Implementation of automatic speech rating systems for second language (L2) learners of Finnish and Finland Swedish
The ASR systems were trained by run_asr_SLT_kfold.py
The hidden wav2vec 2.0 representations were extracted by extract_static_w2v2_features.py
The speech rating wav2vec 2.0 systems were trained by run_classification_SLT_kfold.py
speech_rating_FI.ipynb and speech_rating_SV.ipynb contain the results of speech rating experiments for L2 Finnish and Finland Swedish, respectively.