Generalized End-to-End Loss for Speaker Verification

Code

• github.com/google/speaker-id/tree/master/lingvo
  Officialnone★ 0
• github.com/coqui-ai/TTS
  pytorch★ 44,894
• github.com/PaddlePaddle/PaddleSpeech
  paddle★ 12,564
• github.com/piotrkawa/audio-deepfake-source-tracing
  pytorch★ 33
• github.com/JeffT13/rd-diarization
  pytorch★ 0
• github.com/Aurora11111/voiceprint
  pytorch★ 0
• github.com/Aurora11111/speaker-recognition-pytorch
  pytorch★ 0
• github.com/JeffT13/VoiceEncoder
  pytorch★ 0
• github.com/resemble-ai/Resemblyzer
  pytorch★ 0
• github.com/luomingshuang/GE2E-SV-TI-Chinese-LMS
  pytorch★ 0

Abstract

In this paper, we propose a new loss function called generalized end-to-end (GE2E) loss, which makes the training of speaker verification models more efficient than our previous tuple-based end-to-end (TE2E) loss function. Unlike TE2E, the GE2E loss function updates the network in a way that emphasizes examples that are difficult to verify at each step of the training process. Additionally, the GE2E loss does not require an initial stage of example selection. With these properties, our model with the new loss function decreases speaker verification EER by more than 10%, while reducing the training time by 60% at the same time. We also introduce the MultiReader technique, which allows us to do domain adaptation - training a more accurate model that supports multiple keywords (i.e. "OK Google" and "Hey Google") as well as multiple dialects.

Tasks

Benchmark Results

Reproductions