kaistmm/seed-pytorch

[INTERSPEECH 2025] Official code for "SEED: Speaker Embedding Enhancement Diffusion Model"

28
/ 100
Experimental

This project helps speech technologists improve the accuracy of speaker recognition systems when dealing with noisy audio. It takes existing speaker embeddings (digital representations of a voice) from a pre-trained model and refines them to be more robust to background noise. The result is more accurate speaker identification, even in challenging acoustic environments, benefiting professionals working on voice authentication or speaker diarization.

Use this if your speaker recognition or verification system struggles with performance degradation due to environmental noise and you need to enhance the robustness of your speaker embeddings.

Not ideal if your primary goal is to train a speaker recognition model from scratch, as this tool focuses on enhancing existing speaker embeddings rather than building them.

speaker-recognition voice-biometrics speech-enhancement audio-processing speech-technology
No License No Package No Dependents
Maintenance 6 / 25
Adoption 8 / 25
Maturity 7 / 25
Community 7 / 25

How are scores calculated?

Stars

57

Forks

3

Language

Python

License

Last pushed

Nov 03, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/kaistmm/seed-pytorch"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.