README.md

metadata

language: pt
datasets:
  - common_voice
metrics:
  - wer
tags:
  - audio
  - automatic-speech-recognition
  - speech
  - xlsr-fine-tuning-week
license: apache-2.0
model-index:
  - name: Wav2Vec2 Large 53 Portugese by Gunjan Chhablani
    results:
      - task:
          name: Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice pt
          type: common_voice
          args: pt
        metrics:
          - name: Test WER
            type: wer
            value:
              wer_result_on_test: null

Wav2Vec2-Large-XLSR-53-Portuguese

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Portuguese using the Common Voice dataset. When using this model, make sure that your speech input is sampled at 16kHz.

Usage

The model can be used directly (without a language model) as follows:

import torch
import torchaudio
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

test_dataset = load_dataset("common_voice", "pt", split="test[:2%]")

processor = Wav2Vec2Processor.from_pretrained("gchhablani/wav2vec2-large-xlsr-pt")
model = Wav2Vec2ForCTC.from_pretrained("gchhablani/wav2vec2-large-xlsr-pt")

resampler = torchaudio.transforms.Resample(48_000, 16_000)

# Preprocessing the datasets.
# We need to read the aduio files as arrays
def speech_file_to_array_fn(batch):
  \tspeech_array, sampling_rate = torchaudio.load(batch["path"])
  \tbatch["speech"] = resampler(speech_array).squeeze().numpy()
  \treturn batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
  \tlogits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)

print("Prediction:", processor.batch_decode(predicted_ids))
print("Reference:", test_dataset["sentence"][:2])

Evaluation

The model can be evaluated as follows on the Portuguese test data of Common Voice.

import torch
import torchaudio
from datasets import load_dataset, load_metric
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import re

test_dataset = load_dataset("common_voice", "pt", split="test") #TODO: replace {lang_id} in your language code here. Make sure the code is one of the *ISO codes* of [this](https://huggingface.co/languages) site.
wer = load_metric("wer")

processor = Wav2Vec2Processor.from_pretrained("gchhablani/wav2vec2-large-xlsr-pt") #TODO: replace {model_id} with your model id. The model id consists of {your_username}/{your_modelname}, *e.g.* `elgeish/wav2vec2-large-xlsr-53-arabic`
model = Wav2Vec2ForCTC.from_pretrained("gchhablani/wav2vec2-large-xlsr-pt") #TODO: replace {model_id} with your model id. The model id consists of {your_username}/{your_modelname}, *e.g.* `elgeish/wav2vec2-large-xlsr-53-arabic`
model.to("cuda")

chars_to_ignore_regex = '[\\,\\?\\.\\!\\-\\;\\:\\"\\“\\'\\�]'
resampler = torchaudio.transforms.Resample(48_000, 16_000)

# Preprocessing the datasets.
# We need to read the aduio files as arrays
def speech_file_to_array_fn(batch):
\t  batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
\t\tspeech_array, sampling_rate = torchaudio.load(batch["path"])
  \tbatch["speech"] = resampler(speech_array).squeeze().numpy()
  \treturn batch

test_dataset = test_dataset.map(speech_file_to_array_fn)

# Preprocessing the datasets.
# We need to read the aduio files as arrays
def evaluate(batch):
  \tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

  \twith torch.no_grad():
  \t\tlogits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits

  \tpred_ids = torch.argmax(logits, dim=-1)
  \tbatch["pred_strings"] = processor.batch_decode(pred_ids)
  \treturn batch

result = test_dataset.map(evaluate, batched=True, batch_size=8)

print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))

Test Result: XX.XX % # TODO: write output of print here. IMPORTANT: Please remember to also replace {wer_result_on_test} at the top of with this value here. tags.

Training

The Common Voice train and validation datasets were used for training. The script used for training can be found here. The parameters passed were:

#!/usr/bin/env bash
python run_common_voice.py \\
    --model_name_or_path="facebook/wav2vec2-large-xlsr-53" \\
    --dataset_config_name="pt" \\
    --output_dir=/workspace/output_models/pt/wav2vec2-large-xlsr-pt \\
    --cache_dir=/workspace/data \\
    --overwrite_output_dir \\
    --num_train_epochs="30" \\
    --per_device_train_batch_size="32" \\
    --per_device_eval_batch_size="32" \\
    --evaluation_strategy="steps" \\
    --learning_rate="3e-4" \\
    --warmup_steps="500" \\
    --fp16 \\
    --freeze_feature_extractor \\
    --save_steps="500" \\
    --eval_steps="500" \\
    --save_total_limit="1" \\
    --logging_steps="500" \\
    --group_by_length \\
    --feat_proj_dropout="0.0" \\
    --layerdrop="0.1" \\
    --gradient_checkpointing \\
    --do_train --do_eval \\