ElevenLabs
AI voice platform for text-to-speech, voice cloning, and conversational agents in 70+ languages
FREEMIUM Verified JUL 3, 2026
Hume AI
Empathic voice AI for developers — Octave 2 expressive text-to-speech and the EVI speech-to-speech API with per-character and per-minute pricing
Hume AI is a developer platform for expressive voice AI: the Octave text-to-speech model (Octave 2 in preview) and the EVI speech-to-speech API. The free tier includes 10,000 TTS characters and 5 EVI minutes per month; paid plans run from $3 to $500/month, with overage falling from $0.15 to $0.05 per 1,000 TTS characters and $0.07 to $0.04 per EVI minute as tiers rise. Best for developers building voice agents, narration, and character voices.
Verified JUL 6, 2026 FREEMIUM Live
What is Hume AI?
Hume AI is a voice-AI research lab and developer platform built around emotional expression. Its two commercial products are Octave, a text-to-speech system built on an LLM that interprets what the text means before deciding how to say it — with voice design from plain-text prompts, voice cloning, and voice conversion — and EVI, the Empathic Voice Interface, a speech-to-speech API that listens to a user’s vocal modulation and replies expressively with native interruption handling and back-channeling. As of July 2026 the current generations are Octave 2 and EVI 4-mini, both in preview alongside the generally available Octave 1 and EVI 3. The company also publishes TADA, an open-source LLM TTS system, and its research base covers 50+ languages, 48+ emotions, and 600+ voice descriptors.
Billing is usage-metered inside monthly subscriptions: every tier includes a TTS character allowance and an EVI minute allowance, with overage rates that drop as tiers rise — from $0.15 down to $0.05 per 1,000 TTS characters and from $0.07 down to $0.04 per EVI minute. Plans run from a permanent free tier (10,000 characters and 5 EVI minutes per month) through Starter at $3, Creator at $14, Pro at $70, Scale at $200, and Business at $500 per month, with custom Enterprise deals adding SOC 2 Type II, GDPR, and HIPAA compliance. Developers integrate through REST and WebSocket APIs plus SDKs for React, TypeScript, Python, iOS/macOS, and .NET; voice cloning is unlimited on every tier.
Who is it for?
Hume AI is aimed at people who build with APIs rather than consumers looking for a point-and-click voice app — the platform playground exists for testing, but the product is the API surface. The free tier and $3 Starter plan make evaluation nearly free, while the per-minute EVI pricing model maps cleanly onto conversational products that bill their own users by usage.
- Developers building voice agents — support lines, booking systems, coaching apps — who want interruption handling and emotional responsiveness out of the box instead of stitching together separate STT, LLM, and TTS services, or who are comparing against orchestration layers like /tools/vapi.
- Content creators and narration services producing audiobooks, podcasts, or video voiceovers who need a specific voice designed from a description or cloned for brand consistency, as an alternative to /tools/elevenlabs.
- Product teams shipping conversational features at scale who care about unit economics: Business-tier overage of $0.05 per 1,000 characters and $0.04 per EVI minute is among the lowest published self-serve rates for expressive voice.
- Researchers and evaluation teams using the Human Feedback API’s survey templates to run structured studies on voice quality and expression.
How much does Hume AI cost?
Starting price: $3/mo · Free tier: yes · Model: freemium
Pricing verified JUL 6, 2026
Price history tracked from June 2026
| Plan | Price | Includes |
|---|---|---|
| Free | $0 | 10,000 TTS characters/month (~10 minutes) · 5 EVI minutes/month · 1 concurrent connection · Unlimited voice cloning (create and use) · Discord support |
| Starter | $3/mo | 30,000 TTS characters/month (~30 minutes) · 40 EVI minutes/month, then $0.07/minute · 5 concurrent connections · 50% off the first month |
| Creator | $14/mo | 140,000 TTS characters/month (~140 minutes), then $0.15/1,000 characters · 200 EVI minutes/month, then $0.07/minute · 5 concurrent connections |
| Pro | $70/mo | 1,000,000 TTS characters/month (~1,000 minutes), then $0.12/1,000 characters · 1,200 EVI minutes/month, then $0.06/minute · 10 concurrent connections |
| Scale | $200/mo | 3,300,000 TTS characters/month (~3,300 minutes), then $0.10/1,000 characters · 5,000 EVI minutes/month, then $0.05/minute · 20 concurrent connections · 3 team seats |
| Business | $500/mo | 10,000,000 TTS characters/month (~10,000 minutes), then $0.05/1,000 characters · 12,500 EVI minutes/month, then $0.04/minute · 30 concurrent connections · 5 team seats |
| Enterprise | Custom | Custom volumes and unlimited team seats · SOC 2 Type II, GDPR, HIPAA compliance · Slack support channel |
What are Hume AI's key features?
- Octave — LLM-based text-to-speech with voice design from text prompts, voice cloning, and voice conversion; Octave 2 available in preview
- EVI (Empathic Voice Interface) — speech-to-speech API with interruption handling, back-channeling, and expressive instruction following; EVI 3 plus EVI 4-mini in preview
- Unlimited voice cloning creation and use on every tier, including Free
- SDKs for React, TypeScript/Node.js, Python, iOS/macOS, and .NET
- Research base spanning 50+ languages, 48+ emotions, and 600+ voice descriptors
- TADA — Hume's open-source LLM TTS system that streams text and audio together to cut latency and hallucinations
- Human Feedback API with science-backed survey templates for voice evaluation studies
What people use Hume AI for
- 01 Building phone or in-app voice agents that respond to user emotion via the EVI speech-to-speech API
- 02 Narrating audiobooks, podcasts, and videos with Octave voices designed from a plain-text description
- 03 Cloning a specific voice for consistent brand narration across marketing and product content
- 04 Adding low-latency conversational voice with native interruption handling to support or coaching apps
- 05 Running human evaluation studies on voice output quality via the Human Feedback API
Pros and cons
| Pros | Cons |
|---|---|
| Cheap to prototype: a permanent free tier plus a $3/month Starter plan with 40 EVI minutes included | Developer-first product — there is no polished consumer app; you build against APIs and SDKs or work in the platform playground |
| Business-tier TTS overage of $0.05/1,000 characters undercuts most premium expressive-TTS rivals at scale | The newest models (Octave 2, EVI 4-mini) are labeled preview, so behavior and rates may still shift |
| EVI handles interruptions and back-channeling natively — conversational behaviors that must be hand-built on plain TTS stacks | Concurrency caps (1 connection on Free, 30 on Business) constrain production voice apps without an Enterprise deal |
| Voice cloning is unlimited on every tier, including Free | High-volume conversational usage adds up: 10,000 EVI minutes beyond the Pro allowance costs about $600 at $0.06/minute |
What are the best Hume AI alternatives?
Cartesia
Low-latency voice AI platform behind the Sonic text-to-speech model, built for real-time agents
USAGE-BASED Verified JUL 5, 2026
Resemble AI
AI voice platform for voice cloning, text-to-speech, and real-time voice agents, plus deepfake detection and audio watermarking
USAGE-BASED Verified JUN 30, 2026
Vapi
Developer platform to build, test, and deploy AI voice agents for phone calls with usage-based per-minute pricing
USAGE-BASED Verified JUN 24, 2026
Murf AI
AI text-to-speech studio with 200+ voices across 35+ languages for voiceovers, dubbing, and voice agents
FREEMIUM Verified JUN 11, 2026
How people make money with Hume AI
- Freelance voice-agent development — build EVI-powered booking, intake, or support phone lines for local businesses via Upwork or direct outreach; per-minute overage billing makes client cost pass-through straightforward
- Audiobook and long-form narration services using Octave voice design — the Pro plan includes 1,000,000 characters (~1,000 minutes) per month with commercial usage on paid plans
Frequently asked questions
Is Hume AI free?
There is a permanent free tier with 10,000 text-to-speech characters (~10 minutes) and 5 EVI conversation minutes per month, one concurrent connection, and unlimited voice cloning. It is enough to evaluate the APIs but not to run anything in production.
How much does Hume AI cost?
Paid plans are Starter $3/month, Creator $14/month, Pro $70/month, Scale $200/month, and Business $500/month, each with monthly TTS-character and EVI-minute allowances. Overage falls from $0.15 to $0.05 per 1,000 TTS characters and $0.07 to $0.04 per EVI minute as tiers rise. Enterprise pricing is custom.
What is EVI?
EVI (Empathic Voice Interface) is Hume's speech-to-speech API: it listens, interprets vocal expression, and replies in an expressive voice with native interruption handling and back-channeling. EVI 3 is generally available and EVI 4-mini is in preview with expanded language support and lower latency.
What is Octave 2?
Octave is Hume's text-to-speech system built on an LLM, so it interprets the meaning of the text rather than just pronouncing it. It supports voice design from text prompts, voice cloning, and voice conversion. Octave 2, the current generation, is available in preview.
Does Hume AI support voice cloning?
Yes. Voice cloning — both creating and using cloned voices — is unlimited on every plan, including the free tier. Commercial usage rights are tied to paid plans per the pricing page.
How is Hume AI different from ElevenLabs?
Both offer expressive TTS and conversational voice APIs. Hume's differentiators are its emotion-science research base (48+ emotions, 600+ voice descriptors), the speech-to-speech EVI API with built-in interruption handling, and aggressive at-scale pricing — $0.05 per 1,000 characters overage on the $500/month Business plan.
What SDKs does Hume AI provide?
Official SDKs cover React, TypeScript/Node.js, Python (sync and async clients), iOS/macOS, and .NET, alongside REST and WebSocket APIs documented on the developer portal.