Merge pull request #3 from livekit-examples/bcherry/updates

Update template

Author: bcherry

README.md

@@ -1,6 +1,6 @@
 # Voice Agent Hackathon Template
 
-Welcome to the Voice Agent Hackathon! This template contains a ready-to-use voice agent built with [AssemblyAI](https://assemblyai.com) and [Rime](https://rime.ai). All you need to run it is a [LiveKit Cloud](https://cloud.livekit.io) project.
+Welcome to the Voice Agent Hackathon! This template contains a ready-to-use voice agent built [LiveKit Agents](https://github.com/livekit/agents). All you need is a [LiveKit Cloud](https://cloud.livekit.io) project.
 
 This starter app is compatible with any [custom web/mobile frontend](https://docs.livekit.io/agents/start/frontend/) or [SIP-based telephony](https://docs.livekit.io/agents/start/telephony/).
 
@@ -46,9 +46,9 @@ Finally, open the [LiveKit Agents Playground](https://agents-playground.livekit.
 
 ## Tips for managing background noise
 
-This hackathon may be a noisy place which can make it tricky to test your agent. Here are some tips to help you:
+If you're in a noisy hackathon environment, it may be tricky to test your agent. Here are some tips to help you:
 
-1. Use headphones with a microphone and noise isolation features (such as Airpods) 
+1. Use headphones with a microphone and noise isolation features (such as AirPods Pro) 
 2. Use the LiveKit [background voice cancellation](https://docs.livekit.io/home/cloud/noise-cancellation/) model (pre-installed in this template)
 3. Turn off your microphone in the [Agents Playground](https://agents-playground.livekit.io/#cam=0&mic=1&screen=0&video=0&audio=1&chat=1&theme_color=cyan) and use text input to test your agent instead.
 
@@ -84,79 +84,8 @@ This project is production-ready and includes a working `Dockerfile`. To deploy
 
 ## Models
 
-This project uses models from AssemblyAI and Rime, as well as GPT-4o-mini from Azure OpenAI. By default, these are served through early-access to LiveKit Inference and no extra account is required.
+This project uses LiveKit Inference with models from AssemblyAI, OpenAI, and Cartesia. No extra account is required to use these models, which have been selected for their quality and performance.
 
-### AssemblyAI customization
+Many more models are available, both through LiveKit Inference and through SDK plugins for a wide variety of third-party APIs. See the [LiveKit Agents documentation](https://docs.livekit.io/agents/models/) for a full list.
 
-To customize the AssemblyAI model, while still using LiveKit Cloud, you can use the following session setup in your [agent code](https://github.com/livekit-examples/voice-agent-hackathon/blob/main/src/agent.py) instead of the version above:
-
-```python
-from livekit.agents import inference
-
-session = AgentSession(
-    stt=inference.STT(model="assemblyai", extra_kwargs={ ... })
-)
-```
-
-Refer to the [source code](https://github.com/livekit/agents/blob/main/livekit-agents/livekit/agents/inference/stt.py#L57) for available parameters (docs for LiveKit Inference are coming soon)
-
-#### AssemblyAI plugin
-
-To use your own AssemblyAI account, or access additional AssemblyAI features, use the AssemblyAI plugin:
-
-```shell
-uv add livekit-agents[assemblyai]
-```
-
-```python
-from livekit.plugins import assemblyai
-
-session = AgentSession(
-    stt=assemblyai.STT()
-)
-```
-
-Refer to the [plugin documentation](https://docs.livekit.io/agents/integrations/stt/assemblyai/) for more information.
-
-### Rime customization
-
-To use a different Rime voice, while still using LiveKit Cloud, just change the voice name after the colon in your [agent code](https://github.com/livekit-examples/voice-agent-hackathon/blob/main/src/agent.py):
-
-```python
-session = AgentSession(
-    tts="rime/arcana:andromeda"
-)
-```
-
-Refer to the [Rime voices list](https://docs.rime.ai/api-reference/voices) for more information.
-
-#### Rime plugin
-
-To use your own Rime account, or access additional features, use the Rime plugin:
-
-```shell
-uv add livekit-agents[rime]
-```
-
-```python
-from livekit.plugins import rime
-
-session = AgentSession(
-    tts=rime.TTS(model="arcana", speaker="andromeda")
-)
-```
-
-Refer to the [plugin documentation](https://docs.livekit.io/agents/integrations/tts/rime/) for more information.
-
-
-## Other large language models
-
-Refer to the [source code](https://github.com/livekit/agents/blob/main/livekit-agents/livekit/agents/inference/llm.py) for available models (LiveKit Inference docs are coming soon).
-
-```python
-session = AgentSession(
-    llm="azure/gpt-4o-mini"
-)
-```
-
-Or, use an [LLM plugin](https://docs.livekit.io/agents/integrations/llm/) for a wider range of models and more configuration options.
+Happy hacking!

pyproject.toml

@@ -5,11 +5,11 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "voice-agent-hackathon"
 version = "1.0.0"
-description = "Simple voice agent built with LiveKit, AssemblyAI, and Rime"
+description = "Simple voice agent built with LiveKit Agents"
 requires-python = ">=3.9"
 
 dependencies = [
-    "livekit-agents[turn-detector,silero]~=1.2.11",
+    "livekit-agents[turn-detector,silero]~=1.2",
     "livekit-plugins-noise-cancellation~=0.2",
     "python-dotenv",
 ]

src/agent.py

@@ -2,17 +2,13 @@
 
 from dotenv import load_dotenv
 from livekit.agents import (
-    NOT_GIVEN,
     Agent,
-    AgentFalseInterruptionEvent,
     AgentSession,
     JobContext,
     JobProcess,
-    MetricsCollectedEvent,
     RoomInputOptions,
     WorkerOptions,
     cli,
-    metrics,
 )
 from livekit.plugins import noise_cancellation, silero
 from livekit.plugins.turn_detector.multilingual import MultilingualModel
@@ -54,72 +50,18 @@ def prewarm(proc: JobProcess):
 
 
 async def entrypoint(ctx: JobContext):
-    # Logging setup
-    # Add any other context you want in all log entries here
-    ctx.log_context_fields = {
-        "room": ctx.room.name,
-    }
-
     # Set up a voice AI pipeline using OpenAI, Rime, AssemblyAI, and the LiveKit turn detector
+    # For a list of all available models, as well as configuration options see
+    # https://docs.livekit.io/agents/models/
     session = AgentSession(
-        # This starter template uses GPT-4o-mini via LiveKit Cloud.
-        # For a list of available models, see https://github.com/livekit/agents/blob/main/livekit-agents/livekit/agents/inference/llm.py
-        # Or, for a wider range of models, see plugins at https://docs.livekit.io/agents/integrations/llm/
-        llm="azure/gpt-4o-mini",
-        # This starter template uses AssemblyAI via LiveKit Cloud.
-        # To send extra parameters, use the following session setup instead of the version above:
-        # 1. add `from livekit.agents import inference` to the top of this file
-        # 2. Use the following session setup instead of the version above:
-        #     stt=inference.STT(model="assemblyai", extra_kwargs={ ... })
-        # See available configuration at https://github.com/livekit/agents/blob/main/livekit-agents/livekit/agents/inference/stt.py#L57
-        #
-        # Or to use your own AssemblyAI account:
-        # 1. Install livekit-agents[assemblyai]
-        # 2. Set ASSEMBLYAI_API_KEY in .env.local
-        # 3. Add `from livekit.plugins import assemblyai` to the top of this file
-        # 4. Use the following session setup instead of the version above
-        #     stt=assemblyai.STT()
-        # See available configuration at https://docs.livekit.io/agents/integrations/stt/assemblyai/
-        stt="assemblyai",
-        # This starter template uses Rime via LiveKit Cloud
-        # To change the voice, alter the voice name (currently "luna") after the colon.
-        # See available voices at https://docs.rime.ai/api-reference/voices
-        #
-        # Or, to use your own Rime account:
-        # 1. Install livekit-agents[rime]
-        # 2. Set RIME_API_KEY in .env.local
-        # 3. Add `from livekit.plugins import rime` to the top of this file
-        # 4. Use the following session setup instead of the version above
-        #     tts=rime.TTS(model="arcana", speaker="luna")
-        # See available configuration at https://docs.livekit.io/agents/integrations/tts/rime/
-        tts="rime/arcana:luna",
+        stt="assemblyai/universal-streaming:en",
+        llm="openai/gpt-4.1-mini",
+        tts="cartesia/sonic-2:9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
         turn_detection=MultilingualModel(),
         vad=ctx.proc.userdata["vad"],
         preemptive_generation=True,
     )
 
-    # sometimes background noise could interrupt the agent session, these are considered false positive interruptions
-    # when it's detected, you may resume the agent's speech
-    @session.on("agent_false_interruption")
-    def _on_agent_false_interruption(ev: AgentFalseInterruptionEvent):
-        logger.info("false positive interruption, resuming")
-        session.generate_reply(instructions=ev.extra_instructions or NOT_GIVEN)
-
-    # Metrics collection, to measure pipeline performance
-    # For more information, see https://docs.livekit.io/agents/build/metrics/
-    usage_collector = metrics.UsageCollector()
-
-    @session.on("metrics_collected")
-    def _on_metrics_collected(ev: MetricsCollectedEvent):
-        metrics.log_metrics(ev.metrics)
-        usage_collector.collect(ev.metrics)
-
-    async def log_usage():
-        summary = usage_collector.get_summary()
-        logger.info(f"Usage: {summary}")
-
-    ctx.add_shutdown_callback(log_usage)
-
     # Start the session, which initializes the voice pipeline and warms up the models
     await session.start(
         agent=Assistant(),
@@ -137,7 +79,7 @@ async def log_usage():
     await session.generate_reply(
         instructions="""
         The user has just finished getting their first voice agent up and running.
-        Welcome them to the Voice Agent Hackathon and wish them good luck!
+        Wish them good luck with their hackathon project!
         """,
         allow_interruptions=False,
     )

tests/test_agent.py

@@ -5,7 +5,7 @@
 
 
 def _llm() -> llm.LLM:
-    return inference.LLM(model="azure/gpt-4o-mini")
+    return inference.LLM(model="openai/gpt-4.1-mini")
 
 
 @pytest.mark.asyncio