This is the PyTorch implementation of speaker embedding (d-vector) trained with GE2E loss.
The original paper about GE2E loss could be found here: Generalized End-to-End Loss for Speaker Verification
To use the script provided here, you have to organize your raw data in this way:
- all utterances from a speaker should be put under a directory (speaker directory)
- all speaker directories should be put under a directory (root directory)
- speaker directory can have subdirectories and utterances can be placed under subdirectories
You have to specify two things here:
- use
-eor--extensionsto specify the extensions of utterances to be extracted and separate them with commas e.g.wav,flac,mp3(do not leave SPACES in between) - use
-sor--save_dirto specify the directory for saving processed utterances
And a good thing about this script is that you can extract utterances from multiple root directories. For example:
python prepare.py -s data -e wav,flac VCTK-Corpus/wav48 LibriSpeech/train-clean-360Only DATA_DIR and MODEL_DIR have to be specified here.
For more details, check the usage with python train.py -h.
python train.py DATA_DIR MODEL_DIR \
-i 1000000 \
-s 10000 \
-d 100000 \
-n 64 \
-m 10 \
-l 128During training, event logs will be put under MODEL_DIR.
To continue the training from a saved checkpoint, just specify the checkpoint path with -c or --checkpoint_path.
Note that you can still specify other optional arguments because they might be different from the ones in the previous training.
python train.py DATA_DIR MODEL_DIR \
-i 500000 \
-s 100000 \
-d 100000 \
-n 64 \
-m 10 \
-l 128 \
-c SAVED_CHECKPOINTThe dimension reduction result (using t-SNE) of some utterances from LibriSpeech.
The GE2E-Loss module is borrowed from cvqluu/GE2E-Loss.