Speech recognition and voices

Camlin Speech: Australian speech recognition and voices, hosted in Sydney

Streaming recognition with natural-pause endpointing, Australian text-to-speech with lip-sync data, one API.

Try it

Both demos run on the speech portal without an account.

Talk to the recogniser

Open the mic and watch your words appear as you speak.

Hear Camlin Voice

Type a phrase, pick an Australian voice, and watch the avatar speak it.

Send us your hardest audio

Names, places and reference numbers your callers say. We run it and send back what the engine heard.

For developers: API keys, streaming recognition, vocabulary biasing, and one-call text-to-speech that returns audio and visemes together.

Read the docs

How speech flows

Spoken words become checked business fieldsA man on the phone says “Reference 2448 0193, the 14th of October, $120.” The words stream in as they are said and the turn ends on the natural pause; the avatar on a mobile, listening, hands the journey a reference, a date and an amount, each with a confidence check.“Reference 2448 0193, the14th of October, $120.”Streamed as it is spokenTO THE JOURNEYREFERENCE2448 0193DATE14 OctAMOUNT$120.00Eight recognition modes, sixteen field types; every field carries a confidence check.
  1. 1

    Someone speaks

    On the phone, in web chat, to an avatar or by SMS.

  2. 2

    Words stream in

    The turn closes on a natural pause, not a fixed timer.

  3. 3

    Fields are checked

    Dates, numbers, IDs and references, with retries.

  4. 4

    Camlin Voice answers

    With the lip-sync track in the same response.

Camlin Voice and lip-sync

Australian voices, recorded in Sydney. Each response carries the audio and the lip-sync track together, 52 ARKit blendshapes at 60 fps, so the voice on your phone line can drive an avatar too.

Plans

Monthly, in Australian dollars, plus GST. Card checkout is coming; for now, request a plan and we set up your keys.

Starter sandbox

A$99 a month, ex GST

API keys, streaming recognition and Camlin Voice, for building and testing.

Team pilot

A$499 a month, ex GST

For a team taking a real service to pilot.

Sovereign rollout

Scoped with you

A dedicated deployment for an agency or carrier, planned with your team.

Where it runs and what is kept

  • The Camlin engine, our own engine built on open weights, runs by default, built, supported and hosted in Sydney.
  • The major cloud speech engines stay available per journey, when a tender or a language needs them.
  • One exception: stock avatar voices are made by a speech provider outside Australia. Camlin Voice’s own voices are recorded in Sydney.
  • What is stored and for how long is set out in the privacy policy.

Procurement detail is on the trust page.

Questions

Where is audio processed?

For Australian accounts, audio is processed in Sydney on the Camlin engine, our own engine built on open weights.

Which accents and terms does it handle, and can I add vocabulary?

It is built for Australian English. Register the names, brands and reference formats your callers say in a per-tenant lexicon, and recognition leans toward them.

How do I get access?

Try the live demos on the speech portal with no account. To build with it, request a plan on the speech portal and we set up your account.

Can I use another speech engine?

The Camlin engine is the default. The major cloud speech engines are available per journey when a project requires them.

Can I run it myself?

Ask us about the fit. Running Camlin Speech in your own environment is scoped with your team.

Does it work with Camlin Avatar and Companion?

Yes. Camlin Voice returns the audio and the lip-sync track in one response, which is what drives the avatar on a live call and the Companion on a website.

If you are...

Get started

Build on Camlin Speech

Starter sandbox from A$99 a month, ex GST.