ginic/gender_split_100_female_5_wav2vec2-large-xlsr-53-buckeye-ipa

license: mit language:

• —en pipeline_tag: automatic-speech-recognition ---

About

This model was created to support experiments for evaluating phonetic transcription with the Buckeye corpus as part of https://github.com/ginic/multipa. This is a version of facebook/wav2vec2-large-xlsr-53 fine tuned on a specific subset of the Buckeye corpus. For details about specific model parameters, please view the config.json here or training scripts in the scripts/buckeye_experiments folder of the GitHub repository.

Experiment Details

Still training with a total amount of data equal to half the full training data (4000 examples), vary the gender split 30/70, but draw examples from all individuals. Do 5 models for each gender split with the same model parameters but different data seeds.

Goals:

• —Determine how different in gender split in training data affects performance

Params to vary:

• —percent female (--percent_female) [0.0, 0.3, 0.7, 1.0]
• —training seed (--train_seed)