About Speaker Embeddings

[Shmuel-Gruel]

Hello again, I notice that for training, if using the pretrained speaker encoder and SR, the same speaker embedding is used regardless of the SR augmentation.
(edit: also the audio file and spectrogram are the same, regardless of SR)
I just wanted to double check if this is by design?

Thanks so much for helping everyone :)

[OlaWod]

Yes. SR is only used to help content representation learning, i.e., multiple SR augmented wavs correspond to one same content representation.

[Shmuel-Gruel]

Excellent, thank you.
One more question, are the test.txt files used anywhere?

[OlaWod]

Nowhere.
It is only used for experimental test. We randomly select seen speaker's unseen utterances from this file for test.

[Shmuel-Gruel]

Understood! Thank you for your insight