feat: add BreezeBlue TTS extension

ai_agents/.env.example

@@ -164,6 +164,15 @@ MISTRAL_API_KEY=
 #   curl -H "Authorization: Bearer $MISTRAL_API_KEY" https://api.mistral.ai/v1/audio/voices
 MISTRAL_TTS_VOICE=
 
+# Extension: breezeblue_tts_python
+# BreezeBlue TTS API key (brz_...), created at
+# https://breezeblue.ai/app/developer/api-keys
+BREEZE_API_KEY=
+# Voice for the guarder/tests. Voices are account-visible; the public catalog
+# works with any key. List them with:
+#   curl -H "xi-api-key: $BREEZE_API_KEY" "https://api.breeze.blue/v1/voices?voice_type=default&language_code=en"
+BREEZE_TTS_VOICE_ID=
+
 # Extension: qwen3_tts_python
 # Qwen3 TTS uses local models from HuggingFace, no API key required
 # Configure model, speaker, and mode in the extension properties

ai_agents/agents/integration_tests/tts_guarder/tests/bin/start

@@ -9,7 +9,7 @@ EXTENSION_NAME=${EXT_NAME:-elevenlabs_tts_python}
 
 export PYTHONPATH=.:ten_packages/system/ten_runtime_python/lib:ten_packages/system/ten_runtime_python/interface:ten_packages/system/ten_ai_base/interface:ten_packages/extension/${EXTENSION_NAME}:$PYTHONPATH
 # some tts extension does not support sample rate comparison, so we need to set the enable_sample_rate parameter
-if [[ "${EXTENSION_NAME}" == "humeai_tts_python" || "${EXTENSION_NAME}" == "openai_tts_python" || "${EXTENSION_NAME}" == "openai_tts2_python" || "${EXTENSION_NAME}" == "mistral_tts_python" || "${EXTENSION_NAME}" == "ezai_tw_tts_python" || "${EXTENSION_NAME}" == "typecast_tts_python" ]]; then
+if [[ "${EXTENSION_NAME}" == "humeai_tts_python" || "${EXTENSION_NAME}" == "openai_tts_python" || "${EXTENSION_NAME}" == "openai_tts2_python" || "${EXTENSION_NAME}" == "mistral_tts_python" || "${EXTENSION_NAME}" == "breezeblue_tts_python" || "${EXTENSION_NAME}" == "ezai_tw_tts_python" || "${EXTENSION_NAME}" == "typecast_tts_python" ]]; then
     export ENABLE_SAMPLE_RATE="False"
 else
     export ENABLE_SAMPLE_RATE="True"

ai_agents/agents/ten_packages/extension/breezeblue_tts_python/README.md

@@ -0,0 +1,84 @@
+# breezeblue_tts_python
+
+TEN text-to-speech extension for [BreezeBlue](https://docs.breezeblue.ai/),
+using its HTTP streaming API (`POST /v1/text-to-speech/{voice_id}/stream`).
+
+## Features
+
+- Streaming synthesis: raw PCM16, 24 kHz, mono. No conversion is needed, the
+  stream matches the TEN `pcm_frame` contract.
+- Public voice catalog and your own saved/cloned voices (`voice_id`).
+- Models `breeze-tts-2` (English, Chinese) and `breeze-tts-2-multilingual`.
+  When `model_id` is omitted the vendor picks one from `language_code`.
+- Expressive controls: `instructions`, `speed`, `volume`, `guidance_scale`,
+  `use_speaker_boost`.
+
+## Configuration
+
+Everything goes in `params`:
+
+- `api_key` (string, required): BreezeBlue key (`brz_...`), sent as the
+  `xi-api-key` header. Create one at
+  <https://breezeblue.ai/app/developer/api-keys>.
+- `voice_id` (string, required): voice to synthesize with. The voice is part of
+  the request path and there is no server-side default. List the public
+  catalog with:
+  `curl -H "xi-api-key: $BREEZE_API_KEY" "https://api.breeze.blue/v1/voices?voice_type=default&language_code=en"`
+- `model_id` (string): defaults to the vendor's automatic choice.
+- `language_code` (string): ISO 639-1 two-letter code, e.g. `en`, `zh`.
+- `instructions` (string): performance direction, in the language of the text.
+- `speed` (0.5-2.0), `volume` (0.01-2.0), `guidance_scale` (1.0-10.0),
+  `use_speaker_boost` (bool): may also be nested under `voice_settings`. Flat
+  keys win over nested ones.
+- `base_url` (string): API base (default `https://api.breeze.blue`).
+
+Any other key in `params` is ignored. The vendor rejects unknown request
+fields with HTTP 422, so only the fields above are forwarded. The frontend's
+generic `sample_rate` is dropped for the same reason (output is fixed at
+24 kHz).
+
+```json
+{
+  "type": "extension",
+  "name": "tts",
+  "addon": "breezeblue_tts_python",
+  "extension_group": "tts",
+  "property": {
+    "params": {
+      "api_key": "${env:BREEZE_API_KEY}",
+      "voice_id": "${env:BREEZE_TTS_VOICE_ID}",
+      "language_code": "en"
+    }
+  }
+}
+```
+
+## Limits and behavior
+
+- Text is limited to 1000 characters per request by default (up to 2000 for
+  approved accounts). Longer text is rejected, not truncated.
+- `speed` and `volume` apply to HTTP streaming (not to the realtime WebSocket,
+  which this extension does not use).
+- Billing is per character: 10 units = 1 credit, and CJK characters count as 2
+  units.
+- A new account is limited to 2 concurrent generations.
+- 401/403 responses are reported as an invalid-key error. Other HTTP errors,
+  such as 402 (no credits), 422 and 429, are reported as general errors with
+  the vendor's `detail` message.
+
+## Tests
+
+Unit tests (`tests/test_config.py`, `tests/test_client.py`) mock the HTTP layer.
+The TTS guarder needs `BREEZE_API_KEY` and `BREEZE_TTS_VOICE_ID` in
+`ai_agents/.env`:
+
+```bash
+task tts-guarder-test EXTENSION=breezeblue_tts_python CONFIG_DIR=tests/configs
+```
+
+## Architecture
+
+- **Extension**: `BreezeBlueTTSExtension` extends `AsyncTTS2HttpExtension`.
+- **Config**: `BreezeBlueTTSConfig` extends `AsyncTTS2HttpConfig`.
+- **Client**: `BreezeBlueTTSClient` extends `AsyncTTS2HttpClient` and streams
+  the response with httpx (HTTP/2, shared SSL context).

ai_agents/agents/ten_packages/extension/breezeblue_tts_python/__init__.py

@@ -0,0 +1,6 @@
+#
+# 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

ai_agents/agents/ten_packages/extension/breezeblue_tts_python/addon.py

@@ -0,0 +1,21 @@
+#
+# 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_runtime import (
+    Addon,
+    register_addon_as_extension,
+    TenEnv,
+)
+
+from .extension import BreezeBlueTTSExtension
+
+
+@register_addon_as_extension("breezeblue_tts_python")
+class BreezeBlueTTSExtensionAddon(Addon):
+
+    def on_create_instance(self, ten_env: TenEnv, name: str, context) -> None:
+
+        ten_env.log_info("BreezeBlueTTSExtensionAddon on_create_instance")
+        ten_env.on_create_instance_done(BreezeBlueTTSExtension(name), context)