ChivoxAI
sign in
liveSpeech intelligence · English & Mandarin
AI-powered assessment for education

Speech & PronunciationAssessment for EdTech

Chivox brings AI-powered speech and pronunciation assessment to EdTech products, delivering phoneme-level accuracy, fluency, prosody, and actionable feedback for English and Mandarin learners.

Try the demo
20 years of speech R&DHigh-stakes testing experience1.2B global users
54321你好T3 ✓T3 ✓nǐhǎoT3 + T3 → T2 + T3TONE SANDHI · DETECTEDSCORE92thinkPHONEMEACCURACYHEARD/s/WEAKDROPPED—/s//θ//ɪ/✓/ŋ//ŋ//k//k/3 ISSUES · DETECTEDMISSING · WEAK · MISPRONOUNCEDALL CORRECTEDPHONEME DIAGNOSIS · SUPERVISEDSCORE5892

/customers

Trusted inside learning products around the world.

Platforms, publishers and education partners using Chivox speech assessment in production.

PearsonPearsonNowUkanNowUkanPrometheanPrometheanEdmodoEdmodoRentrisRentrisNAVERNAVERJumpStartJumpStartHuahua LearningHuahua Learning
Students using speech assessment in a classroom learning environment

/why chivox

Speech research, proven in exams and real products.

Chivox’s technical roots go back to Cambridge in 2007. Since then, the engine has been shaped by language-learning products, large-scale education programmes and high-stakes speaking assessment.

  • 20 years of speech-assessment R&D
  • 10+ years supporting high-stakes language testing
  • Experience across 1,000+ education projects
Our story Customers

Products and partners

188
Countries & regions

Across education products

1.2B
Global cumulative users

Learner audio corpus

52M
Hours of recording data

Hardware deployments

36M
Offline devices

/application scenarios

See where speech assessment fits.

Browse all six real-world workflows. Scroll through the cards, or open the full comparison to evaluate them side by side.

View all scenarios
Young learner speaking into a headset while a parent reviews progress in a bright home learning space
Child-tuned English assessment

Young learner products

Speech assessment for young voices.

Open scenario
Learner completing an online speaking activity with automated feedback
Learning platforms

Online education

Give every learner useful speaking feedback, at any scale.

Open scenario
Teacher reviewing simultaneous spoken responses in a classroom
Face-to-face classes

Classroom training

Hear the whole room, not only the learner who raises a hand.

Open scenario
Students completing a computer-based speaking assessment in a school lab
Tests & benchmarking

School assessment

Make speaking assessment consistent, explainable and easier to operate.

Open scenario
Child practising spoken language with a connected learning device
Devices & companions

Smart hardware

Put a responsive language coach inside the device.

Open scenario
Guided speech-practice session using structured pronunciation feedback
Guided intervention

Speech practice & therapy

Make practice progress visible between sessions.

Open scenario
6 scenariosMore scenarios to the right →

/scoring-engines

Two scoring engines. One assessment platform.

English and Mandarin use different acoustic models and scoring rubrics. Learner age, classroom context and device behavior belong to the use-case layer—not a third engine.

See the shared platform
01

English

/θ/ · stress · liaison

Phoneme accuracy, word stress, liaison, fluency and CEFR-aligned dimensions.

Explore scoring engine
02

Mandarin

nǐ hǎo · 1 2 3 4

Tone contour, pinyin, neutral tone, erhua, tone sandhi and PSC-standard words.

Explore scoring engine

/integrate

The assessment engine, delivered your way.

Product teams embed it. Agents call it. Both receive the same assessment evidence.

SDK & API

Embed in your product

Native SDK plus REST and WebSocket for mobile, web and backends.

Best for EdTech apps embedding assessment in-product.

  • JS / Native SDKs for product surfaces
  • REST and WebSocket streaming paths
  • Enterprise licensing through BD
iOS · Android · Web · ServerExplore

MCP server

Give an LLM tools

Sixteen English and Mandarin assessment tools for any MCP client.

Best when an LLM should call scoring tools directly.

  • 16 tools for English and Mandarin
  • Cursor, Claude Desktop, custom agents
  • Self-serve evaluation-point billing
Evaluation-point pricingExplore

Function calling

Keep your orchestration

The same engine through familiar tool contracts over REST or WebSocket.

Best if you already own the orchestrator and tool loop.

  • Same tool contracts as the MCP surface
  • Fits your own agent / workflow stack
  • Pricing matches MCP evaluation points
Same pricing as MCPExplore
/need help choosing?Not sure which integration path fits?Compare ownership, orchestration, transport and pricing across all three routes.Compare all three

Get started

Ready to score speech in your product?

MCP and function calling developers can self-serve free credits today. SDK/API licensing and regional distribution go through the shared inquiry flow.

Start MCP / function callingSee a demo
Solution mappingIntegration supportProduction onboarding