
audEERING
An empathic AI speech SDK spun out of the Technical University of Munich

Tool details
audEERING is an empathic AI company spun out of the Technical University of Munich (TUM). It parses emotion, arousal and physiological state in the voice into structured signals and delivers them to B2B customers as SDKs and APIs. It is neither a consumer app nor a general-purpose AI product, but an audio understanding engine built on in-house acoustic features plus deep models.
Core capabilities
- devAIce®: SDK, Web API and plug-in form factors bundle more than ten composable modules and return emotion dimensions and categories in one unified output
- devAIce® XR: Unity / Unreal plug-ins that add tone-of-voice emotion sensing directly to VR/AR and game engines
- AI SoundLab: web-based audio data collection and voice biomarker analysis, with data stored in Germany and GDPR as the headline promise
- On-device real-time inference: the SDK covers Windows/Linux/macOS/Android/iOS and ARM, and runs in real time even on Raspberry Pi-class hardware
- openSMILE toolchain: it maintains this de facto standard feature library for affective computing, offering more than six thousand dimensions of acoustic features
- Multi-dimensional output: arousal, valence and dominance continuous values are returned alongside four emotion labels
Where it fits Call-centre quality assurance, headphone and in-car interaction, mental-health screening research, VR social emotion feedback, and psychology or linguistics projects that need bulk audio annotation; teams building a consumer voice assistant or plain speech-to-text need not consider it.
Distinctive strengths Its academic foundation is among the deepest in the category: the team includes pioneers of speech emotion computing such as Björn Schuller, it has long organised the INTERSPEECH ComParE challenges, and it does not merely wrap someone else's API. The tone analysis in Jabra Engage AI is driven by it.
Editorial take Technical substance and verifiable benchmarks are rare here, and the official whitepaper reports explicit CCC and cross-language UAR figures. The trade-off is a somewhat high barrier to entry: subscriptions and SDK licences are not publicly priced, self-serve purchase is limited to the Prepaid minute packages, and the mobile wrappers were removed in version 4.0. To embed emotion recognition into your own product, this is one of the few dependable choices; to validate an idea quickly, start with the open-source openSMILE.
Pricing
Subscription€35 (Prepaid Entry, 600 minutes)Prepaid minutes package for the Web API, including the BASE package and 1 module of your choice, billed by the minute
The same package at a larger volume, with a lower price per minute
Subscriptions and licences for the SDK, XR plug-ins and AI SoundLab are quoted by model combination, deployment mode and usage volume
Application-based discount or fee waiver, limited to non-commercial research; proof of enrolment or employment and a project description are required, and outputs must credit audEERING
Additional notes
Subscriptions and SDK licences are not publicly priced and require a quote; self-serve purchase covers only the two Prepaid Web API minute packages, with no free tier and no trial. openSMILE and the open-source speech models on Hugging Face are free for research use, and any commercial product integration requires a commercial development licence.
FAQ
User reviews
User Reviews
No reviews yet
No one has reviewed this tool yet