VOICE/EMOTION
A voice AI developer platform. Its Octave text-to-speech engine produces expressive, emotionally aware speech, its EVI speech-to-speech interface holds real-time voice conversations that respond to tone of voice, and it supports voice cloning.
- PRICE
- Freemium
- SET-UP
- Advanced
- DATA
- No AU option published
- FOUNDED
- 2021
- HQ
- New York, USA
- WEBSITE
- hume.ai
- REVIEWED
- October 2026
SPEC SHEET
001 / WHAT IT DOES
- 01Octave text-to-speech that adjusts emotion and delivery from text instructions
- 02EVI speech-to-speech interface for real-time voice conversations
- 03Voice cloning and voice design from short samples or text descriptions
- 04Expression measurement models that analyse tone in voice, face and language
- 05APIs and SDKs for embedding voice into web, mobile and telephony apps
BEST FOR
- Developers building voice agents that need expressive, natural speech
- Product teams prototyping voice interfaces for apps or kiosks
- Content producers needing narration with controllable emotion
TYPICAL USES
002 / THE HONEST TAKE
STRENGTHS
- +Speech sounds more natural and expressive than basic text-to-speech
- +Real-time conversational voice through one API
- +Free tier makes it cheap to prototype
- +Voice cloning included on paid plans
WATCH OUTS
- –Developer-focused, so non-technical teams need help to use it
- –Leadership changes add some uncertainty to long-term direction
- –Emotion detection from voice raises privacy and fairness questions
- –Data hosting location is not published, so ask before using it for calls
003 / FOR AUSTRALIAN BUSINESSES
No Australian hosting option is published, so assume data is stored and processed overseas.
Hume does not publish an Australian hosting region or let customers choose where data is processed. Its API data usage policy says customer-submitted data is not used to train its general models, and custom model data is used only for that customer. Ask Hume where audio and transcripts are stored, how long they are kept and which sub-processors handle conversations.
PRIVACY ACT & PRACTICAL NOTES
Voice cloning needs explicit, written consent from the person whose voice is used, and a synthetic voice should never be presented as a real person endorsing your business, which would risk misleading conduct under Australian Consumer Law. If EVI takes calls from Australian customers, their recordings are personal information under the Privacy Act 1988, and callers should be told they are speaking with AI. Hume's CEO joined Google DeepMind under a licensing deal in early 2026, so confirm the product roadmap before committing to a long build.
Client data that can't leave Australia? There are ways to get the same job done onshore.
See private AI →004 / WHAT IT COSTS
Freemium
A free tier for experimenting, then monthly plans that include a set amount of text-to-speech characters and conversation minutes, with pay-as-you-go overage. Enterprise pricing is available for high volumes. Check hume.ai/pricing for plan details.
Check current prices on hume.ai ↗︎005 / CONNECTS WITH
No integrations listed in our review. Check the vendor for an API or connectors.
Want it wired into your CRM, inbox or Xero? That's what we build.
006 / ALTERNATIVES
Also worth a look
- ElevenLabsOffshoreFreemium→
- VapiOffshorePaid→
- Retell AIOnshore possiblePaid→
- Murf.aiAU optionFreemium→
- KrispOffshorePaid→
- SpeechifyOffshoreFreemium→
Also compared against: OpenAI Realtime API, Cartesia, Google Cloud Text-to-Speech.
007 / QUESTIONS
Anything else is quicker to answer on a call. Thirty minutes, free, with the people who would build it.
Yes. In January 2026 Hume's chief executive and a small group of engineers joined Google DeepMind under a non-exclusive licensing agreement, while Hume continues to operate and sell its voice APIs to other customers. Check the product roadmap and support commitments in your contract before building a long-term voice product on it.
We set up and connect tools like this for Australian businesses, with the privacy settings done right from day one.
Book a free call→