refactor(): refactor cartesia tts (#444)

agents/examples/experimental/property.json

@@ -3009,7 +3009,7 @@
               "api_key": "${env:DEEPGRAM_API_KEY}",
               "language": "en-US",
               "model": "nova-2",
-              "sample_rate": "16000"
+              "sample_rate": 16000
             }
           },
           {
@@ -3036,10 +3036,10 @@
             "name": "cartesia_tts",
             "property": {
               "api_key": "${env:CARTESIA_API_KEY}",
-              "cartesia_version": "2024-06-10",
+              "language": "en",
               "model_id": "sonic-english",
-              "voice_id": "f9836c6e-a0bd-460e-9d3c-f7299fa60f94",
-              "sample_rate": "16000"
+              "sample_rate": 16000,
+              "voice_id": "f9836c6e-a0bd-460e-9d3c-f7299fa60f94"
             }
           },
           {
@@ -3253,7 +3253,7 @@
               "api_key": "${env:DEEPGRAM_API_KEY}",
               "language": "en-US",
               "model": "nova-2",
-              "sample_rate": "16000"
+              "sample_rate": 16000
             }
           },
           {

agents/ten_packages/extension/cartesia_tts/BUILD.gn

@@ -0,0 +1,19 @@
+#
+# This file is part of TEN Framework, an open source project.
+# Licensed under the Apache License, Version 2.0.
+# See the LICENSE file for more information.
+#
+import("//build/feature/ten_package.gni")
+
+ten_package("cartesia_tts") {
+  package_kind = "extension"
+
+  resources = [
+    "__init__.py",
+    "addon.py",
+    "extension.py",
+    "manifest.json",
+    "property.json",
+    "tests",
+  ]
+}

agents/ten_packages/extension/cartesia_tts/README.md

@@ -0,0 +1,29 @@
+# cartesia_tts
+
+<!-- brief introduction for the extension -->
+
+## Features
+
+<!-- main features introduction -->
+
+- xxx feature
+
+## API
+
+Refer to `api` definition in [manifest.json] and default values in [property.json](property.json).
+
+<!-- Additional API.md can be referred to if extra introduction needed -->
+
+## Development
+
+### Build
+
+<!-- build dependencies and steps -->
+
+### Unit test
+
+<!-- how to do unit test for the extension -->
+
+## Misc
+
+<!-- others if applicable -->

agents/ten_packages/extension/cartesia_tts/__init__.py

@@ -1,6 +1,6 @@
-from . import cartesia_tts_addon
-from .extension import EXTENSION_NAME
-from .log import logger
-
-
-logger.info(f"{EXTENSION_NAME} extension loaded")
+#
+# This file is part of TEN Framework, an open source project.
+# Licensed under the Apache License, Version 2.0.
+# See the LICENSE file for more information.
+#
+from . import addon

agents/ten_packages/extension/cartesia_tts/addon.py

@@ -0,0 +1,19 @@
+#
+# This file is part of TEN Framework, an open source project.
+# Licensed under the Apache License, Version 2.0.
+# See the LICENSE file for more information.
+#
+from ten import (
+    Addon,
+    register_addon_as_extension,
+    TenEnv,
+)
+
+
+@register_addon_as_extension("cartesia_tts")
+class CartesiaTTSExtensionAddon(Addon):
+
+    def on_create_instance(self, ten_env: TenEnv, name: str, context) -> None:
+        from .extension import CartesiaTTSExtension
+        ten_env.log_info("CartesiaTTSExtensionAddon on_create_instance")
+        ten_env.on_create_instance_done(CartesiaTTSExtension(name), context)

agents/ten_packages/extension/cartesia_tts/cartesia_tts.py

@@ -0,0 +1,42 @@
+#
+#
+# Agora Real Time Engagement
+# Created by XinHui Li in 2024.
+# Copyright (c) 2024 Agora IO. All rights reserved.
+#
+#
+
+from dataclasses import dataclass
+from typing import AsyncIterator
+from cartesia import AsyncCartesia
+
+from ten_ai_base.config import BaseConfig
+
+
+@dataclass
+class CartesiaTTSConfig(BaseConfig):
+    api_key: str = ""
+    language: str = "en"
+    model_id: str = "sonic-english"
+    request_timeout_seconds: int = 10
+    sample_rate: int = 16000
+    voice_id: str = "f9836c6e-a0bd-460e-9d3c-f7299fa60f94"
+
+class CartesiaTTS:
+    def __init__(self, config: CartesiaTTSConfig) -> None:
+        self.config = config
+        self.client = AsyncCartesia(api_key=config.api_key, timeout=config.request_timeout_seconds)
+
+    def text_to_speech_stream(self, text: str) -> AsyncIterator[bytes]:
+        return self.client.tts.sse(
+            language=self.config.language,
+            model_id=self.config.model_id,
+            output_format={
+                "container": "raw",
+                "encoding": "pcm_s16le",
+                "sample_rate": self.config.sample_rate,
+            },
+            stream=True,
+            transcript=text,
+            voice_id=self.config.voice_id,
+        )