nmeripo / Text-Independant-Speaker-Identification

Implementation of Google AI's GE2E (PyTorch)

Text Independant Speaker Verification

Implementation of Google's GE2E

Dataset

Get VoxCeleb1 Dataset from here

Get QUT-NOISE Database from here

Tensorboard visualization

tensorboard --logdir runs

Results

EER = 0.4735

TODO

• Implement gradient scaling and gradient clipping
• Tensorboard Integration
• Comments/Documentation * Upload weights
• Implement Evaluation
• Implement EER
• Unknown Noise & SNR Analysis