Hume VoiceEQ

What it does

Hume VoiceEQ is an evaluation platform for voice models and agents. It takes real conversation data and live audio, including voice interactions streamed over WebRTC, WebSocket, or SIP as well as batch audio, and produces generated evaluation suites, stress-test results, failure diagnoses, and checks that improvements hold up over time. Measurements combine audio-based expression, timing, and conversational signals with ratings from trained human raters.

Teams use VoiceEQ to find the conversations that break a voice system, group them into the use cases and failure modes behind them, and rerun the same tests after each change. Model builders can compare checkpoints and diagnose weaknesses across TTS, speech-to-speech, ASR, and cascaded systems. Agent teams can choose models, define quality criteria for their specific use case, and validate realistic interactions. Because scoring starts from the audio rather than the transcript alone, it captures how a response sounds as well as what it says.

VoiceEQ is aimed at teams building or deploying voice models and agents who need evaluation beyond task success. Access is by request through the product page rather than self-serve signup, and pricing is handled directly with Hume. Results depend on Hume's measurement models and its human rater pool, so teams should validate them against their own conversations.

What it can do

• Generate targeted evaluation suites from real-world conversation data, use cases, and success criteria.
• Stress-test voice AI under conditions such as accents, background audio, interruptions, and code-switching.
• Diagnose failures by clustering calls into use cases and failure modes, using expression measurement from the audio.
• Combine automated expression and timing metrics with human judgment from trained raters via the Human Feedback API.
• Compare checkpoints and benchmark TTS, speech-to-speech, ASR, and cascaded models.
• Run repeatable regression testing to validate improvements.

Works with

• WebRTC
• WebSocket
• SIP / live phone line
• Batch audio

Before you choose

• Access is by request, so plan for an onboarding conversation before running evaluations.
• Human ratings and expression measurements are core to the offering; the Expression Measurement API is currently in limited release, so request access to it separately if you need direct endpoint use.

Listed under

• Coding
• AI voice agents
• Benchmarking
• Developer tools
• Emotion measurement
• Evaluation
• Speech

Get started

Use the 'Request access to VoiceEQ' option on the product page.

1. Request access to VoiceEQ through the product page.
2. Connect conversation data or live audio over WebRTC, WebSocket, SIP, or batch audio.
3. Define success criteria and use cases, then run generated evaluation suites and review diagnoses.

Confirm the applicable plan, usage charges and commercial terms with the provider.

We couldn’t find published prices for Hume VoiceEQ. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Measure and improve real-world voice AI performance | Hume AI ↗www.hume.ai
• VoiceEQ Platform | Hume AI ↗www.hume.ai
• Expression APIs | Hume AI ↗www.hume.ai
• Expression Measurement API | Hume API ↗dev.hume.ai

Company website

hume.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.