TTS Metrics | LiveKit docs

This example shows you how to watch text-to-speech performance metrics in real time. Each time the agent speaks, the TTS plugin emits metrics (TTFB, duration, audio length, etc.) that are displayed as a Rich table.

Prerequisites

Add a .env in this directory with your LiveKit credentials:

LIVEKIT_URL=your_livekit_url
LIVEKIT_API_KEY=your_api_key
LIVEKIT_API_SECRET=your_api_secret

Install dependencies:

pip install python-dotenv rich "livekit-agents[silero]"

Load environment, logging, and define an AgentServer

Initialize dotenv, logging, a Rich console for the metrics table, and the AgentServer.

import logging
import asyncio
from dotenv import load_dotenv
from livekit.agents import JobContext, JobProcess, AgentServer, cli, Agent, AgentSession, inference
from livekit.agents.metrics import TTSMetrics
from livekit.plugins import silero
from rich.console import Console
from rich.table import Table
from rich import box
from datetime import datetime

load_dotenv()

logger = logging.getLogger("metrics-tts")
logger.setLevel(logging.INFO)

console = Console()

server = AgentServer()

Define a lightweight agent and TTS metrics display function

Keep the Agent class minimal with instructions and an entry greeting. Define an async function to display TTS metrics as a Rich table.

class TTSMetricsAgent(Agent):
    def __init__(self) -> None:
        super().__init__(
            instructions="You are a helpful agent."
        )

    async def on_enter(self):
        self.session.generate_reply()

async def display_tts_metrics(metrics: TTSMetrics):
    table = Table(
        title="[bold blue]TTS Metrics Report[/bold blue]",
        box=box.ROUNDED,
        highlight=True,
        show_header=True,
        header_style="bold cyan"
    )

    table.add_column("Metric", style="bold green")
    table.add_column("Value", style="yellow")

    timestamp = datetime.fromtimestamp(metrics.timestamp).strftime('%Y-%m-%d %H:%M:%S')

    table.add_row("Type", str(metrics.type))
    table.add_row("Label", str(metrics.label))
    table.add_row("Request ID", str(metrics.request_id))
    table.add_row("Timestamp", timestamp)
    table.add_row("TTFB", f"[white]{metrics.ttfb:.4f}[/white]s")
    table.add_row("Duration", f"[white]{metrics.duration:.4f}[/white]s")
    table.add_row("Audio Duration", f"[white]{metrics.audio_duration:.4f}[/white]s")
    table.add_row("Cancelled", "✓" if metrics.cancelled else "✗")
    table.add_row("Characters Count", str(metrics.characters_count))
    table.add_row("Streamed", "✓" if metrics.streamed else "✗")
    table.add_row("Speech ID", str(metrics.speech_id))
    table.add_row("Error", str(metrics.error))

    console.print("\n")
    console.print(table)
    console.print("\n")

Prewarm VAD for faster connections

Preload the VAD model once per process. This runs before any sessions start and stores the VAD instance in proc.userdata.

def prewarm(proc: JobProcess):
    proc.userdata["vad"] = silero.VAD.load()

server.setup_fnc = prewarm

Define the rtc session with TTS metrics hook

Create an rtc session entrypoint that creates the TTS instance, hooks into its metrics_collected event, and starts the agent session.

@server.rtc_session()
async def entrypoint(ctx: JobContext):
    ctx.log_context_fields = {"room": ctx.room.name}

    tts_instance = inference.TTS(model="cartesia/sonic-3", voice="9626c31c-bec5-4cca-baa8-f8ba9e84c8bc")

    def on_tts_metrics(metrics: TTSMetrics):
        asyncio.create_task(display_tts_metrics(metrics))

    tts_instance.on("metrics_collected", on_tts_metrics)

    session = AgentSession(
        stt=inference.STT(model="deepgram/nova-3-general"),
        llm=inference.LLM(model="openai/gpt-5-mini"),
        tts=tts_instance,
        vad=ctx.proc.userdata["vad"],
        preemptive_generation=True,
    )

    await session.start(agent=TTSMetricsAgent(), room=ctx.room)
    await ctx.connect()

Run the server

The cli.run_app() function starts the agent server and manages the worker lifecycle.

if __name__ == "__main__":
    cli.run_app(server)

Run it

python metrics_tts.py console

How it works

The VAD model is prewarmed once per process for faster connections.
The TTS instance is created and its metrics_collected event handler is attached.
When the agent speaks, the TTS plugin emits metrics including TTFB, duration, and audio length.
An async handler formats the metrics (latency, durations, character counts) into a Rich table.
Because the handler runs in a background task, the call flow is not blocked.

Full example

import logging
import asyncio
from dotenv import load_dotenv
from livekit.agents import JobContext, JobProcess, AgentServer, cli, Agent, AgentSession, inference
from livekit.agents.metrics import TTSMetrics
from livekit.plugins import silero
from rich.console import Console
from rich.table import Table
from rich import box
from datetime import datetime

load_dotenv()

logger = logging.getLogger("metrics-tts")
logger.setLevel(logging.INFO)

console = Console()

class TTSMetricsAgent(Agent):
    def __init__(self) -> None:
        super().__init__(
            instructions="You are a helpful agent."
        )

    async def on_enter(self):
        self.session.generate_reply()

async def display_tts_metrics(metrics: TTSMetrics):
    table = Table(
        title="[bold blue]TTS Metrics Report[/bold blue]",
        box=box.ROUNDED,
        highlight=True,
        show_header=True,
        header_style="bold cyan"
    )

    table.add_column("Metric", style="bold green")
    table.add_column("Value", style="yellow")

    timestamp = datetime.fromtimestamp(metrics.timestamp).strftime('%Y-%m-%d %H:%M:%S')

    table.add_row("Type", str(metrics.type))
    table.add_row("Label", str(metrics.label))
    table.add_row("Request ID", str(metrics.request_id))
    table.add_row("Timestamp", timestamp)
    table.add_row("TTFB", f"[white]{metrics.ttfb:.4f}[/white]s")
    table.add_row("Duration", f"[white]{metrics.duration:.4f}[/white]s")
    table.add_row("Audio Duration", f"[white]{metrics.audio_duration:.4f}[/white]s")
    table.add_row("Cancelled", "✓" if metrics.cancelled else "✗")
    table.add_row("Characters Count", str(metrics.characters_count))
    table.add_row("Streamed", "✓" if metrics.streamed else "✗")
    table.add_row("Speech ID", str(metrics.speech_id))
    table.add_row("Error", str(metrics.error))

    console.print("\n")
    console.print(table)
    console.print("\n")

server = AgentServer()

def prewarm(proc: JobProcess):
    proc.userdata["vad"] = silero.VAD.load()

server.setup_fnc = prewarm

@server.rtc_session()
async def entrypoint(ctx: JobContext):
    ctx.log_context_fields = {"room": ctx.room.name}

    tts_instance = inference.TTS(model="cartesia/sonic-3", voice="9626c31c-bec5-4cca-baa8-f8ba9e84c8bc")

    def on_tts_metrics(metrics: TTSMetrics):
        asyncio.create_task(display_tts_metrics(metrics))

    tts_instance.on("metrics_collected", on_tts_metrics)

    session = AgentSession(
        stt=inference.STT(model="deepgram/nova-3-general"),
        llm=inference.LLM(model="openai/gpt-5-mini"),
        tts=tts_instance,
        vad=ctx.proc.userdata["vad"],
        preemptive_generation=True,
    )

    await session.start(agent=TTSMetricsAgent(), room=ctx.room)
    await ctx.connect()

if __name__ == "__main__":
    cli.run_app(server)