feat: Add Kokoro TTS FastAPI template (#353) (#403)

* feat: Add Kokoro TTS FastAPI template (#353)  - Add CPU-optimized docker-compose.yml with source build - Add GPU-optimized docker-compose-gpu.yml for NVIDIA support - Add comprehensive template.toml with OpenAI-compatible API docs - Add kokoro-tts.svg logo and meta.json entry - Support streaming audio, timestamps, and multi-language TTS - Resolves #353

* updated the meta.json for the build errors

* removed the docker-compose-gpu.yml file

* Update docker-compose.yml

---------

Co-authored-by: Mauricio Siu <47042324+Siumauricio@users.noreply.github.com>

blueprints/kokoro-tts/docker-compose.yml

@@ -0,0 +1,23 @@
+services:
+  kokoro-tts:
+    build:
+      context: https://github.com/remsky/Kokoro-FastAPI.git#master
+      dockerfile: docker/cpu/Dockerfile
+    restart: unless-stopped
+    ports:
+      - 8880
+    environment:
+      - MODEL_PATH=/app/models
+      - DEVICE=${DEVICE:-cpu}
+      - HOST=0.0.0.0
+      - PORT=8880
+    volumes:
+      - kokoro-models:/app/models
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8880/health"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 60s
+volumes:
+  kokoro-models:

blueprints/kokoro-tts/kokoro-tts.svg

@@ -0,0 +1,12 @@
+<svg width="64" height="64" viewBox="0 0 64 64" fill="none" xmlns="http://www.w3.org/2000/svg">
+  <rect width="64" height="64" rx="12" fill="#3B82F6"/>
+  <path d="M16 24C16 20.6863 18.6863 18 22 18H42C45.3137 18 48 20.6863 48 24V40C48 43.3137 45.3137 46 42 46H22C18.6863 46 16 43.3137 16 40V24Z" fill="white"/>
+  <path d="M24 28H40V32H24V28Z" fill="#3B82F6"/>
+  <path d="M24 34H36V38H24V34Z" fill="#3B82F6"/>
+  <circle cx="32" cy="42" r="2" fill="#3B82F6"/>
+  <path d="M20 22L44 22" stroke="#3B82F6" stroke-width="2" stroke-linecap="round"/>
+  <path d="M20 26L44 26" stroke="#3B82F6" stroke-width="2" stroke-linecap="round"/>
+  <path d="M20 30L44 30" stroke="#3B82F6" stroke-width="2" stroke-linecap="round"/>
+  <path d="M20 34L44 34" stroke="#3B82F6" stroke-width="2" stroke-linecap="round"/>
+  <path d="M20 38L44 38" stroke="#3B82F6" stroke-width="2" stroke-linecap="round"/>
+</svg>

blueprints/kokoro-tts/template.toml

@@ -0,0 +1,72 @@
+[variables]
+main_domain = "${domain}"
+device = "cpu"
+
+[config]
+env = [
+  "DEVICE=${device}",
+  "MODEL_PATH=/app/models",
+  "HOST=0.0.0.0",
+  "PORT=8880",
+  "PYTHONUNBUFFERED=1",
+  "UV_SYSTEM_PYTHON=1"
+]
+
+[[config.domains]]
+serviceName = "kokoro-tts"
+port = 8880
+host = "${main_domain}"
+path = "/"
+
+[[config.mounts]]
+filePath = "README.md"
+content = """# Kokoro TTS FastAPI
+
+This template provides a Dockerized FastAPI wrapper for the Kokoro-82M text-to-speech model.
+
+## Features
+
+- Multi-language support (English, Japanese, Chinese)
+- OpenAI-compatible speech endpoint
+- CPU and GPU support
+- Web interface for monitoring
+- RESTful API with comprehensive documentation
+- Streaming audio generation
+- Word-level timestamps and phonemes
+
+## Usage
+
+- **Web Interface**: Access the web UI at `https://${main_domain}/web`
+- **API Documentation**: Available at `https://${main_domain}/docs`
+- **Health Check**: Monitor service health at `https://${main_domain}/health`
+
+## Configuration
+
+The service runs on port 8880 and supports both CPU and GPU inference.
+For GPU support, ensure your Dokploy instance has NVIDIA GPU support enabled.
+
+## API Endpoints
+
+- `POST /v1/audio/speech` - Generate speech from text (OpenAI compatible)
+- `POST /dev/captioned_speech` - Generate speech with timestamps
+- `POST /dev/phonemize` - Convert text to phonemes
+- `POST /dev/generate_from_phonemes` - Generate audio from phonemes
+- `GET /health` - Health check endpoint
+- `GET /docs` - Interactive API documentation
+- `GET /web` - Web interface
+
+## Model Information
+
+- Model: Kokoro-82M
+- License: Apache-2.0
+- Repository: https://github.com/remsky/Kokoro-FastAPI
+- HuggingFace Model: https://huggingface.co/hexgrad/Kokoro-82M
+
+## Notes
+
+- This template builds the image from source during deployment
+- Uses CPU-optimized Dockerfile by default
+- Initial build may take several minutes due to model download
+- Ensure sufficient disk space for model storage
+- For GPU support, change dockerfile path to `docker/gpu/Dockerfile` in docker-compose.yml
+"""

meta.json

@@ -3195,6 +3195,25 @@
       "personal"
     ]
   },
+  {
+    "id": "kokoro-tts",
+    "name": "Kokoro TTS",
+    "version": "latest",
+    "description": "Dockerized FastAPI wrapper for the Kokoro-82M text-to-speech model with multi-language support and OpenAI-compatible endpoints.",
+    "logo": "kokoro-tts.svg",
+    "links": {
+      "github": "https://github.com/remsky/Kokoro-FastAPI",
+      "website": "https://github.com/remsky/Kokoro-FastAPI",
+      "docs": "https://github.com/remsky/Kokoro-FastAPI#readme"
+    },
+    "tags": [
+      "text-to-speech",
+      "ai",
+      "voice",
+      "fastapi",
+      "openai-compatible"
+    ]
+  },
   {
     "id": "kokoro-web",
     "name": "Kokoro Web",