1.6 KiB
1.6 KiB
Phoenix Voice Orchestrator
Category: ORCHESTRATION_SERVICES
Publisher: Phoenix Cloud Services
Status: PUBLISHED
Version: 1.0.0
Overview
Phoenix Voice Orchestrator provides text-to-speech and speech-to-text orchestration with audio caching, multi-provider routing, and moderation. Features deterministic caching for cost and latency optimization, PII scrubbing, and multi-model routing.
Key Features
- Audio caching: Hash-based deterministic caching
- Multi-provider routing: ElevenLabs, OpenAI, Azure TTS, OSS fallback
- PII scrubbing: Automatic PII detection and removal
- Moderation: Content moderation before synthesis
- Multi-model support: High quality vs low-latency routing
- CDN delivery: Cached audio via CDN
- OSS fallback: Open-source TTS for baseline
API Endpoints
Synthesize Voice
POST /voice/synthesize
Content-Type: application/json
{
"text": "Hello, world!",
"voiceProfile": "en-US-female",
"format": "mp3",
"latencyClass": "STANDARD"
}
Get Cached Audio
GET /voice/audio/{hash}
Transcribe Speech
POST /voice/transcribe
Content-Type: multipart/form-data
{
"audio": <file>,
"language": "en-US"
}
Pricing
- Model: Usage-based
- Per Synthesis: $0.02
- Per Minute: $0.10
- Per Transcription: $0.05
- Free Tier: 100 syntheses/month
Compliance
- SOC 2 Type II
- GDPR compliant
SLA
- Uptime: 99.9%
- Latency: <500ms p95
Architecture
- Hash-based cache key generation
- Provider adapter pattern
- PII detection and scrubbing
- CDN integration for delivery