Create Voice Synthesie pluging with playht service

livekit-plugins/livekit-plugins-playht/README.md

@@ -0,0 +1,13 @@
+# LiveKit Plugins PlayHT
+
+Agent Framework plugin for voice synthesis with [PlayHT](https://play.ht/) API.
+
+## Installation
+
+```bash
+pip install livekit-plugins-playht
+```
+
+## Pre-requisites
+
+You'll need USER ID and API Secret KEY from PlayHT. It can be set as an environment variable: `PLAYHT_USER_ID`, `PLAYHT_API_KEY`

livekit-plugins/livekit-plugins-playht/livekit/plugins/playht/__init__.py

@@ -0,0 +1,24 @@
+
+from .models import TTSEngines
+from .tts import DEFAULT_VOICE, TTS, Voice
+from .version import __version__
+
+__all__ = [
+    "TTS",
+    "Voice",
+    "DEFAULT_VOICE",
+    "TTSEngines",
+    "__version__",
+]
+
+from livekit.agents import Plugin
+
+
+class PlayHTPlugin(Plugin):
+    def __init__(self) -> None:
+        super().__init__(__name__, __version__, __package__)
+
+    def download_files(self) -> None:
+        self.download_files(self)
+
+Plugin.register_plugin(PlayHTPlugin())

livekit-plugins/livekit-plugins-playht/livekit/plugins/playht/log.py

@@ -0,0 +1,3 @@
+import logging
+
+logger = logging.getLogger("livekit.custom_tts_plugins.playht")

livekit-plugins/livekit-plugins-playht/livekit/plugins/playht/models.py

@@ -0,0 +1,7 @@
+from typing import Literal
+
+TTSEngines = Literal[
+    'PlayHT2.0',
+    'PlayHT1.0',
+    'PlayHT2.0-turbo'
+]

livekit-plugins/livekit-plugins-playht/livekit/plugins/playht/tts.py

@@ -0,0 +1,183 @@
+from __future__ import annotations
+
+import asyncio
+import base64
+import dataclasses
+import json
+import os
+import io
+from dataclasses import dataclass
+from typing import Any, List, Literal
+from pyht import Client, TTSOptions, Format
+
+import aiohttp
+from livekit.agents import tts, utils, tokenize
+from livekit import rtc
+
+from .log import logger
+from .models import TTSEncoding, TTSEngines
+
+_Encoding = Literal["mp3", "pcm"]
+
+
+def _sample_rate_from_format(output_format: TTSEncoding) -> int:
+    split = output_format.split("_") 
+    return int(split[1])
+
+
+def _encoding_from_format(output_format: TTSEncoding) -> _Encoding:
+    if output_format.startswith("mp3"):
+        return "mp3"
+    elif output_format.startswith("pcm"):
+        return "pcm"
+
+    raise ValueError(f"Unknown format: {output_format}")
+
+
+@dataclass
+class Voice:
+    id: str
+    name: str
+    voice_engine: TTSEngines
+
+
+DEFAULT_VOICE = Voice(
+    id="s3://peregrine-voices/mel22/manifest.json",
+    name="Will",
+    voice_engine="PlayHT2.0"
+)
+
+API_BASE_URL_V1 = "https://api.play.ht/api/v2"
+AUTHORIZATION_HEADER = "AUTHORIZATION"
+USERID_HEADER = "X-USER-ID"
+PLAYHT_TTS_SAMPLE_RATE = 24000
+PLAYHT_TTS_CHANNELS = 1
+
+
+@dataclass
+class _TTSOptions:
+    api_key: str
+    user_id: str
+    voice: Voice
+    base_url: str
+    sample_rate: int
+
+
+class TTS(tts.TTS):
+    def __init__(
+            self,
+            *,
+            voice: Voice = DEFAULT_VOICE,
+            api_key: str | None = None,
+            user_id: str | None = None,
+            base_url: str | None = None,
+            http_session: aiohttp.ClientSession | None = None,
+    ) -> None:
+        super().__init__(
+            capabilities=tts.TTSCapabilities(
+                streaming=False,
+            ),
+            sample_rate=PLAYHT_TTS_SAMPLE_RATE,
+            num_channels=PLAYHT_TTS_CHANNELS,
+        )
+        api_key = api_key or os.environ.get("PLAYHT_API_KEY")
+        if not api_key:
+            raise ValueError("PLAYHT_API_KEY must be set")
+
+        user_id = user_id or os.environ.get("PLAYHT_USER_ID")
+        if not user_id:
+            raise ValueError("PLAYHT_USER_ID mus be set")
+
+        self._opts = _TTSOptions(
+            voice=voice,
+            user_id=user_id,
+            api_key=api_key,
+            base_url=base_url or API_BASE_URL_V1,
+            sample_rate=self.sample_rate,
+        )
+        self._session = http_session
+
+    def _ensure_session(self) -> aiohttp.ClientSession:
+        if not self._session:
+            self._session = utils.http_context.http_session()
+
+        return self._session
+
+    async def list_voices(self) -> List[Voice]:
+        async with self._ensure_session().get(
+                f"{self._opts.base_url}/voices",
+                headers={
+                    "accept": "application/json",
+                    AUTHORIZATION_HEADER: self._opts.api_key,
+                    USERID_HEADER: self._opts.user_id
+                },
+        ) as resp:
+            return _dict_to_voices_list(await resp.json())
+
+    def synthesize(self, text: str) -> "ChunkedStream":
+        return ChunkedStream(text, self._opts, self._ensure_session())
+
+
+class ChunkedStream(tts.ChunkedStream):
+    """Synthesize using the chunked api endpoint"""
+
+    def __init__(
+            self, text: str, opts: _TTSOptions, session: aiohttp.ClientSession
+    ) -> None:
+        super().__init__()
+        self._text, self._opts, self._session = text, opts, session
+
+    @utils.log_exceptions(logger=logger)
+    async def _main_task(self) -> None:
+        stream = utils.audio.AudioByteStream(
+            sample_rate=self._opts.sample_rate, num_channels=1
+        )
+        request_id = utils.shortuuid()
+        segment_id = utils.shortuuid()
+        parent_path = os.path.dirname(os.path.abspath(__file__))
+        client = Client(self._opts.user_id, self._opts.api_key)
+        options = TTSOptions(
+            voice=self._opts.voice.id,
+            sample_rate=PLAYHT_TTS_SAMPLE_RATE,
+            format=Format.FORMAT_WAV,
+            speed=1
+        )
+
+        try:
+            response = client.tts(text=self._text, voice_engine=self._opts.voice.voice_engine, options=options)
+            audio_buffer = io.BytesIO()
+
+            for bytes_data in response:
+                for frame in stream.write(bytes_data):
+                    self._event_ch.send_nowait(
+                        tts.SynthesizedAudio(
+                            request_id=request_id,
+                            segment_id=segment_id,
+                            frame=frame,
+                        )
+                    )
+
+                audio_buffer.write(bytes_data)
+
+            for frame in stream.flush():
+                self._event_ch.send_nowait(
+                    tts.SynthesizedAudio(
+                        request_id=request_id, segment_id=segment_id, frame=frame
+                    )
+                )
+        except Exception as e:
+            print(e)
+
+
+def _dict_to_voices_list(data: dict[str, Any]):
+    voices: List[Voice] = []
+    for voice in data["text"]:
+        voices.append(
+            Voice(
+                id=voice["id"],
+                name=voice["name"],
+                voice_engine=voice["voice_engine"]
+            )
+        )
+    return voices
+

livekit-plugins/livekit-plugins-playht/livekit/plugins/playht/version.py

@@ -0,0 +1 @@
+__version__ = "1.0.0"

livekit-plugins/livekit-plugins-playht/package.json

@@ -0,0 +1,6 @@
+{
+    "name": "livekit-plugins-playht",
+    "private": true,
+    "version": "1.0.0"
+  }
+

livekit-plugins/livekit-plugins-playht/pyproject.toml

@@ -0,0 +1,3 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"

livekit-plugins/livekit-plugins-playht/setup.py

@@ -0,0 +1,44 @@
+
+import os
+import pathlib
+
+import setuptools
+import setuptools.command.build_py
+
+here = pathlib.Path(__file__).parent.resolve()
+about = {}
+with open(
+    os.path.join(here, "livekit", "plugins", "playht", "version.py"), "r"
+) as f:
+    exec(f.read(), about)
+
+
+setuptools.setup(
+    name="livekit-plugins-playht",
+    version=about["__version__"],
+    description="Agent Framework plugin for voice synthesis with PlayHT's API.",
+    long_description=(here / "README.md").read_text(encoding="utf-8"),
+    long_description_content_type="text/markdown",
+    url="https://github.com/livekit/agents",
+    cmdclass={},
+    classifiers=[
+        "Intended Audience :: Developers",
+        "Topic :: Multimedia :: Sound/Audio",
+        "Topic :: Scientific/Engineering :: Artificial Intelligence",
+        "Programming Language :: Python :: 3",
+        "Programming Language :: Python :: 3.11",
+        "Programming Language :: Python :: 3.12",
+        "Programming Language :: Python :: 3 :: Only",
+    ],
+    keywords=["webrtc", "realtime", "audio", "livekit", "playHT"],
+    license="Apache-2.0",
+    packages=setuptools.find_namespace_packages(include=["livekit.*"]),
+    python_requires=">=3.9.0",
+    install_requires=["livekit-agents[codecs]>=0.8.0.dev0", "pyht", "aiohttp", "livekit"],
+    package_data={"livekit.plugins.playht": ["py.typed"]},
+    project_urls={
+        "Documentation": "https://docs.livekit.io",
+        "Website": "https://livekit.io/",
+        "Source": "https://github.com/livekit/agents",
+    },
+)