Hume AI Logo

Foundation Models & Agent Frameworks · Data Provider / Broker

Hume AI

Evaluation and data infrastructure for emotionally intelligent voice AI.

Available information varies by company and source.

Profile record updated:

Company facts

Official name
Hume AI, Inc.
Entity type
COMPANY
Founded
2021
Headquarters
United States
Company size
50–200
Market cluster / sector
Foundation Models & Agent Frameworks
Market role
Data Provider / Broker
Official website
hume.ai

What Hume AI does

Hume AI operates a B2B software and data infrastructure model for voice AI developers and enterprises. It creates value by helping customers evaluate, simulate, benchmark and improve conversational AI systems using proprietary APIs, hosted tooling, human feedback workflows and custom dataset services. The software layer supports ongoing testing and regression management, while the data and human evaluation layer improves model quality and trust. This combination makes Hume AI part SaaS platform, part API infrastructure and part specialist managed data service provider.

Category differentiation

Hume AI is not a consumer voice assistant or a foundational LLM provider. It sells B2B evaluation, human feedback and data infrastructure for teams building voice AI systems.

Strategic context

AI-supported assessment from the existing company research; distinguish interpretation from sourced facts.

Hume AI, Inc. is a private US software company that provides evaluation, simulation, expression measurement and human-feedback infrastructure for voice AI systems. Its product set includes Kairos Simulate + Evaluate for testing and regression tracking, an Expression Measurement API for analysing emotional and expressive signals in speech, a Human Feedback API for rapid human scoring, and custom data services for training and evaluation. The company serves voice AI engineers, machine learning teams, researchers and product teams building conversational systems. Hume AI generates revenue through a mix of software subscriptions, usage-based API billing and custom enterprise or managed data engagements. Its commercial position is centred on becoming the data and evaluation layer for emotionally intelligent voice AI rather than a consumer application or foundation model provider. The business sells directly to organisations building voice agents and speech systems that need better evaluation quality, benchmark comparability and human-grounded training data.

Business model & monetisation

Hume AI monetises through tiered SaaS subscriptions for hosted evaluation products, usage-based pricing for APIs, and service fees for bespoke data collection and dataset generation. Enterprise contracts add custom pricing for higher-scale deployments, advanced workflows and tailored support. The Human Feedback API also supports managed-service style revenue tied to evaluation tasks and turnaround requirements.

Hosted evaluation platform subscriptions
Software subscription
Expression and evaluation API usage
Pay-per-Use
Human Feedback API engagements
Service Fee
Custom data collection and dataset generation
Service Fee

Products & capabilities

Recorded products with linked evidence; content review pending.
ProductDescriptionSource
Kairos Simulate + EvaluateKairos Simulate + Evaluate is a hosted evaluation platform designed for voice AI systems, enabling teams to simulate conversations between agents and humans and build large-scale evaluation suites based on real-world scenarios. It provides regression tracking and continuous evaluation workflows, allowing developers to identify failures and iterate rapidly during model development. The platform functions as a core evaluation and simulation layer for voice-native AI pipelines.
  • “Kairos lets voice AI teams build evaluation suites from real-world use cases, simulate agent-to-agent and human-to-agent conversations, and track regressions over time, all in record speed.”

Expression Measurement APIExpression Measurement API is a real-time and offline analysis API that quantifies emotional and expressive characteristics in speech using a large taxonomy of voice signals. It outputs hundreds of metrics across dozens of emotional categories and supports multilingual processing, enabling developers to integrate emotion-aware analysis directly into voice AI systems. The API provides granular measurement of expressivity, naturalness, and emotional signals for conversational AI applications.
  • “Real-time expression measurement and offline analysis across 48+ emotion categories and 50+ languages with 600+ output metrics”

Human Feedback APIHuman Feedback API is an API-driven human evaluation service that provides structured ratings, qualitative feedback, and aggregated insights from pre-screened human raters. It includes built-in participant screening, fraud detection, and quality monitoring, delivering rapid turnaround for large-scale evaluation tasks. The system enables human-in-the-loop model development by converting study setup to results through a single API call.
  • “Pre-screened human raters return per-sample scores, free-response feedback, and aggregated analysis. In hours, not days.”

Data SolutionsData Solutions is a bespoke data collection and dataset generation offering tailored to specific voice and emotional AI use cases. It delivers custom-built datasets aligned with client scenarios, supporting both training and evaluation workflows for conversational AI systems. The service focuses on human-grounded data collection processes rather than generic off-the-shelf datasets.
  • “Custom data collection built around your specific scenarios, use cases, and evaluation requirements.”

VoiceEQ BenchVoiceEQ Bench is a public benchmarking platform that evaluates the human-perceived quality of voice AI systems across multiple dimensions including recognition, understanding, expression, and conversation. It ranks models using human judgement to provide an objective comparison standard for voice AI technologies. The platform serves as a research-grade evaluation reference for developers and organisations building conversational systems.
  • “a benchmark for measuring the human quality of voice AI... ranked by human judgment.”

SLM JudgeSLM Judge is a leaderboard platform that evaluates automated scoring models by measuring how closely they align with human ratings. It provides comparative insights into the reliability of different evaluation models, helping teams select trustworthy automated evaluators. The system focuses on validating judge models rather than ranking primary AI systems.
  • “Which SLMs track human ratings most closely, so you know which automated evaluator you can actually trust.”

POLARIS7 WORKSPACE

Go deeper on Hume AI and its market context

Access extended corporate structures, historical M&A and partnership links, technical footprint data, and automated Watchlist briefings grounded in primary sources.

  • Corporate & subsidiary hierarchy
  • Competitor & partner network
  • Automated Watchlist briefings

Competitors & alternatives

Track competitor shifts

Receive automated telemetry on strategic moves, acquisitions, and market share shifts across Hume AI's competitive landscape.

Monitor company
  • AI model evaluation platform powered by human feedback data.

  • Expert intelligence platform for AI evaluation and deployment.

  • Independent AI model benchmarking and selection platform.

  • Enterprise AI data annotation and model evaluation provider.

  • Enterprise platform for AI data labelling, evaluation and expert services.

Recent recorded signals

Dates refer to the source publication. Older entries are historical context, not evidence of a new event.

  • Evaluating Google’s multi-speaker TTS: A case study in why private evaluations matter

    hume.ai

    Product & Platform Launch · Recorded impact score: 2/5

    Hume AI published an independent evaluation of Google's Gemini multi-speaker TTS models, using bespoke rubrics and 48 two-speaker scripts rated by three human raters through its Feedback API. Across five Gemini models from three generations, same-gender speaker separation emerged as the main weakness, with male-male pairs averaging 2.87 versus 4.65 for mixed-gender pairs. Gemini 2.5 Pro TTS, 3.8 Flash TTS, and 3.8 Flash-Lite TTS scored within 0.02 overall, but differed on specific dimensions. Comparisons with ElevenLabs v3, Higgs Audio v2, and VibeVoice 1.5B showed similar same-gender gaps, suggesting this challenge is not unique to Google.

    • Hume AI published the evaluation on September 29, 2026, covering five Google Gemini TTS models across three generations: Gemini 2.5 Flash TTS, 2.5 Pro TTS, 3.1 Flash TTS, 3.8 Flash TTS, and 3.8 Flash-Lite TTS.
    • The evaluation used 48 two-speaker scripts and rated five dimensions: speaker separation, voice stability, seamlessness, conversation dynamics, and emotional containment.
  • Introducing the Hume Voice Controllability Leaderboard

    hume.ai

    Product & Platform Launch · Recorded impact score: 4/5

    Hume AI published the Voice Controllability Leaderboard on September 17, 2026, extending its Real World VoiceEQ Benchmark. The new leaderboard evaluates 17 text-to-speech models on their ability to follow direction across voice design, voice instruction, and role fit, using human ratings as the gold standard. Results show no single model dominates every controllability dimension, with Google and ElevenLabs leading most categories but exhibiting distinct weaknesses.

    • On September 17, 2026, Hume AI published the Voice Controllability Leaderboard.
    • The leaderboard evaluates 17 text-to-speech models for controllability across voice design, voice instruction, and role fit.
  • Introducing the Hume Voice Replication Leaderboard

    hume.ai

    Product & Platform Launch · Recorded impact score: 3.5/5

    On September 10, 2026, Hume AI published the Voice Replication Leaderboard, a new component of its Real World VoiceEQ Benchmark. The leaderboard ranks 11 text-to-speech models on how accurately they reproduce a specific person's voice across 25 reference voices in standard, expressive, and accented categories. Models were rated on same-speaker similarity, quality, naturalness, and objective embedding similarity; no single model led all dimensions. OpenBMB VoxCPM2 topped same-speaker similarity, while Cartesia sonic-3.6-beta achieved the highest naturalness and Inworld TTS-2 the highest quality.

    • On September 10, 2026, Hume AI launched the Voice Replication Leaderboard as an extension of its Real World VoiceEQ Benchmark.
    • The leaderboard evaluated 11 TTS models using 25 reference voices: 5 standard, 5 expressive, and 15 accented voices (6 native English and 9 non-native accents).

Careers & open positions

Open positions indexed from verified career portals and applicant tracking systems.

PositionDepartmentLocationPosted
Senior Software Engineer - Microservices(Senior)EngineeringHume AI HQ
Senior Technical Writer(Senior)EngineeringHume AI HQ
Senior Software Engineer - Frontend UI(Senior)EngineeringHume AI HQ
Product Designer(Mid-Level)ProductHume AI HQ
Senior Software Engineer - Data Plane(Senior)EngineeringHume AI HQ

Explore company relationships

Questions about Hume AI

What is Hume AI?

Hume AI is a private B2B software company that provides evaluation, simulation, expression measurement and human-feedback infrastructure for voice AI systems.

Who uses Hume AI?

Voice AI engineers, machine learning teams, AI researchers and product teams use Hume AI to test, benchmark and improve conversational systems.

How does Hume AI make money?

Hume AI makes money through software subscriptions, API usage-based pricing and paid custom data and human evaluation services.

Sources & coverage

This profile uses public, official and technically observable information. Missing information does not prove that a product or relationship does not exist. The list below does not imply that every profile statement has been verified.

3 publicly documented primary sources and citations linked across the market graph.

Start with a company or market. Track what changes.

Create your Polaris7 workspace to explore extended company profiles, organize relevant players into Watchlists, and receive focused briefings when material changes occur.