README.md

metadata

language: fr
datasets:
  - common_voice
metrics:
  - wer
  - cer
tags:
  - audio
  - automatic-speech-recognition
  - speech
  - xlsr-fine-tuning-week
license: apache-2.0
model-index:
  - name: XLSR Wav2Vec2 French by Jonatas Grosman
    results:
      - task:
          name: Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice fr
          type: common_voice
          args: fr
        metrics:
          - name: Test WER
            type: wer
            value: 16.86
          - name: Test CER
            type: cer
            value: 5.65

Wav2Vec2-Large-XLSR-53-French

Fine-tuned facebook/wav2vec2-large-xlsr-53 on French 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 = "fr"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-french"
SAMPLES = 5

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
"CE DERNIER A ÉVOLUÉ TOUT AU LONG DE L'HISTOIRE ROMAINE."	CE DERNIER EST VOLÉ TOUT AU LONG DE L'HISTOIRE ROMAINE
CE SITE CONTIENT QUATRE TOMBEAUX DE LA DYNASTIE ACHÉMÉNIDE ET SEPT DES SASSANIDES.	CE SITE CONTIENT QUATRE TOMBEAUX DE LA DYNASTIE ASHÉMÉNIDE ET SEPT DES SASANNIDES
"J'AI DIT QUE LES ACTEURS DE BOIS AVAIENT, SELON MOI, BEAUCOUP D'AVANTAGES SUR LES AUTRES."	JAI DIT QUE LES ACTEURS DE BOIS AVAIENT SELON MOI BEAUCOUP DAVANTAGES SUR LES AUTRES
LES PAYS-BAS ONT REMPORTÉ TOUTES LES ÉDITIONS.	LE PAYS-BAS AN REMPORTAIT TOUTES LES ÉDITIONS
IL Y A MAINTENANT UNE GARE ROUTIÈRE.	IL A MA ANDIN GARD DETIRON

Evaluation

The model can be evaluated as follows on the French 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 = "fr"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-french"
DEVICE = "cuda"

CHARS_TO_IGNORE = [",", "?", "¿", ".", "!", "¡", ";", ":", '""', "%", '"', "�", "ʿ", "·", "჻", "~", "՞", 
                   "؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
                   "=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ", "‹", "›", "©", "®", "—", "→", "。"]

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):
\tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

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

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:

My model may report better scores than others because of some specificity of my evaluation script, so I ran the same evaluation script on other models (on 2021-04-21) to make a fairer comparison.

Model	WER	CER
jonatasgrosman/wav2vec2-large-xlsr-53-french	16.86%	5.65%
Ilyes/wav2vec2-large-xlsr-53-french	24.76%	8.06%
facebook/wav2vec2-large-xlsr-53-french	25.45%	10.35%
MehdiHosseiniMoghadam/wav2vec2-large-xlsr-53-French	28.22%	9.70%