language	datasets	metrics	tags	license	model-index
en	common_voice	wer cer	audio automatic-speech-recognition speech xlsr-fine-tuning-week	apache-2.0	name results XLSR Wav2Vec2 English by Jonatas Grosman task dataset metrics name type Speech Recognition automatic-speech-recognition name type args Common Voice en common_voice en name type value Test WER wer 18.98 name type value Test CER cer 8.29

Wav2Vec2-Large-XLSR-53-English

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

The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint

Usage

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

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

LANG_ID = "en"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-english"
SAMPLES = 10

test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = batch["sentence"].upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"], 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)
predicted_sentences = processor.batch_decode(predicted_ids)

for i, predicted_sentence in enumerate(predicted_sentences):
    print("-" * 100)
    print("Reference:", test_dataset[i]["sentence"])
    print("Prediction:", predicted_sentence)

Reference	Prediction
"SHE'LL BE ALL RIGHT."	SHE'LL BE ALL RIGHT
SIX	SIX
"ALL'S WELL THAT ENDS WELL."	ALL AS WELL THAT ENDS WELL
DO YOU MEAN IT?	DO YOU MEAN IT
THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE, BUT STILL CAUSES REGRESSIONS.	THE NEW PATCH IS LESS INVASIVE THAN THE OLD ONE BUT STILL CAUSES REGRESSION
HOW IS MOZILLA GOING TO HANDLE AMBIGUITIES LIKE QUEUE AND CUE?	HOW IS MOSLILLAR GOING TO HANDLE ANDBEWOOTH HIS LIKE Q AND Q
"I GUESS YOU MUST THINK I'M KINDA BATTY."	RUSTIAN WASTIN PAN ONTE BATTLY
NO ONE NEAR THE REMOTE MACHINE YOU COULD RING?	NO ONE NEAR THE REMOTE MACHINE YOU COULD RING
SAUCE FOR THE GOOSE IS SAUCE FOR THE GANDER.	SAUCE FOR THE GUICE IS SAUCE FOR THE GONDER
GROVES STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD.	GRAFS STARTED WRITING SONGS WHEN SHE WAS FOUR YEARS OLD

Evaluation

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

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

LANG_ID = "en"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-english"
DEVICE = "cuda"

CHARS_TO_IGNORE = [",", "?", "¿", ".", "!", "¡", ";", "；", ":", '""', "%", '"', "<22>", "ʿ", "·", "჻", "~", "՞",
                   "؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
                   "{", "}", "=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ", "‹", "›", "©", "®", "—", "→", "。",
                   "、", "﹂", "﹁", "‧", "～", "﹏", "，", "｛", "｝", "（", "）", "［", "］", "【", "】", "‥", "〽",
                   "『", "』", "〝", "〟", "⟨", "⟩", "〜", "：", "！", "？", "♪", "؛", "/", "\\", "º", "−", "^", "ʻ", "ˆ"]

test_dataset = load_dataset("common_voice", LANG_ID, split="test")

wer = load_metric("wer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/wer.py
cer = load_metric("cer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/cer.py

chars_to_ignore_regex = f"[{re.escape(''.join(CHARS_TO_IGNORE))}]"

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
model.to(DEVICE)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    with warnings.catch_warnings():
        warnings.simplefilter("ignore")
        speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = re.sub(chars_to_ignore_regex, "", batch["sentence"]).upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)

# Preprocessing the datasets.
# We need to read the audio 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(DEVICE), attention_mask=inputs.attention_mask.to(DEVICE)).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)

predictions = [x.upper() for x in result["pred_strings"]]
references = [x.upper() for x in result["sentence"]]

print(f"WER: {wer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")
print(f"CER: {cer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")

Test Result:

In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the evaluation script described above on other models as well (on 2021-06-17). Note that the table below may show different results from those already reported, this may have been caused due to some specificity of the other evaluation scripts used.

Model	WER	CER
jonatasgrosman/wav2vec2-large-xlsr-53-english	18.98%	8.29%
jonatasgrosman/wav2vec2-large-english	21.53%	9.66%
facebook/wav2vec2-large-960h-lv60-self	22.03%	10.39%
facebook/wav2vec2-large-960h-lv60	23.97%	11.14%
boris/xlsr-en-punctuation	29.10%	10.75%
facebook/wav2vec2-large-960h	32.79%	16.03%
facebook/wav2vec2-base-960h	39.86%	19.89%
facebook/wav2vec2-base-100h	51.06%	25.06%
elgeish/wav2vec2-large-lv60-timit-asr	59.96%	34.28%
facebook/wav2vec2-base-10k-voxpopuli-ft-en	66.41%	36.76%
elgeish/wav2vec2-base-timit-asr	68.78%	36.81%