cpierse/wav2vec2-large-xlsr-53-esperanto

Wav2Vec2-Large-XLSR-53-eo

Fine-tuned facebook/wav2vec2-large-xlsr-53 on esperanto 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", "eo", split="test[:2%]") 
processor = Wav2Vec2Processor.from_pretrained("cpierse/wav2vec2-large-xlsr-53-esperanto") 
model = Wav2Vec2ForCTC.from_pretrained("cpierse/wav2vec2-large-xlsr-53-esperanto") 

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):
   speech_array, sampling_rate = torchaudio.load(batch["path"])
   batch["speech"] = resampler(speech_array).squeeze().numpy()
   return 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():
   logits = 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 Esperanto test data of Common Voice.

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

def chunked_wer(targets, predictions, chunk_size=None):
    if chunk_size is None: return jiwer.wer(targets, predictions)
    start = 0
    end = chunk_size
    H, S, D, I = 0, 0, 0, 0
    while start < len(targets):
        chunk_metrics = jiwer.compute_measures(targets[start:end], predictions[start:end])
        H = H + chunk_metrics["hits"]
        S = S + chunk_metrics["substitutions"]
        D = D + chunk_metrics["deletions"]
        I = I + chunk_metrics["insertions"]
        start += chunk_size
        end += chunk_size
    return float(S + D + I) / float(H + S + D)

test_dataset = load_dataset("common_voice", "eo", 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("cpierse/wav2vec2-large-xlsr-53-esperanto")
model = Wav2Vec2ForCTC.from_pretrained("cpierse/wav2vec2-large-xlsr-53-esperanto")
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):
   batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
   speech_array, sampling_rate = torchaudio.load(batch["path"])
   batch["speech"] = resampler(speech_array).squeeze().numpy()
   return 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):
   inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

   with torch.no_grad():
      logits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits

   pred_ids = torch.argmax(logits, dim=-1)
   batch["pred_strings"] = processor.batch_decode(pred_ids)
   return batch

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

print("WER: {:2f}".format(100 * chunked_wer(predictions=result["pred_strings"], targets=result["sentence"],chunk_size=2000)))

Test Result: 12.31 %

Training

The Common Voice train, validation datasets were used for training.