Ollalink Platform Overview
Ollalink is an enterprise-grade AI Backend-as-a-Service providing high-throughput developer APIs for language models, technical translation, speech recognition, and regional Indic voice synthesis.
Capabilities at a Glance
| Capability | Description | Primary Use Case |
|---|---|---|
| Chat & RAG | OpenAI-compatible chat completions with retrieval augmentation, function calling, and structured JSON outputs. | Technical assistants, document Q&A, customer workflows |
| Technical Translation | Preserves technical entities (IP addresses, CIDR, Cisco CLI syntax) while translating vernacular speech. | Documentation, multi-lingual call centers, engineering portals |
| Speech-to-Text (STT) | Whisper-based speech recognition with standard and high-accuracy GPU streaming lanes. | Live call transcription, meeting records, voice commands |
| Text-to-Speech (TTS) | 50-voice loudness-normalized (−23 LUFS) catalog tuned across English, Hindi, and Kannada. | Interactive voice response, interactive learning, audio notifications |
| Voice Cloning & Design | Zero-shot voice cloning from 30 seconds of clean reference audio without model fine-tuning. | Branded voice agents, custom instructor personas |
| Live Translation | Bidirectional audio streaming over WebSockets with real-time translated captions. | Cross-border engineering meetings, bilingual support |
Two Specialized Services
To maintain sub-200ms latency on audio while serving high-throughput text, Ollalink operates two distinct compute planes:
| Service | Base host | Auth | Covers |
|---|---|---|---|
| Gateway | https://ai.ollalink.com |
Authorization: Bearer <key> |
Chat / RAG · translation · vision · standard speech-to-text · text-to-speech |
| Speech GPU | gpu-*.ollalink.com |
X-NH-GPU-Key: <key> |
High-accuracy & realtime speech-to-text · 50-voice TTS · voice cloning · voice design · live translation |
Start with Getting started, then review Authentication.