) is your agent's brain, processing user input and generating a response
             # See all available models at https://docs.livekit.io/agents/models/llm/
-            llm=inference.LLM(model="openai/gpt-5.2-chat-latest"),
+            llm=inference.LLM(model="google/gemma-4-31b-it"),
             # To use a realtime model instead of a voice pipeline, replace the LLM
             # with a RealtimeModel and remove the STT/TTS from the AgentSession
             # (Note: This is for the OpenAI Realtime API. For other providers, see https://docs.livekit.io/agents/models/realtime/)
@@ -92,13 +91,6 @@ class Assistant(Agent):
 server = AgentServer()
 
 
-def prewarm(proc: JobProcess):
-    proc.userdata["vad"] = silero.VAD.load()
-
-
-server.setup_fnc = prewarm
-
-
 @server.rtc_session(agent_name="my-agent")
 async def my_agent(ctx: JobContext):
     # Logging setup
@@ -107,23 +99,35 @@ async def my_agent(ctx: JobContext):
         "room": ctx.room.name,
     }
 
-    # Set up a voice AI pipeline using OpenAI, Cartesia, Deepgram, and the LiveKit turn detector
+    # Set up a voice AI pipeline using AssemblyAI, Fish Audio, and the LiveKit turn detector
     session = AgentSession(
         # Speech-to-text (STT) is your agent's ears, turning the user's speech into text that the LLM can understand
         # See all available models at https://docs.livekit.io/agents/models/stt/
-        stt=inference.STT(model="deepgram/nova-3", language="multi"),
+        stt=inference.STT(model="assemblyai/universal-3-5-pro", language="en"),
         # Text-to-speech (TTS) is your agent's voice, turning the LLM's text into speech that the user can hear
         # See all available models as well as voice selections at https://docs.livekit.io/agents/models/tts/
         tts=inference.TTS(
-            model="cartesia/sonic-3", voice="9626c31c-bec5-4cca-baa8-f8ba9e84c8bc"
+            model="fishaudio/s2.1-pro", voice="fa4c9eb3dccc4806b382b40d61c6b10a"
         ),
-        # VAD and turn detection are used to determine when the user is speaking and when the agent should respond
+        turn_handling=TurnHandlingOptions(
-        # See more at https://docs.livekit.io/agents/build/turns
+            # The LiveKit turn detector determines when the user is done speaking and the agent should respond.
-        turn_detection=MultilingualModel(),
+            # TurnDetector is an end-of-turn model that listens to the user's audio directly, combining
-        vad=ctx.proc.userdata["vad"],
+            # semantic understanding with acoustic cues (intonation, pitch, rhythm) for state-of-the-art accuracy.
-        # allow the LLM to generate a response while waiting for the end of turn
+            # AgentSession supplies the required VAD automatically.
-        # See more at https://docs.livekit.io/agents/build/audio/#preemptive-generation
+            # See more at https://docs.livekit.io/agents/build/turns
-        preemptive_generation=True,
+            turn_detection=inference.TurnDetector(),
+            # Adaptive interruptions use the turn detector to tell a real interruption from a
+            # backchannel like "mhm" or "right", so the agent keeps talking through the latter.
+            interruption={"mode": "adaptive"},
+            # allow the LLM to generate a response while waiting for the end of turn
+            # See more at https://docs.livekit.io/agents/build/audio/#preemptive-generation
+            preemptive_generation={"enabled": True},
+        ),
+        # Expressive mode injects the TTS provider's markup guide into the LLM prompt, so the model
+        # emits inline delivery tags (emotion, pacing, non-verbal sounds) that the TTS renders and
+        # the transcript never shows. Requires a TTS model that supports markup, such as the Fish
+        # Audio model above.
+        expressive=True,
     )
 
     # Start the session, which initializes the voice pipeline and warms up the models

taskfile.yaml

@@ -21,6 +21,14 @@ tasks:
       - task: set_agent_name_if_present
       - task: help_open_sandbox_if_present
 
+  help_install_hint_if_needed:
+    desc: "Print the local install hint, unless the CLI already installed dependencies"
+    status:
+      # Skip when `lk agent init --install` already ran the install task.
+      - test -n "$LIVEKIT_DEPS_INSTALLED"
+    cmds:
+      - echo '{{ indent .INDENT "uv sync" }}'
+
   set_agent_name_if_present:
     status:
       - test -z "$LIVEKIT_AGENT_NAME"
@@ -42,8 +50,7 @@ tasks:
       - echo 'To try your new agent directly in your terminal:'
       - echo ''
       - echo '{{ indent .INDENT "cd" }} {{ .REL_PATH }}'
-      - echo '{{ indent .INDENT "uv sync" }}'
+      - task: help_install_hint_if_needed
-      - echo '{{ indent .INDENT "uv run" }} {{ .PYTHON_MAIN }} download-files'
       - echo '{{ indent .INDENT "uv run" }} {{ .PYTHON_MAIN }} console'
 
   help_open_web_console:
@@ -56,8 +63,7 @@ tasks:
       - echo 'To try your new agent in the web console:'
       - echo ''
       - echo '{{ indent .INDENT "cd" }} {{ .REL_PATH }}'
-      - echo '{{ indent .INDENT "uv sync" }}'
+      - task: help_install_hint_if_needed
-      - echo '{{ indent .INDENT "uv run" }} {{ .PYTHON_MAIN }} download-files'
       - echo '{{ indent .INDENT "uv run" }} {{ .PYTHON_MAIN }} dev'
       - echo ''
       - echo 'Then visit:'