Real-time voices with a human cadence

import os
from livekit.agents import AgentSession
from livekit_cadence import TTS

session = AgentSession(
    stt=...,  # your speech-to-text
    llm=...,  # your LLM
    tts=TTS(
        api_key=os.environ["CADENCE_API_KEY"],
        voice="siya_lively_best",
    ),
)

import os
from livekit.plugins import openai

# No plugin to install: LiveKit's OpenAI TTS works against our API
tts = openai.TTS(
    base_url="https://api.trycomsync.com/v1",
    api_key=os.environ["CADENCE_API_KEY"],
    voice="kavita_lib",
    response_format="pcm",
)

curl https://api.trycomsync.com/v1/audio/speech \
  -H "Authorization: Bearer $CADENCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input": "नमस्ते! आपका order कल deliver होगा।", "voice": "neha_lib"}' \
  -o hello.mp3

import asyncio, json, os, websockets

async def speak():
    url = "wss://api.trycomsync.com/v1/stream"
    headers = {"Authorization": f"Bearer {os.environ['CADENCE_API_KEY']}"}
    async with websockets.connect(url, additional_headers=headers) as ws:
        await ws.recv()  # ready
        for token in ["नमस्ते, ", "आपका order ", "कल deliver होगा।"]:
            await ws.send(json.dumps({"type": "text", "text": token}))
        await ws.send(json.dumps({"type": "flush"}))
        while True:
            msg = await ws.recv()
            if isinstance(msg, bytes):
                play(msg)  # 24 kHz mono 16-bit PCM
            elif json.loads(msg)["type"] == "done":
                break

asyncio.run(speak())