Deciphering animal emotions using machine learning and vocalisations
summary
A large comparative study measures vocal signals and heart rate across ungulates, meerkats, little auks and marmosets, extracts spectrogram features and trains a CNN on images of calls and an XGBoost on acoustic parameters, producing a farm-ready classifier that separates 19 contexts with 92% accuracy (boosted to 96% after noise augmentation). Field deployments on farms reveal wide variation — some organic sites hit 56% stress calls — and detection in noisy recordings still requires better call-finders before a practical emotion translator works reliably.
AI-written summary · about
chapters (8)
- 0:00Animal emotion research intro
- 1:53Vocalisations beyond emotions
- 7:12Defining animal emotions
- 16:46Cross-species vocal patterns
- 26:22AI emotion classifier for pigs
- 34:58Expanding to more species
- 37:53Future steps and validation
- 39:18Applications and impact
related talks
15mRecognizing Emotional Expression in Game Streams
29mBioacoustics and Machine Learning as Key Tools in Conservation
25mThe Sound of Life
39mThe art and science of making computers understand sentiment and emotion
21mEmotion Estimation for Social Robotics
59mTechnical Workshop: Sarcastically Speaking: Unlocking Multi-modal Sentiment Analysis With NLP and Facial Expressions
32mDecoding Emotions from Art to Memes
39mOverview and Marine Bioacoustics and Perch Model