| --- |
| language: |
| - ro |
| license: apache-2.0 |
| base_model: openai/whisper-small |
| tags: |
| - hf-asr-leaderboard |
| - generated_from_trainer |
| datasets: |
| - iulik-pisik/horoscop_neti |
| - iulik-pisik/audio_vreme |
| metrics: |
| - wer |
| model-index: |
| - name: Whisper Small - finetuned on weather and horoscope |
| results: |
| - task: |
| name: Automatic Speech Recognition |
| type: automatic-speech-recognition |
| dataset: |
| name: Vreme ProTV and Horoscop Neti |
| type: iulik-pisik/audio_vreme |
| config: default |
| split: test |
| args: 'config: ro, split: test' |
| metrics: |
| - name: Wer |
| type: wer |
| value: 8.51 |
| pipeline_tag: automatic-speech-recognition |
| --- |
| |
|
|
|
|
|
|
|
|
| # Whisper Small - finetuned on weather and horoscope |
| This model is a fine-tuned version of [openai/whisper-small](openai/whisper-small) on the Vreme ProTV and Horoscop Neti datasets. |
| It achieves the following results on the evaluation set: |
|
|
| - Loss: 0.0004 |
| - Wer: 8.51 |
|
|
|
|
| ## Model description |
|
|
| This is a fine-tuned version of the Whisper Small model, specifically adapted for Romanian language Automatic Speech Recognition (ASR) |
| in the domains of weather forecasts and horoscopes. The model has been trained on two custom datasets to improve its performance |
| in transcribing Romanian speech in these specific contexts. |
|
|
| ## Training procedure |
|
|
| The model was fine-tuned using transfer learning techniques on the pre-trained Whisper Small model. |
| Two custom datasets were used: audio recordings of weather forecasts and horoscopes in Romanian. |
|
|
| ### Training hyperparameters |
|
|
| The following hyperparameters were used during training: |
| - learning_rate: 1e-05 |
| - train_batch_size: 16 |
| - eval_batch_size: 8 |
| - 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: 3000 |
| - mixed_precision_training: Native AMP |
| |
| ### Training results |
| |
| | Epoch | Step | Validation Loss | WER | |
| |:-----:|:----:|:---------------:|:-------:| |
| | 3.85 | 1000 | 0.0332 | 9.1945 | |
| | 7.69 | 2000 | 0.0035 | 10.845 | |
| | 11.54 | 3000 | 0.0005 | 8.4679 | |
| | 15.38 | 4000 | 0.0004 | 8.5127 | |
| |
| |
| |
| ### Framework versions |
| |
| - Transformers 4.39.2 |
| - Pytorch 2.2.1+cu121 |
| - Datasets 2.18.0 |
| - Tokenizers 0.15.2 |
