whisper-large-icelandic-30k-steps-1000h

The "whisper-large-icelandic-30k-steps-1000h" is an acoustic model suitable for Automatic Speech Recognition in Icelandic. It is the result of fine-tuning the model "openai/whisper-large" for 30,000 steps with around 1000 hours of Icelandic data developed by the Language and Voice Laboratory. Most of the data is available at public repositories such as LDC, OpenSLR or Clarin.is

The specific list of corpora used to fine-tune the model is:

• Samrómur 21.05 (114h34m)
• Samrómur Children (127h25m)
• Malrómur (119hh03m)
• Althingi Parliamentary Speech (514h29m)
• L2-Speakers Data (125h55m) Unpublished material

The fine-tuning process was performed during April (2023) in the servers of the Language and Voice Laboratory (https://lvl.ru.is/) at Reykjavík University (Iceland) by Carlos Daniel Hernández Mena.

Evaluation

import torch
from transformers import WhisperForConditionalGeneration, WhisperProcessor

#Load the processor and model.
MODEL_NAME="language-and-voice-lab/whisper-large-icelandic-30k-steps-1000h"
processor = WhisperProcessor.from_pretrained(MODEL_NAME)
model = WhisperForConditionalGeneration.from_pretrained(MODEL_NAME).to("cuda")

#Load the dataset
from datasets import load_dataset, load_metric, Audio
ds=load_dataset("language-and-voice-lab/samromur_children",split='test')

#Downsample to 16kHz
ds = ds.cast_column("audio", Audio(sampling_rate=16_000))

#Process the dataset
def map_to_pred(batch):
    audio = batch["audio"]
    input_features = processor(audio["array"], sampling_rate=audio["sampling_rate"], return_tensors="pt").input_features
    batch["reference"] = processor.tokenizer._normalize(batch['normalized_text'])

    with torch.no_grad():
        predicted_ids = model.generate(input_features.to("cuda"))[0]
    
    transcription = processor.decode(predicted_ids)
    batch["prediction"] = processor.tokenizer._normalize(transcription)
    
    return batch
    
#Do the evaluation
result = ds.map(map_to_pred)

#Compute the overall WER now.
from evaluate import load

wer = load("wer")
WER=100 * wer.compute(references=result["reference"], predictions=result["prediction"])
print(WER)

Test Result: 7.743795695602924

BibTeX entry and citation info

When publishing results based on these models please refer to:

@misc{mena2023whisperlarge30kicelandic,
      title={Acoustic Model in Icelandic: whisper-large-icelandic-30k-steps-1000h.}, 
      author={Hernandez Mena, Carlos Daniel},
      url={https://huggingface.co/language-and-voice-lab/whisper-large-icelandic-30k-steps-1000h},
      year={2023}
}

Acknowledgements

Thanks to Jón Guðnason, head of the Language and Voice Lab for providing computational power to make this model possible.

We also want to thank to the "Language Technology Programme for Icelandic 2019-2023" which is managed and coordinated by Almannarómur, and it is funded by the Icelandic Ministry of Education, Science and Culture. This model is an unexpected result of all the resources gathered by the Programme.

Special thanks to Björn Ingi Stefánsson for setting up the configuration of the server where this model was trained.