license: apache-2.0 tags:

• generated_from_trainer metrics:
• wer model-index:
• name: whisper-large-v2-spanish results: []

whisper-large-v2-spanish

This model is a fine-tuned version of openai/whisper-large-v2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1466
• Wer: 0.0855

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: 16
• 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: 25000
• mixed_precision_training: Native AMP

Training results

Transcription:

from datasets import load_dataset, Audio
import torch
from transformers import WhisperProcessor, WhisperForConditionalGeneration

# device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# load the model
processor = WhisperProcessor.from_pretrained("clu-ling/whisper-large-v2-spanish")
model = WhisperForConditionalGeneration.from_pretrained("clu-ling/whisper-large-v2-spanish").to(device)
forced_decoder_ids = processor.get_decoder_prompt_ids(language="es", task="transcribe")

# load the dataset
commonvoice_eval = load_dataset("mozilla-foundation/common_voice_11_0", "es", split="validation", streaming=True)
commonvoice_eval = commonvoice_eval.cast_column("audio", Audio(sampling_rate=16000))
sample = next(iter(commonvoice_eval))["audio"]

# features and generate token ids
input_features = processor(sample["array"], sampling_rate=sample["sampling_rate"], return_tensors="pt").input_features
predicted_ids = model.generate(input_features.to(device), forced_decoder_ids=forced_decoder_ids)

# decode
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)

print(transcription)

Evaluation:

Evaluates this model on mozilla-foundation/common_voice_11_0 test split.

from transformers.models.whisper.english_normalizer import BasicTextNormalizer
from datasets import load_dataset, Audio
import evaluate
import torch
import re
from transformers import WhisperProcessor, WhisperForConditionalGeneration

# device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# metric
wer_metric = evaluate.load("wer")

# model
processor = WhisperProcessor.from_pretrained("clu-ling/whisper-large-v2-spanish")
model = WhisperForConditionalGeneration.from_pretrained("clu-ling/whisper-large-v2-spanish")

# dataset
dataset = load_dataset("mozilla-foundation/common_voice_11_0", "es", split="test", )#cache_dir=args.cache_dir
dataset = dataset.cast_column("audio", Audio(sampling_rate=16000))

#for debuggings: it gets some examples
#dataset = dataset.shard(num_shards=10000, index=0)
#print(dataset)
   
def normalize(batch):
  batch["gold_text"] = whisper_norm(batch['sentence'])
  return batch

def map_wer(batch):
  model.to(device)
  forced_decoder_ids = processor.get_decoder_prompt_ids(language = "es", task = "transcribe")
  inputs = processor(batch["audio"]["array"], sampling_rate=batch["audio"]["sampling_rate"], return_tensors="pt").input_features
  with torch.no_grad():
    generated_ids = model.generate(inputs=inputs.to(device), forced_decoder_ids=forced_decoder_ids)
    transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
  batch["predicted_text"] = whisper_norm(transcription)
  return batch

# process GOLD text
processed_dataset = dataset.map(normalize)
# get predictions
predicted = processed_dataset.map(map_wer)

# word error rate
wer = wer_metric.compute(references=predicted['gold_text'], predictions=predicted['predicted_text'])
wer = round(100 * wer, 2)
print("WER:", wer)

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.8.1.dev0
• Tokenizers 0.13.2