CoolFace
Modelpublic

Minu1234/whisper-small-phonemes

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face
1likes5downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

Whisper Small En - Minu Kumari

This model is a fine-tuned version of openai/whisper-small on the Polly Generated Voice v0.1 dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.6189
  • —Wer: 32.9596

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • —learning_rate: 1e-05
  • —trainbatchsize: 16
  • —evalbatchsize: 8
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 5
  • —training_steps: 100

Training results

Training LossEpochStepValidation LossWer
3.35821.054.3222100.4484
2.89582.0102.6835100.6726
1.6793.0151.8785100.8969
1.08054.0201.420098.4305
0.72815.0251.107463.2287
0.49166.0300.902766.5919
0.34747.0350.789155.1570
0.25818.0400.725565.9193
0.19969.0450.684976.4574
0.158810.0500.659869.0583
0.129511.0550.649164.3498
0.108112.0600.637754.4843
0.093313.0650.632334.5291
0.082914.0700.628034.9776
0.074915.0750.627635.2018
0.069316.0800.626433.8565
0.06517.0850.622534.3049
0.06218.0900.620733.6323
0.0619.0950.619833.4081
0.058520.01000.618932.9596

Framework versions

  • —Transformers 4.34.1
  • —Pytorch 2.1.0+cu118
  • —Datasets 2.14.5
  • —Tokenizers 0.14.1