Wav2vec2-swedish trained on the common voice dataset.
WER 0.511916 on a dataset of 402mb.
Link to notebook for training.
https://colab.research.google.com/drive/1KkD4PeZwnIwxxxOP1bUE7XTZMK7-SzRj?usp=sharing
Wav2vec2-swedish trained on the common voice dataset.
WER 0.511916 on a dataset of 402mb.
Link to notebook for training.
https://colab.research.google.com/drive/1KkD4PeZwnIwxxxOP1bUE7XTZMK7-SzRj?usp=sharing