metadata

language:
  - ml
license: apache-2.0
tags:
  - whisper-event
datasets:
  - mozilla-foundation/common_voice_11_0
  - google/fleurs
  - thennal/IMaSC
  - thennal/ulca_ml
  - thennal/msc
  - thennal/indic_tts_ml
metrics:
  - wer
model-index:
  - name: Whisper Medium Malayalam - Thennal D K
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice 11.0
          type: mozilla-foundation/common_voice_11_0
          config: ml
          split: test
          args: ml
        metrics:
          - name: Wer
            type: wer
            value: 42.98850574712644
          - name: Cer
            type: cer
            value: 10.390585878818229

Whisper Medium Malayalam - Thennal D K

This model is a fine-tuned version of openai/whisper-medium on a combined dataset sourced from IMaSC, SMC, Indic TTS, FLEURS (train set), Common Voice 11 (train + other set), OpenSLR, and ULCA. It achieves the following results on the evaluation set (Common Voice 11 test split):

Loss: 0.0730
WER: 42.9886
CER: 10.3906

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
train_batch_size: 32
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 4000
mixed_precision_training: Native AMP

Framework versions

Transformers 4.26.0.dev0
Pytorch 1.13.0+cu117
Datasets 2.7.1.dev0
Tokenizers 0.13.2