Service status (updated 2026-11-03): available — live captions and realtime speech-to-text in 14 languages, live translation, text translation (14 languages and Tamil), batch speech-to-text (Hindi, Hinglish and other languages), fast file transcription, speech detection, speaker diarization, text-to-speech (34 voices in English, Spanish, French, German, Brazilian Portuguese and Hindi, plus voice design in all 14 languages), text tools, chat and coding, vision and document OCR, image generation and editing, sound effects, music and songs. Paused — sound-stream, voice studio routes and cloning. Details in the docs.

★ AI INFRASTRUCTURE · ONE API

Build AI-powered products —
without building the AI stack.

One API key for speech-to-text, live captions, text-to-speech, text tools and on-demand AI models — spoken port numbers and IP addresses come back as digits. Plain HTTPS and WebSocket endpoints.

Get started → Read the docs →

ONE API KEY + SPEECH-TO-TEXT + LIVE CAPTIONS + TEXT-TO-SPEECH + TEXT TOOLS + MODELS

HOW IT WORKS

From API key to production in 3 steps.

Get your API key

Create an API key (sk_…) at login.ollalink.com → Platform and keep it on your server.

Call the API

Call the speech hosts (gpu-*.ollalink.com) or text tools and models (ai.ollalink.com) with cURL or any HTTP client — OpenAI-style requests.

Build your product

Deliver transcripts, live captions, spoken audio, embeddings, model answers or generated images into your web, mobile, or enterprise applications.

curl https://gpu-tts.ollalink.com/v1/audio/speech \
  -H "X-NH-GPU-Key: sk_XXXX" \
  -H "Content-Type: application/json" \
  -d '{"input": "BGP peers exchange routes over TCP port 179.", "voice": "delia", "language": "en"}' \
  -o speech.mp3
# MP3 by default here; add "response_format": "wav" for WAV.
# 34 voices: GET https://gpu-tts.ollalink.com/voices

import requests

r = requests.post(
    "https://ai.ollalink.com/v1/chat/completions",
    headers={"X-NH-GPU-Key": "sk_XXXX"},
    json={"model": "chat",
          "messages": [{"role": "user",
                        "content": "Explain what BGP does in two sentences."}]},
    timeout=100,
)
msg = r.json()["choices"][0]["message"]
print(msg["content"])   # the answer; the reasoning is in msg["reasoning_content"]

curl https://gpu-transcribe.ollalink.com/v1/audio/transcriptions \
  -H "X-NH-GPU-Key: sk_XXXX" \
  -F "file=@lecture.mp3" -F "language=en"
# English, Spanish, French, German, Portuguese, Russian; English numbers as digits.
# Hindi, Hinglish and other languages: https://gpu-asrhq.ollalink.com/v1/audio/transcriptions

Works with the tools you already use

IF IT SPEAKS HTTP, IT WORKS WITH OLLALINK

VIEW ALL ENDPOINTS →

OpenAI-style requests speech, transcription, chat, images

cURL & REST HTTP JSON & multipart uploads

Realtime WebSockets 16 kHz mono PCM streaming

Streamed answers "stream": true on chat

One key, every host X-NH-GPU-Key header

Product capabilities

One backend. Multiple AI capabilities.

Connect your application to speech, captions, voice, text tools and on-demand models through Ollalink, with one API key for every service.

Speech API

Speech-to-Text

Upload audio in Hindi, Hinglish and other languages. Hindi comes back in Devanagari with English technical words in Latin script; speech can also be turned into English text, with word timestamps.

Speech API

Fast File Transcription

Audio or video up to 100 MB in English, Spanish, French, German, Portuguese and Russian, cut at pauses, with English numbers as digits. Speech detection returns speech start and end times.

Speech API

Realtime Speech-to-Text

Stream 16 kHz audio over a WebSocket with OpenAI-realtime-style events and receive partial and final transcripts in 14 languages.

Speech API

Live Captions

Stream a speaker's audio and receive partial and final captions in the spoken language; a final arrives about 1 s after the speaker pauses. 14 languages.

Speech API

Text-to-Speech

34 voices in English, Spanish, French, German, Brazilian Portuguese and Hindi, or design a new voice from a description in any of 14 languages. WAV, MP3, PCM and other formats.

Speech API

Live Translation

Caption a meeting or class in any of 14 languages and translate each caption into any of the 14 as it arrives — numbers, IP addresses and commands kept intact — then read it aloud if you like.

Text Tools

Turn spoken numbers into digits in English and Hindi ("open port four forty three" → "open port 443"), and create embeddings for documents and queries.

Models

Chat, Coding & More

Always-on chat and coding with OpenAI-style requests, plus vision and speaker diarization (who spoke when).

Models

Image Generation & Editing

Create pictures from a text prompt up to 1024 px, or edit an uploaded picture with a prompt. Repeatable with a seed.

Models

Sound Effects & Music Clips

Generate sound effects or instrumental music from a text description, 1 to 120 seconds, returned as a WAV file.

Models

Songs from Lyrics

Compose a song with vocals from a style description and tagged lyrics; runs as a job you poll, with M4A or FLAC download.

Models

Document OCR

Turn PDFs and scans into markdown page by page, tables included, with optional block positions. Reads English, Hindi and Chinese.

Architecture

Two families of hosts. One API key.

Speech runs on the gpu-*.ollalink.com hosts; text tools and on-demand models run on ai.ollalink.com. Both take the same key.

YOUR APPLICATION (Web / Mobile / Backend)

Sends requests securely via your backend proxy

HTTPS REST / WebSockets

Service 01

Text Tools & Models

https://ai.ollalink.com

Auth: X-NH-GPU-Key: <key>
Text tools: POST /v1/normalize (spoken numbers → digits), POST /v1/embeddings, POST /v1/classify
Models: /v1/chat/completions, /v1/images/generations, /v1/images/edits, /v1/audio/generations, /v1/music/generations, /v1/ocr, GET /v1/models
Limits: 600 requests / 60 s per key; a model request must finish within 100 s

Service 02

Speech APIs

gpu-*.ollalink.com

Auth: X-NH-GPU-Key: <key>
Capabilities: Batch STT, Fast File Transcription, Speech Detection, Speaker Diarization, Realtime STT, Live Captions, Live Translation, Text-to-Speech
Streaming: WebSocket, 16 kHz mono PCM, for live captions and realtime speech-to-text
Languages: speech input and translation in English, Hindi, Spanish, French, Chinese, German, Arabic, Portuguese, Russian, Japanese, Korean, Italian, Dutch and Vietnamese (text translation also Tamil); voices in English, Spanish, French, German, Portuguese and Hindi, designed voices in all 14

Developer experience

Endpoints at a glance.

Each capability has its own host and route. Every one takes the same API key in the X-NH-GPU-Key header.

Host	Use it for	Languages	Notes
gpu-asrhq	Batch transcription incl. Hindi and Hinglish; speech → English text.	Hindi, Hinglish and other languages	Multipart upload; optional word timestamps.
gpu-transcribe	Fast file transcription; speech detection (`/v1/audio/segments`); speaker diarization (`/v1/audio/diarization`).	en, es, fr, de, pt, ru, it, nl	Audio or video up to 100 MB; English numbers as digits.
gpu-asr	Realtime speech-to-text over a WebSocket.	14 languages	Partial and final transcripts; commit to finalize at once.
gpu-live	Live captions over a WebSocket; translate them with `/v1/translate` for live translation.	14 languages	A final about 1 s after each pause.
gpu-tts	Text-to-speech — `POST /v1/audio/speech`, voice list at `GET /voices`.	en, es, fr, de, pt, hi; designed voices: the 14 languages	34 voices plus voice design (`"voice": "design"` with a description); up to 4,000 characters per request (Hindi voices: 600, designed voices: 1,000).
ai.ollalink.com	Text translation (`/v1/translate`), text tools (`/v1/normalize`, `/v1/embeddings`, `/v1/classify`), chat, coding and vision (`/v1/chat/completions`), document OCR (`/v1/ocr`), images (`/v1/images/generations`, `/v1/images/edits`), audio (`/v1/audio/generations`), songs (`/v1/music/generations`).	Translation: 14 languages and Tamil; numbers to digits: en, hi	600 requests / 60 s per key.

Server-Side Authentication Security

Keep your AI credentials server-side.

Ollalink authentication is designed for backend-to-backend calls. Never bundle your API key (sk_…) into browser JavaScript or client-side mobile binaries.

Always proxy client requests through your own backend API to keep access control, rate limiting and the key in your hands.

View Authentication Docs →

// ✓ Recommended Backend Route:

app.post('/api/speech', async (req, res) => {

// The API key stays in process.env
const r = await fetch('https://gpu-tts.ollalink.com/v1/audio/speech', {
  method: 'POST',
  headers: { 'X-NH-GPU-Key': process.env.OLLALINK_GPU_KEY,
    'Content-Type': 'application/json' },
  body: JSON.stringify({ input: req.body.text, voice: 'delia', language: 'en' })
});
res.type('audio/mpeg').send(Buffer.from(await r.arrayBuffer()));

});

// ✗ Never expose sk_* keys in frontend scripts

Solutions

What developers are building with Ollalink.

From classrooms and meetings to internal tools, Ollalink provides the speech, voice and model building blocks.

Lecture Transcripts

Turn recorded classes and videos into searchable transcripts, with English numbers written as digits.

Live Captions & Translation for Classes and Meetings

Show captions in the speaker's language as people talk in classes, webinars and calls, and translate them live into any of 14 languages or Tamil.

Hindi & Hinglish Recordings

Transcribe Hindi and Hinglish speech in Devanagari with English technical words kept in Latin script.

Voice for Learning Content

Read lessons and prompts aloud with 34 voices in English, Spanish, French, German, Brazilian Portuguese and Hindi.

Search Building Blocks

Create embeddings for documents and queries, and normalize spoken numbers in transcripts before you index them.

Meeting & Content Assistants

Find who spoke when, then summarize with a text model, read screenshots with a vision model or create images — check model answers before you rely on them.

Pricing

Developer-friendly infrastructure. Zero surprise bills.

Start testing for free in minutes. Add speech capacity and on-demand models as your workload grows.

SANDBOX

Free

For prototyping, testing, and verifying integration pipelines.

Speech, text tools and model APIs
One API key for every service
Community & Docs Support
Standard Sandbox Rate Limits

Get API Key

Speech & Captions

Production access to transcription, live captions and voices.

Live captions and live translation in 14 languages
Batch and fast file transcription
Text-to-speech with 34 voices and voice design
Speech detection and speaker diarization

Start Building

MODELS

Models

Chat, coding, document OCR and the default image model are always on; the others load on their first request.

Text and reasoning models
Vision and document OCR
Image generation and editing
Sound effects, music clips and songs
Embeddings and spoken-number normalization

Talk to Engineering

Your AI backend starts here.

Build intelligent products on one developer infrastructure layer. Connect to speech, captions, voice, text tools and on-demand models today.

Get started free → Read the documentation →

Build AI-powered products — without building the AI stack.

From API key to production in 3 steps.

Get your API key

Call the API

Build your product

Works with the tools you already use

One backend. Multiple AI capabilities.

Speech-to-Text

Fast File Transcription

Realtime Speech-to-Text

Live Captions

Text-to-Speech

Live Translation

Text Tools

Chat, Coding & More

Image Generation & Editing

Sound Effects & Music Clips

Songs from Lyrics

Document OCR

Two families of hosts. One API key.

Text Tools & Models

Speech APIs

Endpoints at a glance.

Keep your AI credentials server-side.

What developers are building with Ollalink.

Lecture Transcripts

Live Captions & Translation for Classes and Meetings

Hindi & Hinglish Recordings

Voice for Learning Content

Search Building Blocks

Meeting & Content Assistants

Developer-friendly infrastructure. Zero surprise bills.

Free

Speech & Captions

Models

Your AI backend starts here.

Build AI-powered products —
without building the AI stack.