title: Potato — Audio Classification
emoji: 🥔
colorFrom: pink
colorTo: red
sdk: docker
app_port: 7860
pinned: false
license: apache-2.0
tags:
- annotation
- potato
- multimodal
- audio
Potato — Audio Classification
Classify audio clips (emotion / event / speaker).
A live demo of Potato — the free, self-hosted annotation platform for NLP, agentic, and GenAI research, configured entirely through YAML. Visit www.potatoannotator.com for docs, the schema gallery, and more demos.
Try it out
- Enter any username to log in (no password required).
- Read the item shown in the main panel.
- Annotate using the schemes on the right.
- Click Next to continue.
Run your own copy: this demo is generated from a project in the Potato repo. Clone it and start that project directly — no HuggingFace account needed, and no concurrency limit:
python potato/flask_server.py start <example>/config.yaml -p 8000. Duplicating a Space instead requires a HuggingFace PRO, Team or Enterprise plan, because Docker Spaces run on compute.
Annotations in this demo are ephemeral. To collect and keep data, deploy your own Space — see the deployment guide.
About Potato
Potato supports 20+ annotation types — text, spans, images, audio, video, documents, and agent traces — with AI-assisted labeling, quality control, and adjudication.
🥔 Website: www.potatoannotator.com · Documentation · GitHub · All demos