Audio AI
in Health:
Voice Biomarkers
Detecting diseases from the voice? What initially sounds like a futuristic scenario could soon become reality thanks to modern AI technology.
Diseases Affect
Voice Production
Our entire body influences the sound of our voice. Chest, throat and head must be coordinated when we make a sound. Many diseases affect this complex process. They generate specific features in the speech, which AI voice analysis can detect. These so called biomarkers can be used to distinguish a sick person from a healthy one.
Step 1:
Motoric Functionalities
Numerous organs and muscle groups are involved in speech production, e.g. the vocal folds, tongue and cheek muscles. Disease-related motoric disorders are therefore directly noticeable in the speech.
Step 2:
Cognitive Control
and Voice Analysis
Verbal articulation is very demanding on the brain. The fine motor muscles in particular must be perfectly controlled in order to clearly speak syllables and words. Cognitive disorders therefore show characteristic symptoms in the flow of speech early on.
Step 3:
Physiological
Aspects of the Voice
The sound of the human voice is directly influenced by the current anatomical and physical status of the speaker. For example, the breathing frequency will increase while doing sports.
COVID-19 negative
COVID-19 positive
Detecting COVID-19 with up to 82%
Respiratory diseases like COVID-19 change voice characteristics such as intonation, intensity and tempo. Trained with sound and voice data, audEERING’s AI systems can recognize these changes.
In cooperation with the University of Augsburg, we carry out a research study on a voice-based COVID-19 test. We collect data samples, audio recordings, and related health data. The result: Our AI system can then achieve an accuracy of up to 82%.
Accurate Performance
Detecting COVID-19
from the voice
Data Security
Highest data security standards
for health related data
Immediate Results
Fast results based on a few
voice recordings
AI SoundLab: Web-based Health Platform
When developing health AI, precision is paramount. For health applications, the best data quality and the highest possible data security are required at the same time
Our AI SoundLab platform processes data from healthy and sick people under the highest security standards in Germany. Health data enriched by AI technology enables quicker diagnosis from completely human experts. AI SoundLab can monitor health developments, provide analysis for diagnosis, and collect data for the development of voice biomarkers.
Detecting Parkinson’s with emteq
Emteq is a hardware manufacturer of emotion sensor technology. Their emteq glasses analyze and monitor speech and other biometric data over time. audEERING contributes the audio intelligence technology for speech analysis. Together we were able to identify a patient with Parkinson’s with an accuracy of 92 percent.
Health AI:
The Future of Medicine
Health AI will never replace the doctor. However, we believe that AI-based health applications will be the next big step in medicine. Faster, cheaper and more accurate diagnoses for millions of people worldwide.
-
SCALABLE: mass deployable sensor,
only standard microphone needed -
EASY ACCESS: for anyone, anytime,
anywhere, easier access to specialist know-how -
EASY IMPLEMENTATION: in care processes,
telemedicine and apps -
SHORT & LONG TERM APPLICATION:
on-the-spot and monitoring -
EASY FUSION: combination with e.g. vision or patches,
increased prediction probability - SECURE: data stored on device, fingerprint extraction
- ENVIRONMENTAL BENEFITS: less waste
audEERING’s Health
Platform AI SoundLab
More info ›
Voice Study
for COVID-19
In collaboration with the University of Augsburg audEERING conducted a study about developing a voice-based COVID-19 test. We have collected thousands of voice recordings in this scientific study.